This action might not be possible to undo. Are you sure you want to continue?

# Relaying in Interference Limited Networks: Models, Bounds, and Strategies

DISSERTATION

for the Degree of

Doctor of Philosophy (Electrical Engineering)

Onur S ¸ ahin January 2010

**Relaying in Interference Limited Networks: Models, Bounds, and Strategies DISSERTATION
**

Submitted in Partial Fulﬁllment of the Requirement for the Degree of Doctor of Philosophy (Electrical Engineering) at the POLYTECHNIC INSTITUTE OF NEW YORK UNIVERSITY by Onur S ¸ ahin October 2009

Approved:

Department Head 20 Copy No.

ii Approved by the Guidance Committee:

Major: Electrical Engineering

Elza Erkip Associate Professor of Electrical and Computer Engineering

Shivendra S. Panwar Professor of Electrical and Computer Engineering

Osvaldo Simeone Assistant Professor of Electrical and Computer Engineering New Jersey Institute of Technology Newark, NJ

iii Microﬁlm or other copies of this dissertation are obtainable from UMI Dissertation Publishing Bell & Howell Information and Learning 300 North Zeeb Road P. O. Box 1346 Ann Arbor. Michigan 48106-1346 .

Turkey.iv VITA Onur S ¸ ahin was born in Denizli. Since then. in 1981 and received his B.S. NY. His research interests include information theory and communication theory with the emphasis on cooperative systems and multi-user networks. During his internships. Ankara. degree in Electrical and Electronics Engineering from Middle East Technical University. . In 2005 and 2007. Turkey in 2003. he has been working towards his PhD degree at Polytechnic Institute of New York University. he worked as a summer intern at Philips Research North America. he worked on cooperative protocols for ad-hoc systems and cognitive radio systems for high-deﬁnition video transmission over wireless channels. Briarcliﬀ.

v To my beautiful family for their love .

one person existed at any time. I am fortunate to have Yi˘ git Atılgan as a great friend and roommate for more than 4 years. I could never imagine his visit to Poly would pave the way towards completing my work. Getting PhD is already a diﬃcult task. I would like to thank Zinan Lin. I appreciate their endless support while I ﬁnd myself getting lost at various things. I extremely enjoyed spending all these years next to ¨ u Alay. Osvaldo Simeone whom I had chance to know in the last year of my study. I am extremely grateful to Prof. He has been a great friend and I enjoyed all our discussions particularly the ones at West Village cafes on Sundays. Throughout all these years Poly introduced great friends. I will always remember his endless help when getting stuck in various points at my research. Melda Y¨ uksel and Deniz G¨ und¨ uz for their support during my ﬁrst years at school and in New York. I am indebted to her for guiding my study with patience and support for the problems I encountered beyond school.vi ACKNOWLEDGEMENT It is impossible to distinguish the eﬀort spent on the technical outcome of my PhD study from the daily life in many aspects. I want to foremost express my appreciation to my advisor Professor Elza Erkip. However. from most joyful times to over depressed moods. Ozg¨ gan Bakano˘ glu. . and it becomes even harder in New York City. It is always well beyond expectations to work with such a nice and understanding person who is also a great scholar. Whether we should have spent more energy for our thesis than our out-of-school life may be an interesting question to consider but in any case we enjoyed the city together. It has always been a pleasure to be her student bearing great conﬁdence at all times. Sintayehu Dehnie. Several people have been in the center of this process and made tremendous eﬀects mostly easing the harsh road of research which desperately pulls one to obsession and loneliness. Ka˘ Mehmet Turan. Ya˘ gız S¨ utc¨ u and H¨ useyin Akcan with whom I experienced unique friendships and amusing moments. He is one of the best gifts to me from New York City.

My only wish is that they raised us as they desired mostly with joy and pride. my beautiful family has always been in my heart and mind. . Ceren Erdem. I sincerely appreciate my dad and mom’s eﬀorts and struggles all these years to raise all of us without hesitating any sacriﬁces from their life. Taylan. Utku. and Bari¸ s for making me happy anytime and completing life in all aspects. Finally. anthropology and mathematics. My dear Cece. Since the ﬁrst day in New York. thanks to my dear G¨ ulnaz. Thanks for pure love that made everything more meaningful. Deniz. This thesis might not have ﬁnished without her day-care service in Cihangir where a signiﬁcant part of it is written. I always felt the luckiest person to have such a great family whom I miss more and more everyday.vii My dear friends Mustafa Avcı. shined in the darkest times and showed me beautiful colors I couldn’t even imagine before. respectively. PhD’s at musicology. Duygu Parmaksızoglu and Mert G¨ urb¨ uzbalaban have always made me feel relieved. This thesis is a very basic gift to them. Thanks for warming my life through Bosphorus and physically in New York. We formed one of the most interesting academic backgrounds together. I am grateful for the splendid times we spent together and sure to come more.

and the relay receives and transmits in the same band as the sources (in-band relay reception/ transmission ). take place over orthogonal links (out-of-band relay reception and/or transmission ). various conﬁgurations are explored: the relay reception. Relaying. and Strategies by Onur S ¸ ahin Advisor: Elza Erkip Submitted in Partial Fulﬁllment of the Requirements for the Degree of Doctor of Philosophy (Electrical Engineering) October 2009 Interference is inherent in wireless communication systems and can severely degrade communication rates. Bounds. To model and explore these interference-limited systems. Depending on the relay’s transmission and reception bands. Novel relaying techniques. an interference channel (IC) with a relay is considered where the relay operates over an .viii AN ABSTRACT Relaying in Interference Limited Networks: Models. initially. is a well-known technique primarily used to provide signiﬁcant rate improvements to point-to-point communications. In this dissertation. to manage interference in interference-limited multi-terminal networks are proposed. the impact of relaying in interference-limited networks is investigated. denoted by signal relaying. interference forwarding and interference cancelation. or both reception and transmission. on the other hand.

The optimality of signal relaying is shown to entail separability among the IC and the relay channel. Outer bounds on the capacity region are obtained and shown to be tight for relatively large source-to-relay channel gains and the channel conditions named as mixed relay-interference conditions and very strong relay-interference conditions. which cannot be exploited by IC-OBR. an interference channel assisted by a full-duplex relay that operates in the transmission band of the source-destination pairs is explored. The essence of interference forwarding in addition to signal relaying in improving the communication rates is demonstrated and the capacities are established in special cases. An achievable scheme with rate-splitting at the sources and decode-and-forward (DF) transmission at the relay is obtained and compared with its time-division counterpart. Next. Finally. It is shown that in-band transmission by the relay uniquely facilitates the reduction of interference at the terminals through interference cancelation feature at the relay. This model is denoted by IC with an out-of-band reception/in-band transmission relay (IC-OIR). An achievable scheme based on partial decode-and-forward transmission and dirty-paper coding at the relay is proposed. The conditions under which signal relaying and/or interference forwarding operations ensure optimality in terms of capacity are identiﬁed. the relay is assumed to transmit in the band of IC while reception is employed over the channels with orthogonal bands and ﬁnite capacities. . exhibiting substantial reduction in encoding/decoding complexity. The overall system is referred to as IC with an out-of-band relay (IC-OBR).ix orthogonal band with respect to the underlying IC. Interference forwarding and cancelation operations are demonstrated to provide signiﬁcant throughput gain depending on the network topology.

. . . . . . .3 Analysis of IC-OBR Type-I . . . . . . . . 3. . . . . . . . . .1. . . . . . . . .x Contents List of Figures 1 Introduction 1. . .3 Related Work .2. . . . . . . . 3. . . . . . . . . . . . . . . . . . . . .1 Decode-and-Forward Transmission 2. . . . . . . . . . . 3. . . . . . . . . . . . . 2. 3. . . . .2. . . . . . . 3. . . . . . . . . . . . .3 Symmetric IC-OBR Type-I and Type-II . . . . 3.1 Motivation . . . . . . . . . . . . . .3 Capacity Results for OBRC Variable Bandwidth Allocation . . . . . .2 System Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . . . . . . . . . . . .4 Capacity Results for OBRC Variable Bandwidth Allocation . . . . . . . . . . . .1 Outer Bound . . . . . . . . . 3. . . . . . . . . . . . . . . . . . .4. . . . . . . . . . . . . . . . . . .1. . . . . . . . . . . . . .3 Capacity Results for Fixed OBRC Bandwidth Allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Interference Channel . . . . .3. .3. . . . . . . . . . 2 Background and Related Work 2. . . . . . . . . .1 Han-Kobayashi Achievable Scheme 2. . . 3. 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 IC-OBR Type-II . .2 Achievable Rate Region . . . . . . . . . . . . . . .2 Relay Channel . . . . 2. . . 3. . . . . . . . . .4. . . . . . . xiii 1 1 5 6 8 8 9 11 13 13 15 18 18 19 20 22 22 23 23 24 24 26 33 35 36 38 46 3 Interference Channel with an Out-of-band Reception/Out-of-band Transmission Relay 3. . . . . . . . . . . . . . .3.2. . . .2.2. . . . . . . . . . . . 3.1 Introduction . .2 Capacity Results . . . . . . . . . . . . .2 Capacity Results for Fixed OBRC Bandwidth Allocation . . . . .4 Achievable Rates for Fixed and Variable OBRC Bandwidth Allocation . . . . . . .1 Outer Bounds .4 Analysis of IC-OBR Type-II . . .3 Dissertation Outline . . . .3. . . . . . 2. . . . . . . . 3. . .2 Key Contributions . .1 IC-OBR Type-I . . . . . . . . . . . . . . . . 3. . . . . . . . . . . . . . . 1. . . . . . . 3. . . . . . 3. . .

. . . . . . . . . Proposition 10 Proposition 11 Proposition 12 Propositions 13 Propositions 15 Proposition 16 Proposition 17 Proposition 19 Proposition 20 Proposition 21 . . . . . . . . . 83 5 Interference Channel with an In-band sion Relay 5. . . . . . . . . . . . . . . . . . . . . . . . . 67 4. . . . . . . . . . . . . . . . . . . . . . . . .1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . and 14 and 18 . . . . . . .5 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13 Proof A. . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 Proof A. . . . . . 76 4. .4 Proof A. . . . . . . . . . . . . . . . . . . . . 64 4. .4 Orthogonal Transmission . . . . . . . . . .8 Illustration of the Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proposition 9 .14 Proof of of of of of of of of of of of of of of Proposition 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Proof A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proposition 8 . . .10 Proof A. . . . . . . . . . 75 4. . . . . . . . . . 63 4. . . . . . . . . . . . . . . .5 Sum-Rate Maximization . . . . . . . . 47 4 Interference Channel with an Out-of-band Reception/In-band Transmission Relay 60 4. . . . . . . . . . . .3 A General Achievable Region and Outer Bounds for IC-OIR . . . . . . . . . . . .xi 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 4. . . . . . . . . . . . . . . . . . . . . . . . . . . 5. .6 Discussion of the Results . . .1 Proof A. . . . . . . . . . . . . . . . . . . .5 Proof A. . . .2 System Model . . . . . . . . . . . . 5. . . . . . . . . .4 An Achievable Region . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . . 6 Conclusion A A. .6 Sum-Capacity of IC-CR under Mixed Relay-Interference Conditions . . .12 Proof A. . . . . . . . .3 Rate Splitting at the Sources . . . . . . .2 Proof A. . . . . . . . . . . . . . . . . . . . . . . . 5. . . . . .9 Chapter Summary . . Proposition 6 . . . . .6 Proof A. . . . . . . . . . . . . . . . .7 Proof A.1 Introduction . . . . Reception/In-band Transmis. . . . . . . . . . . . . .11 Proof A. . . . . . . . . . . . . 60 4. . .3 Proof A. 5. . . . .1 Outer Bounds for the IC-OIR and IC-CR . . . 5. . . . . .7 Chapter Summary . . . . . . . . . . . . . 74 4.5 Very Strong Capacity Region of IC-OIR . . . . . . . . . . . . . . . . . . . . . 85 85 86 86 89 90 91 93 97 100 100 102 103 105 106 107 107 108 111 112 114 114 115 116 . . . . . . . . . . . . . . . . . .7 Capacity Region of a Degraded IC-CR . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4. . . . . . . . . . . . . . . . . . . .2 System Model . . . . . . . . . . . . . 5. . . . .

. 118 126 129 131 133 . . . . . . . . . . . . . . . . . . .18 Proof Bibliography of of of of Proposition Proposition Proposition Proposition 22 23 25 28 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15 Proof A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xii A. . . . . .16 Proof A. . . .17 Proof A. .

. and iii) η1R = η2R = ηR1 = ηR2 = 1. . .36)) versus b2 for IC-OBR Type-I with variable OBRC bandwidth allocation 2 i=1 ηiR + ηRi = η. . . Achievable sum-rate (from Proposition 8. Achievable sum-rate (from Proposition 6. . . Illustration of the OBRC conditions leading to the sum-capacity in 2 Proposition 16: c1 ≥ c2 . . (ii ) IC-OBR Type II: The OBRC is divided into two orthogonal channels with ηM AC . a21 = a12 = 2. . . . . and c2 = 1. .1 Gaussian interference channel in standard form.3 3. . .xiii List of Figures 2. . .2 2. . namely i) η = 0 (no relay).36)) versus b2 for IC-OBR Type-I with variable OBRC bandwidth allocation (η = 1. The OBR channel (OBRC) has η channel uses for each channel use of IC. b2 1 = b2 = 1.26)) with signal relaying and interference forwarding (only common information transmission over the IC) and outer bound (from Proposition 5. . c2 = 3.5 . . . (A.35) with signal relaying (only common information transmission over the IC). . . . . . . i = 1. . . . .5. . 2 2 2 b2 2 = c1 = 6. (i ) IC-OBR Type-I: The OBRC is divided into four Gaussian orthogonal channels with ηiR . . P1 = P2 = P1R = P2R = PR1 = PR2 =10dB. . .3. . . Achievable sum-rate from Proposition 6 for IC-OBR Type-I with symmetric IC. . . . . . . . .4 52 53 3. ηRi . . All powers are set to 10 dB. . . . . . . . . Gaussian relay channel. . c1 = b1 = 2). . . P1 = P2 = P1R = P2R = PR1 = PR2 =10dB. . . . (which satisﬁes the conditions in Proposition 7). and outer bound (from Proposition 5 (A. . . . . . . . . . . . . ii) η1R = η2R = 2 2 2 ηR1 = ηR2 = 1. . . . . . (A. . c1 = 3. . . . . . b1 = 1. .5 (which satisﬁes the conditions in Proposition 8). . . (A. channel uses each. . . .23)-(A. . .5. a21 = 2. . c1 = c2 = 0. Block-Markov encoding scheme for the relay channel. . . . .3 51 3. . . . . . ηBC channel uses. . . . . . . . . .1 2. . . b1 = c2 = 1. 2. ηM AC C (b2 1 P1R ) ≥ ηBC C (c1 PR ). η = 1.1. 10 14 14 49 3. . . a12 = 3. Interference Channel (IC) with an out-of-band relay (OBR). . . . .2 50 3. . . for ﬁxed OBRC bandwidth allocation versus a for various OBRC link capacities. . . . . . . . .

xiv 3.6 Illustration of the OBRC conditions leading to the sum-capacity in 2 2 Proposition 16: c1 ≥ c2 , ηM AC C (b2 1 P1R ) < ηBC C (c1 PR ), ηM AC C (b2 P2R ) ≥ ∗ 2 c ξ P ηBC C 1+2c2 ξ∗ R where where ξ ∗ is the optimal power allocation that PR

2

maximizes the sum-rate in Proposition 16 (ξ = 1 − ξ ∗ ). . . . . . . . . 3.7 Achievable sum-rate ((3.22a)-(3.22c)) and outer bound ((3.18a)-(3.18d)) for a symmetric IC-OBR Type II channel with respect to relay-todestination channel gains, c (a = 2, b = 1, P = Ps = 10). . . . . . . . 3.8 Achievable sum-rate and outer bound for an IC-OBR Type-II channel with respect to R − D1 channel gain, c1 and S2 − D1 channel gain a21 ∈ {0.1, 0.9.1.8} (a12 = 0.5, b1 = 1, b2 = 10, c2 = 1 and all node powers are equal to 10). . . . . . . . . . . . . . . . . . . . . . . . . . 3.9 Achievable sum-rates and outer bound for a symmetric IC-OBR TypeII with respect to the relay-to-destination gain c (b = 1, ηM AC = ηBC = 0.5, P1 = P2 = P1R = P2R = PR =10dB; Fig. i) a = 3, Fig. ii) a = 10). 3.10 Achievable sum-rate (from Proposition 16, (3.19)) with signal relaying (DF) and outer bound (3.25) and optimal parameters (ηM AC , ηBC , ξ ) of the DF scheme (3.19) for an IC-OBR Type-II with respect to S1 − R channel gain b1 (b2 = 2, c1 = 2, c2 = 0.3, η = 1, all node powers are equal to 10 dB, a21 = 1.8, a12 = 0.5.) . . . . . . . . . . . . . . . . . . 3.11 Achievable sum-rate (from Proposition 16, (3.19)) with signal relaying (DF) and outer bound (3.25) for an IC-OBR Type-II with respect to S1 − R channel gain, b1 and (b2 , c1 ) ∈ {(3, 2), (10, 5), (20, 10)} (c2 = 1, ηM AC + ηBC = η with η = 1, all node powers are equal to 10 dB, a21 = 1.8, a12 = 0.5.) . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12 Achievable sum-rate (from Proposition 19, (3.23a)-(3.23d)) with signal relaying and interference forwarding (DF, only common information transmission by S2 and only private information transmission by S1 over the IC) and outer bound (3.25) versus b2 for IC-OBR Type-II with variable OBRC bandwidth allocation 2 i=1 ηiR + ηRi = η, (η = 1, a21 = 1, a12 = 0.5, P1 = P2 = P1R = P2R = PR =10dB, c1 = 4, c2 = 1.5, b1 = 1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 4.2 Gaussian interference channel with an out-of-band reception/in-band transmission relay (IC-OIR). . . . . . . . . . . . . . . . . . . . . . . . Parallel source-destination pairs with a common orthogonal relay. The 2 2 total relay power satisfy ξ1 c + ξ2c ≤ 1. This is an equivalent system for an IC-OIR under very strong interference. . . . . . . . . . . . . . . . Achievable region of a symmetric IC-OIR with diﬀerent C1 , C2 values. The system parameters are P1 = P2 = PR = 10, a11 = a22 = 1, c1 = c2 = 1, a21 = a12 = 2. The C1 , C2 → ∞ curve also corresponds to the IC-CR model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗

53

54

55

56

57

58

59 64

72

4.3

77

xv 4.4 Achievable sum-rate of a symmetric IC-CR for diﬀerent transmission techniques. The system parameters are P1 = P2 = PR = 10, a11 = a22 = 1, c1 = c2 = 1, µ = a21 /a22 = a12 /a11 . . . . . . . . . . . . . . . . Achievable sum-rate of a one-sided IC-CR for diﬀerent transmission techniques. The system parameters are P1 = P2 = PR = 10, a11 = a22 = 1, c1 = 1, a12 = c2 = 0, µ = a21 /a22 . . . . . . . . . . . . . . . . . Optimal relay power allocations that maximize achievable sum-rate of the one-sided IC-CR in Fig. 4.5. The system parameters are P1 = P2 = PR = 10, a11 = a22 = 1, c1 = 1, a12 = c2 = 0, µ = a21 /a22 . . . . . Achievable sum-rate (4.6a)-(4.6f) and outer bound (4.18a-4.18d) of a symmetric IC-CR with respect to µ = a21 /a22 = a12 /a11 . The system parameters are P1 = P2 = PR = 10, a11 = a22 = 1, and c1 = c2 = 0.2 . Achievable sum-rate and sum capacity of an IC-CR operating under mixed relay-interference regime. The system parameters are P1 = P2 = PR = 10, a11 = 1, c1 = 1, a21 = 1, µ = a22 /a21 = a12 /a11 = c2 /c1 . . . . Gaussian Interference Relay Channel with two source-destination pairs Total optimized sum-rate (bits/channel use) of the symmetric channel with a21 = 1/8, P1 = P2 = PR = 10. When the relay is not present, we have P1 = P2 = 15. . . . . . . . . . . . . . . . . . . . . . . . . . . Optimal power allocation variables of the symmetric channel for rate splitting with a21 = 1/8, P1 = P2 = PR = 10. . . . . . . . . . . . . . Total optimized Rate (bits/channel use) of the symmetric channel as a function of a21 with b1 = 100, c1 = 1.2, P1 = P2 = PR = 10. When the relay is not present, we have P1 = P2 = 15. . . . . . . . . . . . . Optimal power allocation variables of the symmetric system for rate splitting as a function of a21 with b1 = 100, c1 = 1.2, P1 = P2 = PR = 10.

78

4.5

79

4.6

80

4.7

82

4.8

83 87

5.1 5.2

93 94

5.3 5.4

95 96

5.5

A.1 Equivalent model for IC-OBR Type II channel for c1 , c2 → ∞. . . . . 111

1

Chapter 1 Introduction

1.1

Motivation

Modern wireless networks are characterized by the coexistence of multiple

systems and devices. The increasing demand for the utilization of various services in a network simultaneously, such as Wi-Fi, cellular, and bluetooth systems, necessitates an overall design that judiciously incorporates interactions among these systems and the devices within each system. In a communication medium consisting of independent transmitters and receivers, interference between diﬀerent transmitter and receiver pairs is an unavoidable and fundamental phenomenon that limits communication. In the current wireless systems, common approaches to overcome the limitation due to interference can be stated as orthogonalizing the communication links in time/frequency domain or treating the interfering signal as noise while the desired signal is decoded. Despite the substantial reduction in complexity, it is well-known that these schemes can be strictly suboptimal in terms of transmission rates [9]. In the last decade, multiple-input multiple-output (MIMO) scheme has emerged as the foremost technique to improve the rate and reliability of communication systems [18] [19]. The gains oﬀered by MIMO systems rely on their potential to combat fading, introduce spatial diversity and increase spatial multiplexing. The relay channel, on the other hand, is another basic communication model that forms the building

corresponding achievable rates and the capacities of relaying systems in a point-to-point communication scenario are investigated extensively. to elaborate fundamental relaying operations in multi source-destination systems. Despite the inherit loss arising from distributed nature of the nodes. These techniques are not merely straightforward extensions of the standard pointto-point relaying schemes which basically aim at improving the decodability of the desired message.e. Our aim is to develop novel transmission techniques by incorporating joint source and relay transmission schemes and to get insight on the implications of these schemes on the practical systems such as next generation wireless systems. another goal in this thesis is to establish the capacity of various systems that are investigated under interference limited system framework. To understand the ultimate transmission rates one can achieve in the systems of interest and determine the optimal transmission strategies. we explore the eﬀect of relaying on the communication rates in multi-terminal networks where the system is prone to interference due to simultaneous transmission at the transmitter nodes. Relaying mimics MIMO communications by establishing interactions among the distributed nodes that serve as multiple antennas both at the transmitter and receiver sides. Speciﬁcally.2 block of larger networks [6]. various communication systems with multi-terminal network aided by a relay node are explored in this thesis. Even though the strategies. we explore a model where the relay transmits and receives via orthogonal bands with respect to the band the multiple sources and destination operate in. Initially. the optimal scenarios under which the system complexity is reduced are investigated. In this dissertation. cellular networks with Long Term Evolution (LTE) standards. This scenario of interest is modeled by a two-user interference channel aided by a relay with orthogonal re- . relaying oﬀers substantial gains similar to MIMO communications. but also involve the functions to eﬀectively ease the degradation in the network due to interference at the destination terminals. i. Motivated by these points. relaying in interference limited systems is less understood.

[72] [78]. this model helps us to understand fundamental relaying features in a general interference limited communication network.e. As a crucial design parameter. the relay reduces the strength of interference signal at the destinations [66] [72] [78]. i. our aim is to discover the impact of relaying on the communication rates of multiple terminals in the case where relay transmission is employed in the same band of the terminals.3 ception/transmission bands to the underlying interference channel. We denote such relaying operations as: i) Signal relaying. this assumption at the relay yields considerable utilities in the practical network design and improves the capability of the relay to further manage the interference in the . and iii) Interference cancelation. such as Wi-Fi and cellular systems operating simultaneously in the coverage of a base-station. and denoted by interference channel with an out-of-band relay (IC-OBR). i. ii) Interference forwarding. Compared with the out-of-band transmission of the relay node as mentioned above. For a total relay bandwidth constraint. IC-OBR lends itself in demonstrating signiﬁcant reduction in encoding/decoding complexity through optimality of particular relaying operations such as signal relaying. the relay convey additional information from the sources to the destinations. both ﬁxed and variable bandwidth allocations at the relay channel are explored. Besides its essence in modeling distinct communication systems that operate in the same network concurrently. Next. We investigate two speciﬁc models: (i) IC-OBR Type-I: The out-of-band relay channel (OBRC) is operated by assuming orthogonal transmissions over the four Gaussian links connecting sources to relay and relay to destinations. whereas the relay reception is in orthogonal bands. on the other. and the broadcast channel from relay to destinations. In this manner.e the relay helps the decodability of interference at the destinations. (ii) ICOBR Type-II: The OBRC is more generally operated by orthogonalizing the Gaussian multiple access channel between the two sources and the relay on one side. the impact of such relaying operations on the overall system complexity is a striking question that we further elaborate in this thesis.

one of our primary interests in examining such a model is to reveal which relaying features bear particular importance in treating the interference for utmost beneﬁt.4 network. Similar to the previously mentioned systems. Such situation arises in almost all networks with multiple terminals wishing to communicate with the others in the network. Yet. . [41] [59] [66]. Due to its structure. regardless of the communication technique/standard used. Despite its practical importance. a natural model to explore such a scenario is through a two user interference channel assisted by an in-band reception/in-band transmission relay. Finally. Similar to IC-OBR. The inherit diﬀerence of this model with respect to IC-OBR and IC-OIR is the reception band used by the relay which is indeed the only band used in the network. determining and analyzing the relaying functions suitable for this scenario further indicates the eﬀectiveness and importance of interference management in the communication networks. The fundamental reason behind such utilities relies on the relay’s capability to interact with the received signals at the destinations directly as a consequence of transmission in the bands of the interfering terminals. this property limits the relaying features and interference management capability substantially. We denote the model in this thesis as an interference channel with a relay [43] [66]. Hence. we explore a situation that is of particular interest in relaying/cooperation among the terminals that communicate on the same transmission band in a network. we denote this system as an interference channel with an out-of-band reception/in-band transmission relay (IC-OIR). we model such a scenario with a two user interference channel assisted by a relay node where the relay is connected to the sources via orthogonal and ﬁnite capacity links and the relay transmission is performed in the band of the interfering source terminals.

relying on coding for states known at the transmitter techniques such as partial decode-and-forward transmission [6]. whereas signal relaying is optimal for various channel conditions. compress-and-forward and lattice coding are given and compared for diﬀerent channel conditions. In an IC-OIR. For IC-OBR Type-II. Optimality of signal relaying and interference forwarding jointly for IC-OBR Type-I is also established under some conditions.5 1. The optimality of these schemes incorporated with signal relaying and interference forwarding are established under speciﬁc scenarios. our primary aim is to explore the eﬀect of in-band transmission by the relay.2 Key Contributions In this part. achievable regions due to decode-and-forward. Initially. an equally important task of the relay is to reduce the .1. in interference networks. dirty paper coding [12]. the optimality of signal relaying is shown for various situations such as when the relayto-destination channels of the OBRC form the performance bottleneck with respect to the source-to-relay. we demonstrate that signal relaying attains optimal separation of IC and OBRC and therefore under such conditions. the encoding/decoding complexity is shown to be reduced substantially. we summarize our contributions based on the system models mentioned in Sec. we give a general achievable rate region. hence yielding separable IC and OBRC operations. 1. Moreover. For both IC-OBR Type-I and Type-II. such as signal relaying and/or interference forwarding. For instance. Our results suggest that while relaying is usually utilized as an eﬀective technique for boosting received signal quality. interference forwarding is shown to be crucial under some cases. which requires inseparable operation over the IC and OBRC and therefore joint encoding/decoding over these channels are essential. For IC-OBR Type-I with ﬁxed bandwidth allocation at the OBRC. variable bandwidth allocation at OBRC leads to similar conclusions with its ﬁxed bandwidth allocation counterpart. and rate splitting at the sources [9].

6 interference signal strength at the destinations. in an interference channel with an in-band reception/in-band transmission relay. Moreover. We also demonstrate the throughput gain due to in-band reception and transmission of the relay by comparing the system with an orthogonal counterpart and show that in-band transmission provide signiﬁcant throughput improvement. is shown to be a very crucial technique for IC-OIR to improve communication rates in the network. In a symmetric system. we introduce our ﬁrst model. In particular the results suggests that. we establish the optimality of signal relaying and interference forwarding by demonstrating the capacity region of IC-OIR under very strong relay-interference conditions. under some set of channel gains. In Section 2. optimal power allocations for maximum total throughput show that the relay employs interference forwarding only via common information transmission for most of the channel gains. For . In Section 3. where the relay transmits negatively beamforms with the relay signal. the relay dedicates most of its resources in reducing the interference strength at the destinations through interference cancelation. On the other hand. we focus on two sub-models: IC-OBR Type-I and IC-OBR Type-II. we give some back- ground information about the model of interest and discuss related works. We give an achievable rate region obtained by incorporating rate splitting at the sources [9] and decode-and-forward relaying [6]. for very large source-to-relay channel capacities. Finally. interference channel with an out-of-band relay (IC-OBR). Both capacity results entail the necessity of inseparability of IC and the relay channel for optimal performance. we establish the sum-capacity of IC-OIR under the speciﬁed mixed relay-interference conditions. we incorporate full-duplex feature to reveal the maximum beneﬁt due to relaying. 1.3 Dissertation Outline The dissertation is organized as follows. Depending on the bandwidth allocation at the out-of-band relay channel(OBRC). Interference cancelation.

. In Section 4.7 both systems. We give the sum capacity of this model as well as the capacity region of the ﬁnite-link system under very strong and mixed relay interference conditions. achievable regions and outer bounds are given for ﬁxed and variable bandwidth allocation at the OBRC. we conclude the dissertation. we extend the model to an in-band transmission relay which still receives in orthogonal bands from the sources (IC-OIR). we explore an interference channel with a relay where the IC and the relay channel operate in the same band. In Section 6. Capacity results demonstrating the optimality of relaying operations as well as separability of IC and OBRC are demonstrated. A general achievable scheme including non-linear relaying schemes is given. In Section 5. For large source-to-relay link. We demonstrate a general achievable region and the optimal total throughput is compared with an orthogonal system. we make the analogy of the system to a cognitive model and determine outer bounds.

interference channel (IC) model is used extensively in the literature. To account and understand the role of interference. We also provide literature review on relaying in interference limited systems and consider related works on our models of interest. the source signals interfere with each other at the destinations. Both channels have been investigated extensively in the literature and the capacities are established in some special cases. The best strategy and corresponding achievable rate region is obtained by Han and Kobayashi [9]. However. and ii) relay channel. In an interference channel. we brieﬂy review these channels in terms of achievable schemes developed and established capacity results. complete understanding of any of the two channels has not been succeeded yet.1 Interference Channel Interference is a central phenomenon in any communication system com- posed of multiple transmitter and receiver nodes sharing a common medium. The scheme involves splitting of source messages as common and private parts where the common part is decoded at both . hence the point-to-point communication performances are degraded. 2. In this section.8 Chapter 2 Background and Related Work Relaying in interference limited systems inherits two building blocks of communication systems: i) Interference channel.

.2) where Xi.t + a21 X2.t = X1. the capacity region of a Gaussian IC is still unknown for most of the cases except under the strong [10] and noisy interference [64] [44] [67]. Each transmitter splits its message as Wi = (Wic . In the scheme. In this section we brieﬂy mention about the Han-Kobayashi scheme as well as capacity results known for a Gaussian IC. rate-splitting at the sources is performed to facilitate interference cancelation at the destinations.t + Z2. such that the common splits (W1c .al.t = a12 X1..i. respectively.t are independent identically distributed (i.t + X2.d. P P1 = P2 . (2. The signals received on the IC by the two receivers D1 and D2 in channel use t = 1. 2.1) (2. Y1.1 Han-Kobayashi Achievable Scheme As mentioned. [62] demonstrated that a very simple version of Han-Kobayashi scheme can achieve the rates within 1 bits/s/Hz of capacity region.t . Han-Kobayashi scheme gives the best known achievable re- gion for an IC to date. 2. The following assumptions determine a symmetric Gaussian interference channel as well as symmetric rate of IC where a a21 = a12 . a two-user Gaussian interference channel in its standard form is shown.9 destinations to reduce the eﬀect of interference. In Figure 2. Due to its convenience in characterizing achievable and capacity regions. Etkin et.t Y2. Wip ). n are given as..t ] ≤ Pi . which satisﬁes the power constraint 1 n n t=1 2 E[Xi. for a Gaussian interference channel. and Zi. symmetric interference channel is quite important. Recently. W2c ) are .t ∈ R represents the (real) input symbol of source Si . However.t + Z1.) zero-mean Gaussian noise processes. i = 1.1. R R1 = R2 ..1.

3) is within 1 bit/s/Hz of a symmetric capacity of a Gaussian interference channel. Then. each source employs superposition coding to transmit n n the codeword Xin = Xic (Wic ) + Xip (Wip ). W2p . i = 1. 2. decoded at both destinations. The encoding n n is performed by generating two codebooks with codewords Xic (Wic ) and Xip (Wip ). Proof 1 The proof is based on the rate-splitting idea used in the general Han-Kobayashi scheme as discussed above. However. The following proposition demonstrates that a simple Han-Kobayashi scheme achieves within 1 bit/s/Hz of a symmetric Gaussian IC capacity. X1 c (W1c ) n or X2 c (W2c ). i = 1. The details are given in [62] and omitted here.10 Z1 W1 X1 1 a12 a21 + Y1 W1 W2 X2 1 + Z2 Y2 W2 Figure 2. Han-Kobayashi scheme involves joint de- coding of the message splits (W1c . and hence part of the interference can be removed at n the destinations by stripping-oﬀ the corresponding interfering codeword. at the sources. W1p . 2C a2 P + 1 a2 −1 (2. The private split Wip is decoded at Di only. W1c ) at D2 . W2c ) at D1 and (W2c . 2. each destination simply decodes its own private message Wip . 2. . W2c ) are decoded ﬁrst and the codewords n n X1 c (W1c ) and X2c (W2c ) are stripped-oﬀ. Then. i = 1. Proposition 1 ( [62]): The following rate R = min C (P + a2 P ) + C 1 + 1 a2 − 1. sequential decoding is performed at the destinations such that the common splits (W1c .1: Gaussian interference channel in standard form.

In particular. that is Y2n . Then. Using similar approach it is possible to show that for a12 ≥ 1.6) (2. The following proposition gives the capacity region of a Gaussian IC operating under strong interference conditions. for a12 ≥ 1 and a21 ≥ 1.2). [64] [44] [67] propose giving genie signals to the destinations which help . W2 and the overall system becomes a compound multiple-access channel whose capacity is the intersection of the two multiple-access channels. for a21 ≥ 1. D2 decodes the message W1 whenever D1 is able to decode it. Then. R2 ) is in the capacity region of a Gaussian IC as given in (2.7) In establishing the sum-capacity of a Gaussian IC under noisy-interference conditions. Recently. a21 (2.1. D1 is able to form a new signal n n Y1n = a12 X1 + X2 + n Z1 . D1 is able n to decode W1 .5) (2. (2.11 2. suppose that (R1 . In obtaining the capacity region under strong regime. W2 .2 Capacity Results The capacity region of interference channel is known under strong interfer- ence conditions only [9] [10]. Using the regenerated codeword and the received signal Y1n . the sum-capacity of a Gaussian IC is characterized under the so-called noisy interference regime [64] [44] [67].1)-(2. and hence reconstruct the codeword X1 (W1 ). Proposition 2 ( [9] [10]): The capacity region of a Gaussian interference channel with a21 ≥ 1 and a12 ≥ 1 is given by R1 ≤ C (P1 ) R2 ≤ C (P2 ) 2 R1 + R2 ≤ min C (P1 + a2 21 P2 ).4) Clearly. [10] uses the decodability of the interference signals at the destinations to obtain the corresponding outer bounds. and hence is able to decode W2 as long as D2 decodes its desired message. both destinations can decode W1 . for the generated codebook structure. C (a12 P1 + P2 ) . D1 observes a less noisy version of the signal observed at D2 .

Proposition 4 ( [44] [64] [67]): In a symmetric Gaussian interference channel satisfying the condition a + a3 P ≤ 0. and the maximization is over all random vector X n inden n pendent of Z1 and Z2 .5. Then. a Gaussian X n is an optimal solution of this optimization problem.9) Using Proposition 3 and Corollary 1. In addition the following Proposition and Corollary play a crucial role in demonstrating the eﬀectiveness of these outer bounds and is also used in this thesis in proving various results.8) for n n iid. Corollary 1 ( [67]): For µ = 1. a12 and a21 .8) is iid Gaussian and the optimum solution denoted by W is n log P +N1 2 P +N2 W = n log N1 2 N2 if N1 ≤ N2 if N1 > N2 (2.12 in obtaining meaningful outer bounds for small interfering gains.8) n n are Gaussian vectors with strictly positive deﬁnite covariance matrix and Z2 where Z1 KZ1 and KZ2 . The following Corollary explicitly solves the optimization problem (2.10) . respectively and assuming µ = 1. for any µ ≥ 1 and positive semideﬁnite S. the optimal solution of (2. (2. noises Z1 . Z2 with covariance matrices N1 I and N2 I. the noisy sum-capacity of a symmetric Gaussian IC is characterized as follows. Proposition 3 ( [36]): Consider the following optimization problem n n W = max h(X n + Z1 ) − µh(X n + Z2 ) p(x) subject to Cov(X n ) S (2. respectively.

t are independent identically distributed (i.t + cXR.t . 2. 2.) zero-mean Gaussian noise processes.i.d.2 shows a Gaussian relay channel. 1 n n t=1 2 E[XR..t = X1.t = bX1. (2.t + ZR.t ] ≤ P1 .t ∈ R represents the (real) input symbol of the source with the power constraint 1 n n t=1 2 E[X1 . As a key component and technique of several achievability and capacity results established in this thesis.10) is obtained by simply treating the interference as noise. The capacity of the relay channel is established for a physically degraded scenario and decode-and-forward transmission is shown to be the optimal scheme [6] under the physically degradedness condition. YR. we brieﬂy review the decode-and-forward scheme in the next section.t + Z1. The signals received by the relay and the destination in channel use t = 1.11) Note that the sum-capacity obtained in (2.13 the sum capacity is given by Csum = C 1 + P 1 + a2 P . (2.2 Relay Channel The relay channel was introduced by Van der Meulen [2]. the relay decodes the signal it receives from the source terminal and reencodes the source .2.13) where Xi..11) for the noisy-interference conditions given in (2.. respectively. 2.t ] ≤ PR and ZR. . Cover and El Gamal [6] demonstrated various achievable schemes such as decode-and-forward and compress-and-forward transmission. Fig.1 Decode-and-Forward Transmission The main idea in the decode-and-forward transmission is that. n are given as.t and Z1.12) (2.t Y1.

3. YR |X2 ).1) XRn(W1B) Figure 2. the source encodes and transmits the codeword X1 (W1 . W1 ). (2. and removes b−1 n the eﬀect of XR (W1 ) from the received signal. The decoding and transmission structure of the relay together with the source transmission is performed through a block strategy as shown in Fig. In BM transmission.W1B) XRn(W1B-1) b = B+1 X1n(W1B. the source encodes its messages by incorporating memory such b−1 n b that at each block b. the following rate is achievable.2: Gaussian relay channel. b=B X1n(W1B-1. the relay transmits the codeword XR (W1 in block b + 1. 2.14 Z1 X1 ZR 1 c + Y1 b b=1 X1n(1.W12) XRn(W11) ………. In Fig. message and transmits to the destination. Block-Markov (BM) encoding scheme is shown. sliding window technique is employed such that the source the signals received at block b and b + 1 are used in decoding the messages . R ≤ I (X1 . b=2 X1n(W11. Therefore. At the destination.3: Block-Markov encoding scheme for the relay channel. 2.3.14) n b ) After decoding the corresponding message.W11) XRn(1) + YR R XR Figure 2. The relay knows the messages transmitted in the previous block b − 1.

3 Related Work In this part. X2 . one can choose the distribution as well as the codewords such that X1 = αP1 XR + X1 PR (2. capacity achieving schemes for deterministic relay networks are demonstrated. Then. we also consider relaying and cooperation in networks with multiple nodes which lead to various cooperation models. Gupta and Kumar [20] give an achievable scheme for a large network with multiple relays.17) where XR ∼ N (0. As a closely related area. YR ). Moreover. Aref [7] generalized the techniques developed for the regular relay channel [6] to account for multiple relays. it is possible to show that the following rates are achievable. (W1 R ≤ I (X1 . C P1 + c2 PR + 2c αP1 PR . using decode-and-forward and . YR |X2 ). In [26]. I (X1 . by incorporating irregular encoding/successive decoding and decode-and-forward relaying. More recently.15) (2. PR ) and X1 ∼ N (0. in [7]. (2. the following rate is achievable.18) 2. the following gives an achievable rate. Then. Extension of single source-destination pair to multiple pairs assisted by multiple relays is also investigated in [20].15 b+1 b .16) For a Gaussian system. W1 ). pX1 XR (2. R ≤ min C b2 αP1 . Vast majority of work on cooperation among multiple nodes assume single source-destination pair assisted by multiple relay nodes [7] [26]. Overall. we discuss the related work on relaying in interference limited systems. R ≤ max min{I (X1 . X2 . YR )}. αP1 ).

Related work is also presented in [53]. a broadcast relay channel (BRC) is also introduced in [26]. CF is the shown to be the best candidate for in terms of achievable rates. Both cases where the interference-free destination assists . in which simpliﬁed channel models are considered where the relay only receives from one source. Using cut-set bound argument. The discrete memoryless and Gaussian IC with a relay is further investigated in [49] [55]. The capacity region of this channel is established for various scenarios such as semideterministic and orthogonal partially cooperative BRC. As a natural extension of MARC and BRC. where the relay is assumed to be aware a priori of the users’ messages (cognitive relay) and sophisticated achievable strategies are investigated. It is shown that DF scheme performs very close to the upper bound when the relay is located close to the sources. [77] derives an upper bound on the capacity region by allocating inﬁnite power to the relay. In a BRC. The eﬀect of relaying for one-sided interference channel is explored in [57]. achievable regions for multiple relay. upper bounds on the capacity region are given and compared with the achievable regions due to DF and CF schemes for various node topologies in the network.16 compress-and-forward techniques. However. when the relay operates in the vicinity of the destination. In [39]. As a counterpart to the MARC model. In the MARC model. These works emphasize the fact that forwarding the interference of even a single source may improve the rates of both users. source and destination nodes are demonstrated. a multiple-access relay channel (MARC) model is explored. the IC aided by a relay has been ﬁrst studied in [43]. In particular. a partially cooperative relay broadcast channel is studied where one of the destinations acts as a relay in forwarding the source signal in spite of a dedicated relay node. DF-based strategies at the relays with joint decoding at the destinations are proposed without rate splitting and shown to exhaust the capacity region under some conditions. multiple sources wish to communicate with a common destination are assisted by a relay which can be full-duplex or half-duplex. a dedicated relay helps a source node that wishes to transmit its signal to multiple destinations in broadcast fashion.

For Type II system. whereas in [71] scenarios are found for which sending both correlated private and common messages is optimal. As shown in [76] [73] [71]. [70] explores the sum-capacity of parallel Gaussian interference channel in noisy interference regime and shows that the sum-capacity is achieved by treating interference as noise in each sub-channel as well as independent (separable) transmission at each sub-channel. . optimal operation over parallel ICs. is in general not suﬃcient. whereby the parallel channels are treated independently. DF and CF relaying schemes. and thus a separable approach. achievable rate regions are given for both cases. CF is shown to achieve sum-capacity for asymptotically high relay to destination link. Due to its crucial role in the system complexity. The original work [76] derives conditions under which correlated transmission of private messages is optimal. The optimality of interference-forwarding is established for asymptotically high signal-to-noise and interference-to-noise ratios for Type I system where it is shown that each relay bit asymptotically improves the sum-capacity one-bit. typically entails joint encoding over the parallel channels. separability of IC has been taken attention recently in the literature. Based on Han-Kobayashi rate-splitting technique. As a related set of work. In other words. the signals sent over the parallel ICs need to be generally correlated to achieve optimality. unlike scenarios with a single source or destination.17 the interfered destination (Type I) and vice versa (Type II) are investigated.

We are . and thus to interact and cooperate with devices belonging to diﬀerent systems and networks. many current wireless terminals are equipped with a 3G cellular transceiver along with Wi-Fi interface. Opportunities for cooperation are further enhanced for multistandard terminals that are able to communicate simultaneously over multiple radio interfaces. While this often leads to an overall system performance that is limited by mutual interference. we model the system of interest as an interference channel (IC) and an out-of-band relay (OBR) assisting the source-destination pairs via its orthogonal channels. For instance. To gain insight on the general picture. the presence of many independent wireless devices may also potentially oﬀer new opportunities and performance beneﬁts by allowing cooperation.1 Introduction Modern wireless communication networks are characterized by the coexis- tence of an increasing number of interfering devices and systems.18 Chapter 3 Interference Channel with an Out-of-band Reception/Out-of-band Transmission Relay 3. This chapter focuses on investigating the advantages of cooperation in interference-limited scenarios where cooperation is enabled by orthogonal radio interfaces and multistandard terminals. and denote the system as an interference channel with an out-of-band relay (IC-OBR).

3. namely a . (ii ) IC-OBR Type-II: The OBRC is more generally operated by orthogonalizing the Gaussian multiple access channel between the two source and the relay and the broadcast channel from relay to destinations. In both models.4. lead to the optimal performance. the sources S1 and S2 communicate to their respective destinations D1 and D2 via two orthogonal channels. the two components of an IC-OBR model are an IC and the OBR channel (OBRC). we conclude the chapter in Section 3. We model the latter in two diﬀerent ways: (i ) ICOBR Type-I: The OBRC is operated by assuming orthogonal transmissions (e.5. Finally. As explained above.2 System Models We investigate the IC-OBR models shown in Fig.1.19 interested in understanding the optimal operation on the overall system. we give the system models of IC-OBR Type-I and Type-II channels. Therefore. Conditions for the optimality of signal relaying and/or interference forwarding for IC-OBR Type-II are established in this section. Section 3. The chapter is organized as follows. our aim is to further elaborate the impact of using orthogonal bands for the relay on the overall system design. Notation : We deﬁne C (x) = 1/2 log2 (1 + x).. we are interested in revealing under which scenarios separable operation over the IC and OBRC. In Section 3. Capacity results are established for both cases. via TDMA or FDMA) over the four Gaussian links connecting sources to relay and relay to destination.3 gives a general outer bound and achievable region for IC-OBR Type-I for both ﬁxed and variable OBRC bandwidth allocations. we give outer bounds on the capacity region for IC-OBR Type-II. such that encoding and decoding operations are employed independently for IC and OBRC.2. which we refer to as IC-OBR Type-I and Type-II. Also. in particular the role of an orthogonal relay in interference-limited systems in terms of signal relaying and interference forwarding.g. 3. In Section 3.

t + Z2. we consider the number of messages 2nRi rounded oﬀ to the smallest larger integer. Speciﬁcally.t + a21 X2. The two IC-OBR models studied in the following diﬀer in the way the OBRC is operated. parameter η can be thought of as the ratio between the bandwidth of the OBRC and of the IC. 1 .. wishes to send a message index Wi . shown in Fig. as discussed below.1b) where Xi. the OBRC bandwidth (or equivalently the set of channel uses ηn) is partitioned into four orthogonal Gaussian channels. The signals received on the IC by the two receivers D1 and D2 in channel use t = 1.d. In pratice. .i..t + Z1. so that Ri is the rate of the ith pair (Si .3. Both models assume a half-duplex relay.20 Gaussian IC and the out-of-band relay channel (OBRC).t are independent identically distributed (i. Y1. and Zi. each source Si . with the help of an OBR.t ∈ R represents the (real) input symbol of source Si . (3..t = a12 X1.t = X1. Di ) in terms of bits per IC channel use.t + X2.) zero-mean Gaussian noise processes with unit power. to its destination Di . n are given as. where the latter is characterized by η ≥ 0 channel uses per channel use of the IC.t . Notice that n is the number of channel uses of the IC available for communication of the given messages (which yields ηn channel uses for the OBRC).1a) (3. Specifically.2.1 IC-OBR Type-I In the IC-OBR Type-I model. This can be realized by orthogonal access schemes such as TDMA or FDMA. 2nRi ] 1 . 2.t ≤ Pi . which operates half-duplex. We will use the same convention wherever integer quantities are needed. corresponding to diﬀerent source-to-relay and relay-to-destination pairs. respectively. which satisﬁes the power constraint 1 n n t=1 x2 i. uniformly drawn from the message set [1.1-(i ).t Y2. 3. i = 1. we have two Gaussian channels from sources Si to relay R with fraction of As it is common in the literature.

with fraction of channel uses ηRi . denoted by hi . 2nRi ] into the codewords (Xin . i = 1. i = 1.i. 2. i = 1. . The rationale behind this power constraint assumption arises for the scenarios such as the relay transmits using TDMA with per-symbol power constraints.t .21 channel uses ηiR . 2nRi ] → Rn × RηiR n (n) (n) (3. i = 1. zero-mean Gaussian noise processes with unit power. and x2 Ri.t . or employs FDMA transmission with spectral mask constraints.4) ηiR n which maps a message Wi ∈ [1. i = 1.5) nηR1 nηR2 η1R n η2R n (n) which maps the received signal to the codewords (XR (YR 1 .d. given by fi : [1. with. The signals received by the relay R over the OBRC on the source-to-relay channels for t = 1. and two Gaussian channels from the relay R to destinations Di .2) 2 i=1 (ηiR + ηRi ) = η ...t + ZiR. that are sent to the destinations. XR 2 ) = g 1 . 2. (c ) The decoding functions at the destinations Di . hi : Rn × RηRi n → [1.t ) are i.. n) code for the IC-OBR type-I is deﬁned by: (a ) The encoding functions at the sources Si .t ≤ PiR .t ≤ PRi . 2nR2 . 2.. We have are given by YiR.. ZRi. A (2nR1 . Another model that encompasses this assumption is where the relay communicates with the destinations using two distinct radio interfaces with diﬀerent transceivers. whereas the signals received at the destination over the OBRC on the relay-to-destination channels are given by YRi. YR2 ). (3.. ηiR n and i = 1.3) for t = 1. 2nRi ] (n) (n) (3. 2.6) . respectively.t = ci XRi. i = 1. 2. 2. (b ) The encoding function g (n) at the relay given by g (n) : Rη1R n × Rη2R n → RηR1 n × RηR2 n . 2. ηRi n. (3.t = bi XiR. We assume power constraints 1 n ηRi n t=1 1 n ηiR n t=1 x2 iR.t . 2. i = 1. . XiR ) = fi (Wi ) to be transmitted on the IC and OBRC. (3.t + ZRi. where (ZiR.

i = 1. We have ηM AC + ηBC = η . ηM AC n.t ≤ PiR .t + ZR. obtained by orthogonalizing the MAC and the BC. Note that IC-OBR Type-I is a special case of Type-II.t = ci XR.. YRi ). and the signal received at destination Di over the OBRC is YRi.1-(ii ). i = 1. A (2nR1 . . one being a multiple-access channel (MAC) from S1 and S2 to R. 2nR2 . ηBC n.2. shown in Fig. (n) 3.8) . and the other being a broadcast channel (BC) from R to D1 and D2 .3.3 Symmetric IC-OBR Type-I and Type-II Throughout the paper an IC-OBR Type-I with symmetric IC is deﬁned by a a12 = a21 .7) for t = 1.t + ZRi. . n) code for the IC-OBR Type-II is deﬁned similar to codes for IC-OBR Type-I with the diﬀerence that the encoding function g (n) at the relay is modiﬁed as g (n) : RηM AC n → RηBC n . 2.t ≤ PR .t .t = ηM AC n g (n) (YR ).t = b1 X1R. for t = 1.2 IC-OBR Type-II The IC-OBR Type-II model. the OBRC is orthogonal- ized into two channels.t + b2 X2R.22 ηRi n which maps the received signal via the IC. (3. and x2 R. P P1 = P2 . 3. with fraction of channel uses ηBC .... Yin and OBRC YRi into the estimated ηRi n message Wi = hi (Yin ... ηM AC n ηBC n which maps the received signal YR into the transmitted codeword XR. with fraction of channel uses ηM AC . We have the power constraints 1 n ηBC n t=1 1 n ηiR n t=1 x2 iR.2. The received signal at the relay R over the OBRC is given by YR. 2.t (3.

the bandwidth allocation parameters over the OBRC. η2R ) for IC-OBR Type-I and (ηM AC . 3. 3.3 Analysis of IC-OBR Type-I In this section. η1R . Therefore. we deﬁne the probability of error as the probability that any of the two transmitted messages is not correctly decoded at the intended destination. are considered to be given and ﬁxed. so that a rate pair (R1 . The capacity region is in both cases the closure of the set of all achievable rates. P s P1R = P2R .8) and b b1 = b2 . R2 ) is achievable if a coding scheme can be found that drives the probability of error to zero for the given (feasible2 ) bandwidth allocation parameters. namely (ηR1 . Bandwidth allocation parameters are feasible if ηM AC + ηBC = η for Type-II. ηR2 . (b ) Variable OBRC bandwidth allocation : Here the bandwidth allocation can be optimized.23 whereas we will refer to a IC-OBR Type-II as symmetric if it satisﬁes (3.2.9) 3. R2 ) are then deﬁned for two diﬀerent scenarios: (a ) Fixed OBRC bandwidth allocation : Here. (3. 2 2 i=1 (ηRi + ηiR ) = η for IC-OBR Type-I and . we investigate the IC-OBR Type-I system described in Sec. It is noted that the results for ﬁxed OBRC were partly presented in [72].2.4 Achievable Rates for Fixed and Variable OBRC Bandwidth Allocation Following conventional deﬁnitions. c c1 = c2 . a rate pair (R1 . ηBC ) for Type-II. We consider outer bounds and inner bounds to the achievable rate regions for both ﬁxed and variable OBRC bandwidth allocation. Achievable rates (R1 . R2 ) is said to be achievable if a coding scheme exists that drives the probability of error to zero for some feasible bandwidth allocation parameters.

Y1n ) + min η1R C (b2 1 P1R ) + η2R C (b2 P2R ). R2 ): R1 ≤ 1 n 2 2 I (X1 . the capacity region of the IC-OBR Type-I is contained within the set of rates (R1 .1. we derive an achievable rate region for the IC-OBR Type- I. Y2n ) + min η1R C (b2 1 P1R ) + η2R C (b2 P2R ). Proof 2 Appendix A. an outer bound is given as above but with the union in (3. ηRi . ηR2 C (c2 PR2 ) n (3. allowing the identiﬁcation of the capacity region of IC-OBR for various scenarios. ηR1 C (c1 PR1 ) n 1 n 2 2 R2 ≤ I (X2 . We propose to use a rate splitting scheme similar to the standard approach for .3.2 Achievable Rate Region In this section.10) n where the union is taken with respect to all multi-letter input distributions p(xn 1 )p(x2 ) that satisfy the power constraints 1/n n t=1 2 E[Xi. 2. ηR2 C (c2 PR2 ) n 1 n n 2 R2 ≤ I (X2 . R2 ) satisfying lim closure n n n p(xn 1 .3. With variable OBRC bandwidth allocation. we ﬁrst present a general outer bound to the capacity region of an IC-OBR in terms of multi-letter mutual informations (Proposition 1).24 3. 2.x2 )=p(x1 )p(x2 ) n→∞ (R1 . ηR1 C (c1 PR1 ) n 1 n n 2 R1 ≤ I (X1 . This bound is then specialized to a number of special cases of interest. i = 1. Y2n |X1 ) + min η2R C (b2 2 P2R ).t ] ≤ Pi . such that ηRi ) = η . Proposition 5 (Outer bound for IC-OBR Type-I): For ﬁxed OBRC bandwidth allocation. i = 1.1 Outer Bound In this section.10) taken also with respect to all parameters ηiR . 2 i=1 (ηiR + 3. Y1n |X2 ) + min η1R C (b2 1 P1R ).

. Common Messages) It should be noted that the considered transmission scheme is in general not separable. Wic . Speciﬁcally. (iv ) Wic ∈ [1.2nRiR ] is a private message that is transmitted via the OBRC only. to enable interference cancellation. . is sent over both IC and OBRC in order to allow interference mitigation. this message is conveyed interference-free to Di .. Wic ). . where: (i ) WiR ∈ [1. Notice that transmission of the private messages over the OBRC amounts to signal relaying. (ii ) Wip ∈ [1.. but one of the two is also sent over the OBRC to the interfered destination for interference cancellation. More speciﬁcally. Wip . i = 1.2nRip ] is a private message that is transmitted over the IC. 2. directly to Di . since the private information sent on the OBRC is conveyed without interference to the intended destination. Speciﬁcally. in the sense that correlated messages are sent over the IC and OBRC. This is apparently a reasonable choice for a IC-OBR Type-I.2nRic ] is a common message that is transmitted over IC only and decoded at both destinations. However. part of the common messages. where the private message of each source is to be decoded only by the intended destination and the common is to be decoded at both intended and interfered destinations.. 2nRic ] is a common message that is transmitted over the IC and OBRC. we have the following four-way split of each message Wi .. Wip ) are sent separately over the two parallel channels IC and OBRC. j = i. and thus there is no need for transmission also over the IC. Speciﬁcally. while the private message splits (WiR . we split the message of each user into private and common messages. As for the common message. Wi = (WiR . both parts are sent over the IC. One of the private message splits is sent over the IC and the other one over the OBRC.25 ICs [5] [9]. while transmission of the common parts can be seen as interference forwarding. Wic . the relay conveys Wic to Dj only. Remark 1 (Separability and Private vs.. Notice the since the OBR has orthogonal channels to the IC. j = i. . (iii ) Wic ∈ [1.. private and common parts are further split into two (independent) messages as follows. decoded at Di and treated as noise at Dj .

and the power allocations must satisfy the power constraints P1c + P1p ≤ P1 and P2c + P2p ≤ P1 . except S1 = {2c } and S2 = {1c }.26 Proposition 6 (Achievable Rate Region for IC-OBR Type-I): For ﬁxed OBRC bandwidth allocation. 2 i=1 (ηiR + ηRi ) = η .11a)-(3. 1c }. we discuss some condi- tions under which signal relaying or a combination of signal relaying and interference .3. where conditions (3. With variable OBRC bandwidth allocation.11c) (3. rates (3. and we deﬁne Ric = Ric + Ric .11a) (3. Moreover. Here. 2p.11e) (3.11a)-(3. kj 2 = a12 if j ∈ {1c. N2 = a12 P1p +1.11b) (3.2. 2. 1p. and kj 1 = a21 . provides an achievable rate region for the Gaussian IC-OBR Type-I. 1p}. j = 1.11d) (3. we use the convention Pjc = Pjc . R2 ) with Ri = Ric + Ric + Rip + RiR .11f) Rj ≤ C j ∈S2 N2 R1c + R1R ≤ η1R C (b2 1 P1R ) R2c + R2R ≤ η2R C (b2 2 P2R ) R2c + R1R ≤ ηR1 C (c2 1 PR1 ) R1c + R2R ≤ ηR2 C (c2 2 PR2 ).3 Capacity Results for Fixed OBRC Bandwidth Allocation Consider ﬁxed OBRC bandwidth allocation. kj 2 = 1 if j ∈ {2c.11b) must hold for all subsets S1 ⊆ T1 = {1c. 2 N1 = a 2 21 P2p +1. i = 1. 2p}. 2. the convex hull of the union of all rates (R1 . 2c } and S2 ⊆ T2 = {2c. that satisfy the inequalities Rj ≤ C j ∈S∞ j ∈S1 2 kj 1 Pj N1 j ∈S2 2 kj 2 Pj (3. 3.11f) can be evaluated for all bandwidth allocations satisfying Proof 3 Appendix A. and the parameters kj 1 = 1.

Proposition 7 (Optimality of Signal Relaying Only): For a IC-OBR Type-I 2 2 with ﬁxed OBRC bandwidth allocation and ηR1 C (c2 1 PR1 ) ≤ η1R C (b1 P1R ). R2 ≤ ηR2 C (c2 PR2 ). R2 ): (R1 − R1 . ηR2 C (c2 PR2 ) ≤ η2R C (b2 2 P2R ). a separable scheme (at least in the class of strategies we considered) is not suﬃcient to attain optimal performance. the relay to des2 tinations links form the performance bottleneck. R2 ): 1 n .x2 )=p(x1 )p(x2 ) (R1 . Optimality of Signal Relaying The ﬁrst result below shows that. on the path Si − R − Di . whereas. n t=1 . i = 1. Equivalently. the capacity region C is given by C = lim closure n→∞ n n n p(xn 1 . (3. ηR1 C (c2 1 PR1 ) ≤ η1R C (b1 P1R ) and 2 ηR2 C (c2 2 PR2 ) ≤ η2R C (b2 P2R ). The proposition below is expressed in terms of the capacity region CIC of a regular IC. i. ηR2 C (c2 PR2 )) along the individual rates as C = {(R1 . which is generally unknown in single-letter formulation apart from special cases. Y2n ) + ηR2 C (c2 2 PR2 ) n R1 ≤ the power constraints 1/n by signal relaying only.12) n where the union is taken with respect to the input distribution p(xn 1 )p(x2 ) that satisﬁes 2 ] ≤ Pi . signal relaying (and thus separable operation) achieves capacity.t .27 forwarding are optimal. en2 hanced by (ηR1 C (c2 1 PR1 ). 2 for R1 ≤ ηR1 C (c2 1 PR1 ). the capacity region C is given by the capacity region CIC of the IC. if both signal relaying and interference forwarding are needed.e. if. it is noted that optimality of signal relaying implies that separable operation on the IC-OBR is optimal. According to the discussion above. Y1n ) + ηR1 C (c2 I (X1 1 PR1 ) n 1 n R2 ≤ I (X2 . The capacity region is achieved E [Xi.. 2. R2 − R2 ) ∈ CIC }.

e. Remark 4 Both Proposition 6 and 7 apply also to a general discrete memoryless IC-OBR Type-I (with the caveat of eliminating the power constraint). in any scenario where a single-letter capacity region is known for the regular IC. the single-letter capacity region immediately carries 2 2 over to the IC-OBR Type-I with ηR1 C (c2 1 PR1 ) ≤ η1R C (b1 P1R ) and ηR2 C (c2 PR2 ) ≤ η2R C (b2 2 P2R ).28 Proof 4 The converse follows immediately from Proposition 5 with the conditions 2 ηRi C (c2 i PRi ) ≤ ηiR C (bi PiR ). In other words. Optimality of Interference Forwarding While Proposition 7 provides a general capacity result for the case where the relay-to-destination links set the performance bottleneck on the paths Si − R − Di . and then using the Gaussian IC as a regular IC. for instance. signal relaying is able to fully capitalize on such gains by performing signal forwarding (i. private. Remark 3 Due to Proposition 7. we can obtain a single-letter capacity region expression for an IC-OBR Type-I in the strong interference regime (a21 ≥ 1 and a12 ≥ 1) [9] [10] or the sum-capacity in the noisy or mixed interference regime [64] 2 2 2 [44] [67]. on the OBRC. W2R ) in the notation of Proposition 6. the relay-to-destination links can be used here solely for signal forwarding in a way that every carried bit increases the achievable rates by one bit. . 2 Remark 2 One bit on the OBRC links from relay to Di can clearly increase the end-to-end rates Ri by at most one bit (see also [57]). Therefore. i = 1. Under the assumptions of Proposition 7. and this result cannot be improved by interference forwarding. 2 and by Ahlswede’s multi-letter characterization of the interference channel capacity region [3]. stripped of the OBRC. Achievability follows by sending independent messages (W1R . of rates ηR1 C (c2 1 PR1 ) and ηR2 C (c2 2 PR2 ) from sources S1 and S2 . information bits) over the OBRC.. respectively. as long as ηR1 C (c2 1 PR1 ) ≤ η1R C (b1 P1R ) and ηR2 C (c2 PR2 ) ≤ η2R C (b2 P2R ). by sending independent.

are optimal.14a) (3. (3.13) plays a key role. C (a2 12 P1 + P2 ) − C (P1 + a21 P2 )}. ηR1 C (c2 1 PR1 ) ≤ η1R C (b1 P1R ) and ηR2 C (c2 PR2 ) ≤ η2R C (b2 P2R ). Proposition 8 (Optimality of Joint Signal Relaying and Interference Forwarding. the following rate 2 2 2 Rex21 = min ηR1 C (c2 1 PR1 ) − η1R C (b1 P1R ). we next investigate the capacity region for the complementary scenario in which such condition is not sat2 isﬁed.29 2 2 2 i. i.e R1R = min{ηR1 C (c2 1 PR1 ). η2R C (b2 P2R )} = ηR2 C (c2 PR2 ). and 2 channel conditions a12 ≥ 1 and Rex21 ≥ max {0.14b) (3.e. 9. R1 ≤ C (P1 ) + η1R C (b2 1 P1R ) R2 ≤ C (P2 ) + ηR2 C (c2 2 PR 2 ) 2 2 R1 + R2 ≤ C a2 12 P1 + P2 + η1R C (b1 P1R ) + ηR2 C (c2 PR2 ). Propositions 8 and 9 assume conditions similar to strong interference for standard ICs [10]. Notice that the rationale behind this maximum rate allocation for signal relaying follows from the considerations in Remark 2. and 10). (Propositions 8.14c) which is achieved by joint signal relaying and interference forwarding. The following propositions provide conditions under which both signal relaying and interference forwarding. . and thus non-separable operation. η2R C (b2 P2R ) − ηR2 C (c2 PR2 ) (3. We focus speciﬁcally on two cases characterized by η1R C (b2 1 P1R ) ≤ ηR1 C (c1 PR1 ) 2 and ηR2 C (c2 2 PR2 ) ≤ η2R C (b2 P2R ). 1): For an IC-OBR Type-I with ﬁxed OBRC bandwidth allocation.. while Propositions 10 and 11 are formulated under conditions akin to the mixed interference regime [67]. Under the assumption at hand. the following inequalities characterize the capacity region. This can be interpreted as the excess rate from S2 to D1 on the OBR links once user 1 and 2 allocate the maximum possible rate on the OBRC channel for 2 2 signal relaying. η1R C (b1 P1R )} = η1R C (b1 P1R ) and R2R = 2 2 min{ηR2 C (c2 2 PR2 ).

R1R = ηR1 C (b2 1 P1R ) and R2R = ηR2 C (c2 2 PR2 ). what we are left with is the IC plus an orthogonal link from S2 to D1 of capacity Rex21 . so that the sum-rate bound (3.30 Proof 5 Appendix A. the sum-rate bound at D1 (see (A. i.e. After this. the assumptions in Proposition 8 encompass two diﬀerent situations. In Fig. 2 we have the channel conditions (a2 12 − 1)P1 + (1 − a21 )P2 ≤ 0. so that. we show the maximum achievable sum-rate of Proposition 6 for diﬀerent conﬁgurations of the OBRC channel gains and bandwidths. 2 condition Rex21 ≥ max {0. and thus it can be proved that transmission of only common information by S1 over the IC is optimal. we have 2 (a2 12 − 1)P1 + (1 − a21 )P2 > 0. conversely. Here.26) in Appendix A. Under the conditions of Proposition 8.26)). ηR1 C (c1 PR1 ) = ηR2 C (c2 PR2 ) = 1. To be speciﬁc. In the second case. For comparison. R2c = 0. it can be seen that the capacity region of Proposition 8 is attained without performing interference forwarding. we show the case η1R = η2R = ηR1 = ηR2 = 0.3. S2 can transmit only common information with the caveat that part of it will be sent also over the OBRC (non-separable operation).14c) to receiver D2 forms the performance bottleneck in terms of sum-rate irrespective of a positive excess rate Rex21 (which increases the sum-rate at D1 as per (A. Moreover.2. Therefore. D2 is in the strong interference regime (a12 ≥ 1) for this channel. thus falling within .. i. Moreover. 3.25).3) may be more restrictive than (A. Remark 5 To interpret the results of Proposition 8. 2 2 2 η1R C (b2 1 P1R ) = η2R C (b2 P2R ) = 2. We have a symmetric IC with P1 = P2 = 10 and a21 = a12 = a. As a result. In the ﬁrst. we ﬁrst consider a scenario where the relay-to-destination links form bottleneck with respect to source-to-relay links. it can be seen that it is optimal to exploit the excess rate Rex21 to perform interference forwarding 2 from S2 to D1 with rate equal to R2c = C (a2 12 P1 + P2 ) − C (P1 + a21 P2 ). allocate the maximum possible rate on the OBRC channel for signal relaying.e. C (a2 12 P1 + P2 ) − C (P1 + a21 P2 )} essentially guarantees that also D1 is in the strong interference regime in the channel at hand (IC plus capacity Rex21 ).

(3. As stated in the Proposition. the sum-rate shown in the sum-capacity is 4 bits/channel use of IC larger than the reference case of zero OBRC capacities.28) and strong (a ≥ 1) [10] interference regimes. a ≤ 0. here the excess rate is not large enough to make the sum-rate constraint (3. the following conditions characterize the capacity region R1 ≤ C (P1 ) + η1R C (b2 1 P1R ) R2 ≤ C (P2 ) + ηR2 C (c2 2 PR 2 ) 2 2 R1 + R2 ≤ C P1 + a2 21 P2 + ηR1 C (c1 PR1 ) + ηR2 C (c2 PR2 ). However. given . interference forwarding is useful in increasing the capacity region. Proposition 9 (Optimality of Joint Signal Relaying and Interference Forwarding. which falls under the conditions of Proposition 8 for a ≥ 1. The next result considers a scenario where the complementary upper bound is assumed. η1R C (b1 P1R ) = ηR2 C (c2 PR2 ) = 2. under the assumptions of Proposition 9. and the 2 2 conditions a21 ≥ 1. under the assumptions of Proposition 9. Moreover. for a ≥ 1.26) for D1 is always more restrictive than (3. It can be seen that the sum-rate increases by 2 bits/ch use of IC for all values of a. unlike Proposition 8.e. i.15a) (3. 0 ≤ Rex21 = ηR1 C (c2 1 PR1 ) − η1R C (b1 P1R ) ≤ C (a12 P1 + P2 ) − C (P1 + a2 21 P2 ).15b) (3. the sum-rate bound (A. Remark 6 Similarly to the second case described in Remark 5.31 the assumptions of Proposition 7.15c) which is achieved by joint signal relaying and interference forwarding. we consider 2 2 2 a situation with η2R C (b2 2 P2R ) = ηR1 C (c1 PR1 ) = 3.4. the sum-rate shown in the ﬁgure is the sum-capacity. Finally. In fact.26).14c) at D2 the performance bottleneck with respect to (A.5. from Proposition 7. it is known that in the noisy [44] (a(1 + 10a2 ) ≤ 0. 2): For a IC-OBR Type-I with ﬁxed OBRC bandwidth allocation.14c) for D2 in terms of the sum-rate. Proposition 8 assumes that the excess rate satisﬁes a given lower bound. Proof 6 Appendix A.

Proposition 10 (Optimality of Joint Signal Relaying and Interference Forwarding. showing that all interference is decoded and removed over the IC at D1 without resorting to interference forwarding. We ﬁnally remark that. separably over both IC and OBRC. whereas S2 transmits only common messages in a non-separable way over both IC and OBRC (interference forwarding).5. Remark 7 The assumptions in Proposition 9 imply the strong interference conditions a12 ≥ 1 and a21 ≥ 1. Remark 8 Unlike Propositions 8 and 9 where the overall IC-OBR operates in conditions similar to the strong interference regime of a standard IC. C (P1 ) + C the following condition characterizes the sum capacity R1 + R2 ≤ C (P1 ) + C P2 1 + a2 12 P1 2 + η1R C (b2 1 P1R ) + ηR2 C (c2 PR2 ). the optimal coding strategy for S1 is to transmit private messages only. 4): For an IC-OBR Type-I with ﬁxed OBRC bandwidth allocation . The capacity region of Proposition 2 9 is attained by setting R2c = Rex21 = ηR1 C (c2 1 PR1 ) − η1R C (b1 P1R ). P2 1+a2 12 P1 − C (P1 + a2 21 P2 ) . thanks to the excess rate.5. Proposition 11 (Optimality of Joint Signal Relaying and Interference Forwarding.32 2 the constraint Rex21 ≤ C (a2 12 P1 + P2 ) −C (P1 + a21 P2 ). which is achieved by joint signal relaying and interference forwarding. 3): For an IC-OBR Type-I with ﬁxed OBRC bandwidth allocation and the conditions a12 < 1. This is. the case if we further assume the 2 condition (a2 12 − 1)P1 + (1 − a21 )P2 ≤ 0 (as in the ﬁrst case discussed in Remark 5). however. Accordingly. for a21 ≥ 1+P1 . from the conditions in Proposition 10 and Appendix A. Proof 7 Appendix A. In contrast. Rex21 ≥ max 0. here D2 observes weak interference while D1 . 1+a2 12 P1 the constraint on the excess rate in Proposition 8 becomes Rex21 ≥ 0. the assumptions of Proposition 8 in general do not imply strong interference. is able to decode and strip-oﬀ the interference.

even with the help of the excess rate. P1 = P2 . The sum .33 2 and the conditions a21 ≥ 1. 3. under certain conditions bandwidth can be allocated only to the best channels and in a way that only signal relaying is used. which is achieved by joint signal relaying and interference forwarding. 0 ≤ Rex21 = ηR1 C (c2 1 PR1 ) − η1R C (b1 P1R ) ≤ C (P1 ) + C P2 1+a2 12 P1 − C (a2 21 P2 + P1 ). as opposed to Proposition 9. the achievable sumrate is limited by the rate at D1 in (A. and hence the IC operates in mixed interference regime. Proposition 11 essentially establishes the optimality of signal relaying and interference forwarding when one of the interfering links observes weak interference and the other strong interference. we discuss optimality of the considered strategies under variable OBRC bandwidth allocation.26).4 Capacity Results for OBRC Variable Bandwidth Allocation In this part. Therefore. 2 2 2 2 2 C (b2 2 P2R ) = C (c2 PR2 ). The next proposition shows that. Proposition 12 (Optimality of Signal Relaying Only): Consider an IC-OBR Type-I channel with variable OBRC bandwidth allocation and a12 = a21 ≥ 1. However. Remark 9 The situation in Proposition 11 is similar to the one in Proposition 9 in the sense that. thanks to the ability to allocate bandwidth among the source-to-relay and relay-to-destination channels. Proof 8 Appendix A. C (b2 P2R ) ≥ C (b1 P1R ) and C (c2 PR2 ) ≥ 2C (c1 PR1 ).3.6. the assumptions in Proposition 11 imply the conditions 1 ≤ a21 ≤ 1+P1 1+a2 12 P1 which in turn leads to 0 ≤ a12 ≤ 1. the following condition gives the sum capacity 2 2 R1 + R2 ≤ C P1 + a2 21 P2 + ηR1 C (c1 PR1 ) + ηR2 C (c2 PR2 ). Rex21 .

e. c1 = 3. η = 1 and all transmission powers set to 10dB. Signal relaying increases the sum-rate by the number of bits carried on C (c2 the OBRC links. under appropriate conditions. Consider an IC-OBR Type-I channel with b1 = 1. c2 = 1. a12 = 3. not necessarily restricting the channels to satisfy the 2 condition C (b2 2 P2R ) = C (c2 PR2 ). obtained from Proposition 6 (see (A. The following example answers this question in the aﬃrmative. Here. Parameters are ﬁxed as η = 1.2 ≤ b2 ≤ 1.4 shows the achievable . The result in Proposition 12 begs the question as to whether interference forwarding can be ever useful in the case of variable bandwidth allocation. a21 = 2. However. b2 is varied. here we have focused on this simple case. in Fig. namely η 2 PR2 ). Proof 9 Appendix A.36) in Appendix A. Remark 10 Proposition 12 states that for a symmetric and strong IC.34 capacity is given by η 2 R1 + R2 ≤ C P1 + a2 21 P2 + C (c2 PR2 ) 2 and is achieved by signal relaying and transmitting only common information on the IC. c2 = 3.5. i. 3. 3. and the upper bound obtained from Proposition 5 (see (A.3 we compare the sum-rate achievable by transmitting only common information on the IC and using signal relaying (with optimized bandwidth allocation). to illuminate more clearly the gist of the main result.7). To obtain further insight into the result of Proposition 12. P1 = P2 = P1R = P2R = PR1 = PR2 =10dB and b2 is varied.7). Numerical results show for this example that the achievable rate matches the upper bound for most of the channel gains (except 1.8). it is optimal to allocate all the bandwidth η to the better path to perform signal relaying. a21 = a12 = 2.35) in Appendix A. Fig.7. when the path S2 − R − D2 is better than S1 − R − D1 (see conditions on bi and ci ). c1 = b1 = 2. Notice that it is possible to generalize Proposition 2 12 to arbitrary b2 and c2 .

Fig.4 shows similarly that interference-forwarding is instrumental in improving the achievable sum-rate for b2 ≥ 2.e.75). W1R ).3. However. for b2 < 2. This channel diﬀers from IC-OBR Type-I in that on the OBRC: (i ) The signals transmitted from S1 and S2 are superimposed at the relay. given in (A. W2R )) (see (A. i = 1.26) in Appendix A. transmission strategies at sources and relay. Propositions 8 and 9 for the ﬁxed bandwidth case. private signals are transmitted in a separable way over IC and OBRC.36) for the channel parameters at hand. transmitted . the excess rate Rex21 is positive and interference forwarding is potentially useful as discussed in Sec. 3. Also shown is the outer bound from Proposition 5. i.3) along with optimized bandwidth allocation (ηiR .4 Analysis of IC-OBR Type-II In this section.(W2c . we assume a four-way message split into two private and two common parts.23)-(A.3. such that. W2c . interference forwarding is not needed (Proposition 12 shows that indeed signal forwarding is optimal for b2 ≤ 0.2. as discussed in Remark 1. and more general. For the variable bandwidth case. 2. 2) or both signal relaying and interference forwarding ((W1c . 3. while one of the common messages is sent over both IC and OBRC (and. (ii ) Relay broadcasts to the destinations.35 sum-rate obtained from Proposition 8 (which was obtained from Proposition 6) by assuming transmission of common messages only over the IC and either signal relaying only ((Wic . WiR ). we investigate the IC-OBR Type-II channel described in Sec. 3. 3. Namely. i = 1. These aspects allow novel.3. For b2 ≥ c2 = 1. ηRi ). This is reﬂected by the results presented below that encompass scenarios and techniques that have a counterpart in the IC-OBR Type-I analysis and others that stand out as speciﬁc to IC-OBR Type-II. thus making the design of optimal schemes more complex than on the IC-OBR Type-I. To simplify the analysis. hence.2 for source and destinations. we still consider the same class of strategies described in Sec.

the relay may perform Decode-and-Forward (DF) or Compress-and-Forward (CF) and the sources may encode by using random codes or structured (e.g. The third bound is valid under strong IC conditions.36 in a non-separable way). For instance. we will not give a general achievable rate region but rather focus on speciﬁc conditions under which optimality of certain design choices can be assessed. lattice) codes [48] [47].4. 1): For an IC-OBR Type-II with ﬁxed OBRC bandwidth allocation. Proposition 13 (Outer Bound for IC-OBR Type-II. the capacity . we give three outer bounds on the capacity region of IC- OBR Type-II. the variety of possible approaches is remarkable. The ﬁrst bound requires weak interference conditions on at least one of the links over the IC. For this reason. for a12 ≤ 1 and c1 ≥ c2 . we investigate conditions under which signal relaying (and thus separable operation. Then. Even within this class of strategies. while the second holds for any channel gains. unlike IC-OBR Type-I.. as per the discussion above) or joint signal relaying/ interference forwarding (and thus non-separable strategies) are optimal. We ﬁrst give three outer bounds on the capacity region for IC-OBR TypeII.1 Outer Bounds In the following. 3.

16c) (3.16a)-(3.16e) (3. (3. an outer bound is given as above but with the union in (3. with 0 ≤ ξ ≤ 1 and ξ = 1 − ξ. Y1n |X2 ) + ηM AC C b2 1 P1R n 1 n n I (X1 . n R1 ≤ respectively. ηBC . 2.16f) by.17a) (3. Y1n ) + ηM AC C b2 1 P1R + b2 P2R n 1 n n I (X1 .37 region is included in the following region lim closure n n n p(xn 1 . an outer bound to the capacity region is given as in Proposition 13 by replacing the bounds in (3.16f) taken also with respect to all parameters ηM AC . 1 n n ) + ηBC C c2 . Proof 10 Appendix A. Y2n ) + ηBC C n 1 + c2 2 ξPR n where the union is taken with respect to multi-letter input distributions p(xn 1 )p(x2 ) that satisfy the power constraints 1/n n t=1 2 E[Xi.17b) .x2 )=p(x1 )p(x2 ).16c) and (3. Y2n |X1 ) + ηM AC C b2 2 P2R n 1 c2 2 ξPR n I (X2 .16b) (3.t ] ≤ Pi . Y1n |X2 ) + ηBC C (c2 1 ξPR ) n 1 n 2 I (X2 .16a) (3. such that ηM AC + ηBC = η. Y1n |X2 I (X1 1 PR n 1 n n R2 ≤ I (X2 ) + ηBC C c2 . Y2n |X1 2 PR . Proposition 14 (Outer Bound for IC-OBR Type-II.16f) R1 ≤ R1 ≤ R1 ≤ R2 ≤ R2 ≤ R2 ≤ 1 n 2 I (X1 . 2): For the IC-OBR Type-II with ﬁxed and variable OBRC bandwidth allocation and any channel gains.16d) (3. i = 1. With variable OBRC bandwidth allocation. R2 ): (3. and with respect to parameters ξ. Y2n ) + ηM AC C b2 1 P1R + b2 P2R n 1 n n I (X2 .8. i = 1. 2.0≤ξ +ξ ≤1 n→∞ (R1 .

as implicitly also done for IC-OBR Type-I. The next propositions extend this condition to a IC-OBR Type-II.18b) (3. due to the more . 3. we consider ﬁxed OBRC bandwidth allocation. we consider conditions under which signal relaying is found to be sum-rate optimal for ﬁxed bandwidth allocation. 3): For the IC-OBR Type-II with ﬁxed OBRC bandwidth allocation.4.18d).9. (3. in case the relay-to-destination links form the bottleneck on the OBRC.8. Proposition 7 showed that. an outer bound is given as above but with ηM AC = η in (3.18a)-(3. For the previous model.18a) (3. Optimality of Signal Relaying In this section.18c) (3.38 Proof 11 Appendix A. R1 ≤ C (P1 ) + ηM AC C (b2 1 P1R ) R2 ≤ C (P2 ) + ηM AC C (b2 2 P2R ) 2 2 R1 + R2 ≤ C (P1 + a2 21 P2 ) + ηM AC C (b1 P1R + b2 P2R ) 2 2 R1 + R2 ≤ C (a2 12 P1 + P2 ) + ηM AC C (b1 P1R + b2 P2R ). However. Proposition 15 (Outer Bound for IC-OBR Type-II. signal relaying achieves the entire capacity region. a21 ≥ 1 is given by.18d) With variable OBRC bandwidth allocation. Decode-and-Forward (DF) Relaying We start by focusing on Decode-and-Forward (DF) strategies at the relay.2 Capacity Results for Fixed OBRC Bandwidth Allocation In this section. Proof 12 Appendix A. an outer bound to the capacity region for a12 ≥ 1.

Remark 11 Proposition 16 involves two separate set of channel conditions for the OBRC. the interference conditions on the IC are mixed and the optimal transmission strategy over the IC turns out to prescribe transmission of only private information by S1 (given that D2 is in weak interference) and of only common information by S2 (given that D1 is in a strong interference condition). Proposition 16 (Optimality of Signal Relaying Only.20) with ξ = 1 − ξ ∗ . Moreover. and by 1+P1 1+a2 12 P1 and c1 ≥ c2 . a12 < 1.10. ηBC C c1 ξPR (3. 2 + min ηM AC C b2 1 P1R .19) R1 + R2 ≤ max + ηBC C 0≤ξ +ξ ≤1 C (P1 ) + C c2 2 ξPR 1 + c2 2 ξPR P2 1 + a2 12 P1 .39 complex structure of the IC-OBR Type-II.20) c2 2 ξ PR ∗ 1+c2 2 ξ PR ∗ 2 2 if ηM AC C (b2 1 P1R ) < ηBC C (c1 PR ) and ηM AC C (b2 P2R ) ≥ ηBC C . DF): In an IC-OBR Type-II with ﬁxed bandwidth allocation. where ξ ∗ is ∗ the optimal power allocation that maximizes the sum-rate (3. the results below only establish sum-rate optimality. they are limited to mixed interference [67] (Proposition 16) and strong interference conditions (Proposition 18). a21 ≥ capacity is given by the inequality R1 + R2 ≤ C (P1 ) + C 2 if ηM AC C (b2 1 P1R ) ≥ ηBC C (c1 PR ). The operation over the OBRC is instead diﬀerent under the two set of channel conditions. In both cases. but entails in both cases only signal forwarding (of independent information according to our separable approach. Proof 13 Appendix A. rather than to any interference channel. recall Remark 1). . the sum P2 1 + a2 12 P1 + ηBC C c2 1 PR (3.

3.6. Remark 12 Observing Fig. in the sense of Proposition 8 and Remark 5. The second set of conditions (under which the sum-capacity is (3. and is illustrated in Fig.19)) is given 2 2 by c1 ≥ c2 . the sum-rate optimal operation of the OBRC entails signal relaying for both sources.e. Notice that the same considerations given above regarding optimality of this point for the OBRC alone with DF apply here.5.. since the maximum rate S2 − R 2 (i. Notice that this operating point on the OBRC (see dot in the ﬁgure) is sum-rate optimal if one focuses on the OBRC alone and on DF. one could guess (erroneously. 3.19)) is char2 acterized by c1 ≥ c2 and ηM AC C (b2 1 P1R ) ≥ ηBC C (c1 PR ) and is illustrated in Fig. This set of conditions is thus akin to the one of Proposition 7 for the IC-OBR Type-I. The fact that such excess rate is not to be exploited for interference forwarding by the optimal scheme of Proposition 16 . since the corresponding achievable rate region is given by the intersection of the MAC and BC regions in Fig. In this case. Proposition 16 shows that such operating point is also optimal for communications over the IC-OBR under the given conditions.6. the optimal strategy in terms of sum-rate is for only user 1 to transmit over the OBRC using signal forwarding. 3. as Proposition 16 shows) that signal relaying is suboptimal under the conditions of Fig. 3. between S2 and D1 . Notice that this view is consistent with the assumption of DF at the relay. This rate pair is characterized by the power split ξ ∗ in Proposition 16.5.40 The ﬁrst set of conditions (under which the sum-capacity is (3..e. 3.6 and recalling Propositions 8-10. while the opposite is true for the path S1 − R − D1 . ηBC C (c2 PR )). It corresponds to the case where the sources-to-relay MAC S1 − R constitutes the bottleneck with respect to the relay-to-destinations BC. ηM AC C (b2 1 P1R ) < ηBC C (c1 PR ) and ηM AC C (b2 P2R ) ≥ ηBC C c2 2 ξ PR ∗ 1+c2 2 ξ PR ∗ . ηM AC C (b2 2 P2R )) is larger than the maximum rate R − D2 (i. These conditions seem to suggest an “excess rate”. and transmission over the OBRC at rates given by the operating point indicated in the ﬁgure. Here. not just source 1 as above.

DF): In a symmet√ ric IC-OBR Type-II with ﬁxed bandwidth allocation.22c) 2 R1 + R2 ≤ min C (P1 + a2 21 P2 ).2. We now consider strong interference conditions on the IC. we brieﬂy show that signal relaying may be (asymptotically) optimal also in combination with CF at the relay.11. Proposition 17 (Optimality of Signal Relaying Only.22b) . Proposition 18 (Optimality of Signal Relaying Only. (3. the sum capacity is given by R1 + R2 ≤ C (P + a2 P ) + ηM AC C (2b2 Ps ). the following rates are achievable via signal relaying and CF at the relay: R1 ≤ C (P1 ) + ηM AC C R2 ≤ C (P2 ) + ηM AC C b2 1 P1R 1 + σ2 b2 2 P2R 1 + σ2 2 b2 1 P1R + b2 P2R 1 + σ2 (3. 1+a2 12 P1 1+P1 . Proof 14 Appendix A. Remark 13 The sum-rate (3.21) (3. Compress-and-Forward (CF) Relaying Here. C (a12 P1 + P2 ) + ηM AC C . see Sec. 1+a2 12 P1 assumed in Proposition 16.4. already guarantees strong interference condition at D1 .22a) (3. for 1 ≤ a ≤ 1 + P and ηBC C (c2 PR ) ≥ ηM AC C (2b2 Ps ).21) is achieved by transmission of only common information over the IC (due to the strong interference conditions) and symmetric-rate private messages (signal relaying) over the OBRC. and thus no we will show that interferenceforwarding is instead instrumental in maximizing the sum-rate. For a21 < 1+P1 . CF): In an IC-OBR Type-II with ﬁxed bandwidth allocation. 3. since the condition a21 ≥ need for interference forwarding.41 can be interpreted in light of Remark 8.

a21 ≥ 1 and c1 .18d) on the sum-rate given in Proposition 15 for a symmetric IC-OBR Type-II channel as a function of c with a = 2.22c) and the outer bound (3. as shown by Proposition 18. Proof 15 See Appendix A. We observe that the achievable sum-rate and outer bound are not only asymptotically equal for large c. consider the conditions in Proposition 16.22a)-(3. .22c) is achieved by transmitting common information over the IC. for a12 ≥ 1. P = Ps = 10. given by (3. The above provides the capacity region.18d). Fig.18a)-(3. 1+a2 12 P1 as in Proposition 16.42 where σ 2 satisﬁes σ2 ≥ 2 1 + b2 1 P1R + b2 P2R ηBC ηBC min (1 + ηM AC c2 1 PR ) . but in practice become very close already for c ≥ 6. when the interfering link from S2 to D1 . and private information (signal relaying) over the OBRC.9. which leads to its optimality in the strong interference regime. As discussed in Remark 11. η = 1. Speciﬁcally.7 shows the comparison of achievable sum-rate obtained from (3. signal relaying is optimal in terms of sum capacity and there is no need to exploit the “excess rate” between S2 and D1 . Optimality of Interference Forwarding In this section. if we further have that a21 ≥ 1+P1 . The relay performs CF which becomes optimal asymptotically as the relay-to-destination’s BC quality improves. c2 → ∞. However.22a)-(3. 3. 3. illustrated in Fig.6 (also recall that we assume a12 < 1). The rate (3. b = 1.18a)-(3. we discuss conditions under which strategies based on interference forwarding are optimal for IC-OBR Type-II. Decode-and-Forward (DF) Relaying We start from the scenario discussed in Remark 11 following Proposition 16. (1 + ηM AC c2 2 PR ) −1 .

P2 1 + a2 12 P1 2 + ηM AC C (b2 1 P1R ) + ηBC C (c2 PR ). 1+a2 12 P1 then interference forwarding is potentially useful. DF): In an IC-OBR Type-II with ﬁxed bandwidth allocation. This is shown here by ﬁrst providing an achievable region that involves joint signal relaying and interference forwarding. Remark 14 The scheme achieving (3. the sum-capacity is achieved by such scheme and given by R1 + R2 ≤ C (P1 ) + C Proof 16 Appendix A. and b2 . .23b) (3.23a)-(3.23a) (3. Proposition 19 (Asymptotic Optimality of Joint Signal Relaying and Interference Forwarding.12.23c) (3.43 a21 . is not strong enough to allow decoding and cancellation of the interference on the IC to D2 . This scheme is shown to be sum-rate optimal if weak interference is seen at D2 and a large “excess rate” (in the sense of Remark 11) is available between S2 and D1 so as to essentially drive D2 in the very strong interference regime. a21 < 1+P1 . and ﬁnally discussing some numerical results. the following conditions give an achievable region R1 ≤ C (P1 ) + ηM AC C (b2 1 P1R ) R2 ≤ C P2 1 + a2 12 P1 + ηBC C c2 2 ξPR 1 + c2 2 ξPR c2 2 ξPR 1 + c2 2 ξPR (3. while S2 transmits common information over the IC and both common and private on the OBRC (joint signal relaying and interference forwarding). i. c1 → ∞.23d) is based on transmitting only private information over the IC and OBRC (signal relaying) by S1 .23d) 2 R1 + R2 ≤ C (P1 + a2 21 P2 ) + ηBC C (c1 ξPR ) + ηBC C 2 2 R1 + R2 ≤ C (P1 + a2 21 P2 ) + ηM AC C (b1 P1R + b2 P2R ) with ξ + ξ ≤ 1 via DF and joint signal relaying and interference forwarding. Moreover.e. for a12 < 1. then establishing its asymptotic optimality under given conditions that include the scenario discussed above.

Fig. 3. As shown in [48] [47]. since communication between the two can take place with vanishingly small power. Similar conclusion applies to S2 and D1 .44 To investigate the role of interference forwarding in a non-asymptotic regime. c1 . . the IC in the original model) is a two-way Gaussian channel [1]. rather than the usual randomly-generated codes.1. Due to the latter assumption. Consider a symmetric IC-OBR Type-II with a → ∞. Appendix A.78.1. The OBRC gains are set to b1 = 1. and the sum-rate upper bound obtained from Proposition 13 and given in (A.. can be equivalently seen as a single terminal. Wic ) on the IC is asymptotically optimal in terms of equal rates R = R1 = R2 for a symmetric IC-OBR Type-II.23d). while the second (the OBRC) is a two-way relay channel [38] [48] [47]. The asymptotic optimality derived in Proposition 19 is here shown to be attained for ﬁnite values of b2 . all node powers are equal to 10 dB and ηM AC = ηBC = 1.8}.23a)-(3. For a21 ∈ {0.8 shows the sum-rate obtained from (3. codes. Note that for a21 = 1.e. b2 = 10. S1 and D2 .e.9.9} ≤ 1. speciﬁcally lattice. which is due to the fact that the S2 − D1 pair can exploit more excess rate. c2 = 1 and c1 is varied. 0.78 the condi- tions given in Proposition 12 are satisﬁed and signal relaying alone is optimal. communications over a two-way relay channel can beneﬁt from structured. 0. The proposition below shows that using structured codes to transmit common message splits Wic on the OBRC and transmitting private and common messages (Wip . the equivalent system now has only two terminals wishing to communicate with one another over two parallel channels: The ﬁrst (i. R2c = 0 in the achievable region given in Appendix A.. We also have a12 = 0. Lattice Coding We ﬁnally brieﬂy provide a second scheme that is asymptotically (in the sense deﬁned below) optimal and is based on interference forwarding.12) or also interference forwarding. Therefore.5 and S2 − D1 channel gain takes the values a21 ∈ {0.10. the advantages of interference forwarding becomes substantial with increasing c1 .8 ≥ 1+P1 1+a2 12 P1 = 1.74). by assuming that source 2 either uses only signal relaying (i.1.

5 as given in (3. the lattice coding allows the OBR to broadcast the modulo-sum of the source signals.21).13. the destinations decode the messages received over the relay-to-destination links by using the messages received and decoded via the IC. (3. (3. 3. We observe that DF achieves the sum-capacity for c ≥ 1.21) achievable with the DF scheme of Proposition 13. On the other hand. We have b = 1.9 compares the rate (3. in the structured codes discussed. 3. the received signals over the IC helps the destinations decode the messages broadcasted by the relay. lattice coding yields an achievable sum-rate that approaches the outer bounds.22a)-(3. Proof 17 Appendix A. Also. 3.5 and all node powers are set to 10 dB. as opposed to the schemes discussed before where the OBRC is exploited in decoding and canceling the messages transmitted via the IC.17a)(3. ηM AC = ηBC = 0.22c) achievable with CF scheme of Proposition 18 and rate (3. In Fig.24) ηM AC 2 log 1 2 + b2 Ps ≥ . As discussed in [48] [47] and mentioned above brieﬂy.18a)-(3. 10 and 11) and Type-II (Proposition 19).17b)) and (Proposition 15.24) achievable by lattice coding with the outer bounds (Proposition 14.9 i) we set a = 3 and in Fig. (3.9 ii) a = 10 and study rates as a function of c. Proposition 20 (Asymptotic optimality of Interference Forwarding Only): The symmetric capacity of a symmetric IC-OBR Type-II channel for ηBC C (c2 PR ) and a → ∞ is given by R ≤ C (P ) + ηBC C c2 PR and is achieved by interference forwarding via structured codes.45 Using structured codes for IC-OBR Type-II incorporates interference forwarding which is structurally diﬀerent than its counterparts as discussed in Type-I (Propositions 8.18d)) for a symmetric IC-OBR Type-II. as the interference gain a increases. Therefore. rate (3. conﬁrming the results in Proposition 20. Fig. 9.

3 and all nodes powers are 10 dB and η = 1.25) for ξ = 1 − ξ . from the conditions (3. c1 = 2.4. ηBC C (c1 ξPR ) 2 + min ηM AC C (b2 1 P1R + b2 P2R ). In Fig. ηBC ).16b) and (3. if c1 and b2 are suﬃciently large. R1 + R2 ≤ C (P1 ) + C P2 1 + a2 12 P1 2 + min ηM AC C (b2 1 P1R ). with a12 < 1. c2 = 0. we obtain the following outer bound. We consider a mixed interference scenario with a12 = 0. the sum-rate discussed above are shown for variable S1 − R gain. the DF rate (3.3 Capacity Results for OBRC Variable Bandwidth Allocation Here. it is optimal to use DF by letting the relay .8 and c1 ≥ c2 . We know from the ﬁrst part of Proposition 12 that if b1 is suﬃciently larger than c1 . the total bandwidth is balanced between the S1 − R and R − D1 channels. Moreover. The second part of Proposition 16 proves that.19) where the relay helps the S1 − D1 pair only. The right part of the ﬁgure also shows the optimal bandwidth (ηM AC and ηBC ) and power (ξ ) allocation. In both cases. 3. and the other channel gains are set to b2 = 2.16d) and (3.46 3.16f) for R2 . A similar conclusion is drawn here for b1 ≥ 2 as it can be seen from the optimal power allocation ξ . ηBC C c2 2 ξPR 1 + c2 2 ξPR (3. a21 = 1. which satisfy all the conditions of Proposition 16 except the ones that depend on the bandwidth allocation (ηM AC . for ﬁxed bandwidth allocation. is optimal. for ﬁxed bandwidth allocation. b1 . we brieﬂy investigate the eﬀect of variable bandwidth allocation for the IC-OBR Type-II via numerical results.10. bandwidth allocation (ηM AC . Recall that for given conditions on ﬁxed allocations (ηM AC . In particular.5. ηBC ). Proposition 16 shows the optimality of DF with signal relaying (separable operation). We compare the performance of the DF scheme in Proposition 16 (separable transmission) with an outer bound obtained from Proposition 13. ηBC ) is optimized. This result is akin to Proposition 12 for IC-OBR Type-I.16c) for R1 and (3.

3. The relay facilitates interference forwarding by broadcasting W2c which is used at D1 to remove part of the interference. 3. Fig. It is seen that for (b2 . we have studied one such scenario where two source-destination . a12 = 0.47 help both source-destination pairs (see Remark 11). Speciﬁcally. c1 ) ∈ {(3.11 shows that a similar conclusion holds also when optimizing the bandwidth allocation. Note that since a21 ≤ 1+P1 1+a2 12 P1 = 1.11 compares the achievable sum-rate (3. 3. Fig. the outer bound and the achievable sum-rate match for b1 ≤ 7. 2). Fig. (10. A natural question that arises is to understand the eﬀect of interference forwarding for IC-OBR Type-II with variable bandwidth allocation. a21 = 1. (3. and hence interference forwarding is crucial for larger b2 gains.5. As shown in the ﬁgure.5 Chapter Summary Operation over parallel radio interfaces is bound to become increasingly com- mon in wireless networks due to the large number of multistandard terminals.78. c1 with values from the set (b2 .3. since the interference cannot be decoded and removed over the IC only at D1 . W2c ) only. and b2 . P1 = P2 = P1R = P2R = PR =10dB and b2 is varied. the achievable scheme follows from Proposition 19. In this paper. c1 ) large enough.5. To observe this eﬀect. D1 can be aided by the relay’s interference forwarding. In the ﬁgure.12 essentially shows that this is indeed the situation especially for b2 ≥ 2. 5).23a)-(3. where S1 transmits only private information W1p via the IC whereas S2 transmits common information (W2c . This enables cooperation among terminals across diﬀerent bandwidths and possibly standards.25) for variable b1 . c2 = 1.4.20) (attained by the DF scheme just discussed) with the outer bound (3. the increase in b2 gain helps the OBR forward more interference to D1 . 3. 10)}.23d). b1 = 1. and other conditions as above. η = 1. similar to its IC-OBR Type-I counterpart as discussed in Sec. c1 = 4. We also have c2 = 1. we consider a Type-II channel with the parameters set to η = 1. (20. 3.

joint signal relaying and interference forwarding. and thus non-separable transmission. interfering over a given bandwidth. scenarios that extend the current model to more than two sources and one relay are of interest. This clearly complicates the design. These conditions have also been related to the problem of assessing optimality of either separable or non-separable transmission over parallel interference channels. this is shown to be the case for both ﬁxed and variable (i. Moreover. in the presence of optimized OBRC bandwidth allocation. a separable approach has been shown to be optimal. for both considered models. However. which eases interference mitigation. relaying can assist interfering communications via standard signal relaying but also through interference forwarding. An example of such cases. cooperate with a relay over an orthogonal spectral resource (out-of-band relaying. For both considered models. and expected to oﬀer new research challenges in light of the results of [75]. . Moreover. in some special scenarios of interest. is necessary to attain optimal performance. Overall. optimized) bandwidth allocation over the OBRC. separable schemes are often (but not always) optimal. this paper has derived analytical conditions under which either signal relaying or interference forwarding are optimal.. is the case where the relay has better channel conditions from the sources than to the destination. As discussed in previous work. We have focused on two diﬀerent models that correspond to distinct modes of transmission over the out-of-band relay channel (OBRC). In particular. The analysis in this paper leaves open a number of problems related to interference management via cooperation and through multiple radio interfaces.48 pairs. the analysis shows that.e. OBR). in general.

ηRi . channel uses each.1: Interference Channel (IC) with an out-of-band relay (OBR). ηBC channel uses.49 Z1 X1 X1 X1R b1 1 Z1R + Y1 D1 + ZR1 ZR2 a12 Y1R R Y2R a21 1 + + + XR1 XR2 c1 YR1 b2 Z2R c2 + X2 X2R X2 YR2 D2 Y2 Z2 Z1 X1 X1 X1R b1 ZR 1 a12 c1 + + + ZR1 ZR2 Y1 D1 YR1 YR R XR c2 b2 X2 X2R X2 a21 1 + Z2 + YR2 D2 Y2 Figure 3. (ii ) IC-OBR Type II: The OBRC is divided into two orthogonal channels with ηM AC . 2. (i ) IC-OBR Type-I: The OBRC is divided into four Gaussian orthogonal channels with ηiR . i = 1. . The OBR channel (OBRC) has η channel uses for each channel use of IC.

5 1 1. 2 2 2 b2 1 = b2 = 1.1 1 1 2 1 2 Achievable Sum−Rate with η=2. (which satisﬁes the conditions in Proposition 7).50 8 7 Rate (bits/ch use of IC) Noisy 6 Sum−Capacity 5 4 3 2 1 0 0 Strong Capacity Regime Achievable Sum−Rate with η=0 Achievable Sum−Rate with η=2. b2=c2=6. c2=c2=0.5 3 3. namely i) η = 0 (no relay). c1 = c2 = 0.5 2 2.5 (which satisﬁes the conditions in Proposition 8). for ﬁxed OBRC bandwidth allocation versus a for various OBRC link capacities. All powers are set to 10 dB. 2 2 2 and iii) η1R = η2R = ηR1 = ηR2 = 1.1.5. .3.3. b2 2 = c1 = 6. b2=b2=1. b1 = c2 = 1. b2=c2=1.5.5 2 2 1 0. ii) η1R = η2R = ηR1 = ηR2 = 1.2: Achievable sum-rate from Proposition 6 for IC-OBR Type-I with symmetric IC.5 a Figure 3.

51 3. c1 = b1 = 2).5 3 3. c2 = 3. a21 = a12 = 2.5 4 1 Figure 3.1 3.15 Rate (bits/ch use of IC) 3.85 0 Achievable Sum−Rate Outer Bound 0. (A. P1 = P2 = P1R = P2R = PR1 = PR2 =10dB.2 3.95 2. .9 2.5 1 1.25 3.36)) versus b2 for IC-OBR Type-I with variable OBRC bandwidth allocation (η = 1.3: Achievable sum-rate (from Proposition 6.5 2 b 2.05 3 2.35) with signal relaying (only common information transmission over the IC). and outer bound (from Proposition 5 (A.

Rel. Rel. c1 = 3.5 1 1.7 0 Achievable Sum−Rate(Sig. η = 1.36)) versus b2 for IC-OBR Type-I with variable OBRC bandwidth allocation 2 i=1 ηiR + ηRi = η.5 2 b2 2.&Int.9 2. and c2 = 1.) Outer Bound Achievable Sum−Rate(Sig.) 0.5 4 Figure 3.8 2. b1 = 1. P1 = P2 = P1R = P2R = PR1 = PR2 =10dB. a21 = 2.85 2. (A. (A.23)-(A.75 2.4: Achievable sum-rate (from Proposition 8.52 3.95 2.26)) with signal relaying and interference forwarding (only common information transmission over the IC) and outer bound (from Proposition 5.5.05 3 Rate (bits/channel use of IC) 2.5 3 3. . Forw. a12 = 3.

6: Illustration of the OBRC conditions leading to the sum-capacity in Propo2 2 sition 16: c1 ≥ c2 .D2 BC R1 ¢£¤¦§¨¦© ¡¢£¤¥¦§¨© Figure 3. ηM AC C (b2 P2R ) ≥ ηBC C c2 2 ξ PR ∗ 1+c2 2 ξ PR ∗ where where ξ ∗ is the optimal power allocation that maximizes the sum-rate in Proposition 16 (ξ = 1 − ξ ∗ ).5: Illustration of the OBRC conditions leading to the sum-capacity in Propo2 sition 16: c1 ≥ c2 .D2 BC R1 ¡¢£¤¥¦§¨© ¢£¤¦§¨¦© Figure 3.53 R2 S1.S2 to R MAC R to D1. ηM AC C (b2 1 P1R ) ≥ ηBC C (c1 PR ). ηM AC C (b2 1 P1R ) < ηBC C (c1 PR ). R2 ¢£¤§§¨§© ¡¢£¤¥§§¨© 45o line S1.S2 to R MAC ¢£¤§§¨§© ¡¢£¤¥§§¨© 45o line R to D1. ∗ .

22c)) and outer bound ((3.8 0 2 4 6 8 10 Achievable Sum−Rate (CF) Outer Bound c Figure 3.8 2.18d)) for a symmetric IC-OBR Type II channel with respect to relay-to-destination channel gains.4 Rate (bits/ch use of IC) 3.6 3.2 2 1. .6 2.2 3 2. P = Ps = 10).18a)-(3.22a)-(3.4 2.7: Achievable sum-rate ((3. c (a = 2.54 3. b = 1.

a21=0.4 3. Sum−Rate (Sig.8} (a12 = 0. a21=0. Sum−Rate (Sig.1. a21=0.6 1 3 5 7 Ach.).Forw.4 4.1 Ach.8 2.2 Rate (bits/channel use of IC) 4 3.8 Outer Bound 9 11 13 15 17 19 20 c 1 Figure 3. 0.1. b1 = 1.9 Ach.2 3 2.).8 3.1 Ach. b2 = 10. Sum−Rate (Sig.).6 3. Sum−Rate (Sig. c2 = 1 and all node powers are equal to 10).).8: Achievable sum-rate and outer bound for an IC-OBR Type-II channel with respect to R − D1 channel gain. .).Forw.9.Rel. & Int.Rel.Rel. Sum−Rate (Sig. a21=0.9 Ach.5.Rel. c1 and S2 − D1 channel gain a21 ∈ {0. a21=1.Rel.55 4. & Int.

5 2 c 2.6 Rate (bits/channel use of IC) 3. i) a = 3.4 2.5 1 1. 16) Compress−and−Forward (Prop.2 2 0 0.4 3.5. 14) Outer Bnd (Prop.5 4 3.6 2.9: Achievable sum-rates and outer bound for a symmetric IC-OBR Type-II with respect to the relay-to-destination gain c (b = 1.8 2. ηM AC = ηBC = 0. Fig. P1 = P2 = P1R = P2R = PR =10dB.5 3 3.5 3 3. 18) Lattice Coding (Prop.2 3 2. 20) Outer Bnd (Prop.6 2.4 2.5 1 1. . 11) c Figure 3.56 3.6 Rate (bits/channel use of IC) 3. ii) a = 10).5 2 2.2 2 0 0.5 4 Decode−and−Forward (Prop.2 3 2.4 3. Fig.8 2.

5.19)) with signal relaying (DF) and outer bound (3.6 Rate (bits/channel use of IC) 1.8 BC ηMAC 0. a12 = 0. ξ ) of the DF scheme (3.3. a21 = 1.8 0. η = 1.57 1.2 0 Achievable Sum−Rate Outer Bound 2 4 b1 6 8 10 1 Parameters (Achievable Sum−Rate) ξ η 0. all node powers are equal to 10 dB.6 0.2 0 0 2 4 b1 6 8 10 Figure 3.10: Achievable sum-rate (from Proposition 16.25) and optimal parameters (ηM AC . c2 = 0. ηBC .8.2 1 0.4 0.) .6 0.19) for an IC-OBR Type-II with respect to S1 − R channel gain b1 (b2 = 2. c1 = 2.4 1.4 0. (3.

c =5 Outer Bound: b2=10. a12 = 0. (10.8. 2).11: Achievable sum-rate (from Proposition 16.) . ηM AC + ηBC = η with η = 1.58 Achievable Sum−Rate: b =3. c =2 2 1 Rate (bits/channel use of IC) 2. (3. c1 ) ∈ {(3.5.4 1.25) for an IC-OBR Type-II with respect to S1 − R channel gain.6 1. c1=5 Achievable Sum−Rate: b =20.2 1 0.2 2 1.8 Outer Bound: b =3.c1=10 2 1 Achievable Sum−Rate: b =10. b1 and (b2 . c =2 2 1 b2=20.c1=2 0 1 2 3 4 5 6 7 8 b 1 Figure 3. 10)} (c2 = 1.8 1. c =10 2 1 b2=3.c1=5 Outer Bound: b =20.19)) with signal relaying (DF) and outer bound (3. c =10 2 1 b2=10. all node powers are equal to 10 dB. 5). a21 = 1. (20.

Forw.23a)-(3.12: Achievable sum-rate (from Proposition 19. (η = 1.8 2.2 3 2.23d)) with signal relaying and interference forwarding (DF.5. Relay) Outer Bound 1 2 3 4 b2 5 6 7 8 Figure 3. Relay & Int. (3.4 3.25) versus b2 for IC-OBR Type-II with variable OBRC bandwidth allocation 2 i=1 ηiR + ηRi = η. .4 2. b1 = 1).2 0 Achievable Sum−Rate (Sig. P1 = P2 = P1R = P2R = PR =10dB.) Achievable Sum−Rate (Sig. c2 = 1. c1 = 4. a21 = 1. a12 = 0.6 2.59 Rate (bits/channel use of IC) 3.5. only common information transmission by S2 and only private information transmission by S1 over the IC) and outer bound (3.

60 Chapter 4 Interference Channel with an Out-of-band Reception/In-band Transmission Relay 4. IC-OIR in short. multistandard nodes are equipped with diﬀerent standards that are in general operated on diﬀerent bands. We model this scenario with an interference channel assisted by an an out-of-band reception/in-band transmission relay. we are interested in a model where the relay (multistandard node) transmits in the band of multi-terminal nodes however receives in orthogonal bands.1 Introduction In heterogenous networks. . In conjunction with the practical example given for IC-OBR. we considered a scenario with an out-of-band relay helping multiple source-destination pairs and coined the investigated system with the name IC-OBR. one can consider such a case where the relay receives through Wi-ﬁ frequencies from the cellular nodes. In this part. In the previous chapter. cooperation among independent terminals that belong to diﬀerent standards/subsystems is a possible way to improve the system performance as discussed in the previous section. and transmits to the base-stations using the frequency bands of the cellular nodes. However. This feature gives multistandard nodes the ﬂexibility of choosing among various bands in helping other nodes to maximize the beneﬁt obtained through cooperation in the network.

we investigate optimal operations at the sources and the relay.5 is devoted in determining the very strong relay-interference conditions and the corresponding capacity region for IC-OIR. is to explore various relaying operations and their eﬀect on the overall system performance. 2nRi ]. In the model of interest.t } is independent identically distributed (i.8. wishes to send a message index Wi . Each source Si . Z2 are independent.4.9. . uniformly drawn from the set [1. and {Zi. Section 4.7 investigates a degraded IC-CR and provides its capacity region. n ). the relay receives the source signals over links that .) Gaussian noise with unit variance. In Section 4. 4.2 System Model We consider a Gaussian interference channel (IC) with a relay shown in Fig. we investigate interference channel with a cognitive relay (IC-CR) under mixed relay-interference regime and determine its sum capacity.1.t ∈ R. similar to the IC-OBR model.6. Also Z1 . 2. 4.. we are interested in determining relaying operations which can not be utilized by the out-of-band transmission at the relay but in-band transmission and explore their signiﬁcance on improving the transmission rates. We conclude this section in Section 4. Numerical results illustrating the general achievable region for the IC-OIR and capacity results under very-strong and mixed relay-interference conditions are given in Section 4. The rest of the section is organized as follows. we give a general achievable region IC-OIR and two outer bounds on the capacity region and sum-capacity. respectively. To gain more insight on the model. In Section 4. In particular.2. i = 1. to its destination Di .t codeword.d. Section 4. The sources S1 and S2 communicate simultaneously with their respective destinations D1 and D2 via the Gaussian IC. t = 1. with the help of the relay R.1..i..61 Our aim. The (real) input symbols of source Si are denoted as (Xi. on which we enforce the power constraint 1/n n t=1 2 ≤ Pi for each Xi. The system model is introduced in Section 4.

t + c1 XR..2) (4. The encoding function at the relay R is given by. Note that XR. 2nCi ]. where the messages Wi . (4. i = 1. A (2nR1 .t + a22 X2.1b) which maps a message Wi ∈ [1. n to the destinations simultaneously with the sources. 2nRi ] → Rn × [1.t . 2nRi ] into a codeword X n ∈ Rn for the IC and a message to the ﬁnite-capacity relay link Vi ∈ [1. are known at the relay non-causally. 2. 2nR2 ] → Rn . We assume in-band relay transmission where the relay transmits its symbols XR.62 are orthogonal to each other and to the underlying IC.. We deﬁne Vi ∈ [1.1a) (4. n are given by.t = a11 X1. i = 1.. V2 ).. Hence. fr : [1.. 2nR2 . 2.t + Z2. We implicitly assume that Si − R channels are used n times before transmission over IC takes place. t = 1.t Y2.t .t + a21 X2. the received signals at the destinations at time t = 1. The orthogonal link from Si to the relay has capacity Ci bits/channel use. hence it can be classiﬁed as an interference channel with an out-of-band relay reception/in-band relay transmission relay [66]. (4. i = 1.t is a function of (V1 . Y1. 2. decoding . i = 1. . 2nCi ].t + Z1. 2nCi ]. n) code for the interference channel with a ﬁnite capacity relay (IC-OIR) is deﬁned by the encoding function at the source Si . Finally.t ≤ PR . . The relay power constraint is represented as 1/n n i=1 2 XR. This scenario will be called interference channel with a cognitive relay (IC-CR).3) n which maps the messages V1 and V2 into a relay codeword XR . 2. 2nR1 ] × [1. i = 1.t = a12 X1. where the channel coeﬃcient aij are real.t + c2 XR. When Ci → ∞. as the messages sent by the source Si over the ﬁnite-capacity link to the relay. the model reduces to the cognitive scenario of [41] [59] [53]. 2: fi : [1..

i = 1.5b) (4. (4. Cl(S ) and Co(S ) deﬁne the closure and convex hull operations over the non-empty set S . Deﬁnition 3: A rate pair is said to be achievable for the IC-OIR if there exists a sequence (2nR1 . Deﬁnition 4: The capacity region C for the IC-OIR is the closure of the set of all achievable rate pairs.3 A General Achievable Region and Outer Bounds for IC-OIR In this section. we deﬁne a degraded interference channel with a cognitive relay. Also the notation C (x) = 1/2 log(x) is used throughout the chapter. 4. is. Yin . we obtain a general achievable rate region for IC-OIR and outer bounds on the capacity region of IC-OIR and IC-CR. Deﬁnition 5: In this chapter. 1 . Pe. a12 = a11 a21 = a22 c1 = c2 .2 }. which maps the received signal over the IC.i 1 2n(R1 +R2 ) (W1 . 2nR2 ) codes with Pen → 0 as n → ∞. into the estimated message W As a special case.W2 )∈W1 ×W2 ˆ i = Wi Pr W for i = 1. respectively. 2nRi ].5c) Deﬁnition 2: The average probability of error for this code.4) ˆ i. Deﬁnition 1: An interference channel with a cognitive relay is said to be degraded if it satisﬁes the conditions. is deﬁned as n n Pen = max{Pe.63 function at destination Di . 2. (4. gi : Rn → [1.5a) (4. . 2. where n Pe.

we further split the common and private parts as follows. we perform message splitting to serve the following purposes: (i) to reduce the eﬀect of interference at the destinations as in the Han-Kobayashi scheme [9]. (ii) to exploit the relay via the ﬁnite-capacity links using partial decode and forward [25]. the private part is divided into three splits.4 An Achievable Region To obtain an achievable region for IC-OIR. To incorporate the relay into the transmission. Note that the messages transmitted directly to the destinations . each source message is split into common and private parts. the ﬁrst split is transmitted directly. where the common parts are decoded at both destinations and the private parts are decoded at the desired destination only.64 Z1 X1 X1 X1R C1 1 c1 + Y1 D1 a12 R XR C2 c2 X2 X2R X2 a21 1 + Z2 Y2 D2 Figure 4. and (iii) to use partial knowledge of the interference at the relay in transmitting additional information as in dirty-paper coding [12]. the second split is transmitted with the help of the relay.1: Gaussian interference channel with an out-of-band reception/in-band transmission relay (IC-OIR). 4. Therefore. and the third part is transmitted only by the relay to the destinations using partial knowledge of the interference. The common messages are split into two parts such that ﬁrst part is transmitted directly to the destination and the second part is transmitted with the help of the relay. On the other hand.

2nRip ] is the private message transmitted by Si and R jointly. (ii) Wic ∈ [1. .6a) √ √ (4. Wip ). .6b) 2 ai1 αs Pi + c1 ξs PR Nt1 ai2 αs Pi + c2 ξs PR Nt2 √ √ + s∈S1b 2 a2 i1 αs Pi (4.6e) (4...6c) 2 Rs ≤ C s∈S2a ∪S2b s∈S2a + s∈S2b 2 a2 i2 αs Pi (4. 1p.... 1c}.. 2c}. Wip . S1b ⊆ {1c. where (i) Wic ∈ [1.2.. . i=1. R2 = R2c + R2c + R2p + R2p + R2p . with the assumption that the message splits Wic . In the encoding of Wip .. The message splitting can be represented as Wi = (Wic . . Wic . Wip . are decoded error-free at the relay for the given rate constraints. (iv ) Wip ∈ [1... R2 ) . . we have the following achievable region for IC-OIR. 2nRip ] is the private message transmitted by R only.6d) R1p ≤ C R2p ≤ C 2 c2 1 ξ1p PR 2 a2 21 α2p P2 + 1 (4. Based on this strategy. 2nRip ] is the private message transmitted by Si only. 1p. Wip . with R1c + R1p + R1p ≤ C1 R2c + R2p + R2p ≤ C2 Rs ≤ C s∈S1a ∪S1b s∈S1a (4.. 1c}... 2nRic ] is the common message transmitted by Si and R jointly. 2c}. Wip . (iii) Wip ∈ [1. 2. i = 1. Proposition 21 : An achievable rate region for a Gaussian interference channel assisted by an out-of-band reception/in-band transmission relay (IC-OIR) is given by ach = Cl Co R1 .. 2.6f) 2 c2 2 ξ2p PR 2 2 2 a2 12 α1p P1 + c2 ξ1p PR + 1 and S1a ⊆ {1c. .. the relay performs dirty-paper coding and hence transmits Wip to the destinations interference free from the splits Wic . 2nRic ] is the common message transmitted by Si only. and (v ) Wip ∈ [1.65 need not be decoded at the relay. i = 1. S2b ⊆ {2c. S2a ⊆ {2c. 2p. 2p. where R1 = R1c + R1c + R1p + R1p + R1p .

s∈B 2 ξs ≤ 1 where B = {1c. We can obtain another achievable region by switching the roles of D1 and D2 . 1p} and ai1 = a21 . Pi = P2 otherwise. i. 1c. Remark 15 The rate constraints in (4. . ξ2c ≥ 0. 1p. 2p. On the other hand. ξ2c . Positive correlation for the common splits. Remark 17 Note that the relay allocations ξ1c .66 √ √ 2 2 2 2 2 2 2 Nt 1 = c 2 1 (ξ1p + ξ2p )PR + a21 α2p P2 + (a21 α2p P2 +c1 ξ2p PR ) + 1.14. We denote this relay operation as interference cancelation.6e) and (4. the power allocations α R 1 12 1 p 1 2 1 p p 12 1 p 2 2 2 2 satisfy αic + αic + αip + αip ≤ 1. The notation for channel and power allocations are ai1 = a11 . while D2 views the power allocation to W1p as noise. ai2 = a12 . where C1 . W2p ) at the relay in delivering additional information at reduced interference to D1 . Remark 16 The achievable region in Proposition 21 includes the achievable regions of the IC-CR system given in [41]. ξ1d and ξ2d allow for the relay to negatively and positively correlate its signal with the sources. ai2 = a22 . i=1. 2p}. Pi = P1 for s ∈ {1c.e.2. 2c. negative correlation at the relay for the private splits decreases the received power of the private signals at the destinations. Moreover. improves the received power of these splits at both destinations. 1 → 2 and 2 → 1 in (4. This region can be simply obtained by interchanging R1p and R2p . 1p. ξ1c . 1p. i.e. Denoting this region as ach .6e) and (4.6f). which may be beneﬁcial at the interfered destination. C2 → ∞ and the sources transmit private messages only. Interference cancelation was used to improve the achievable rates for the asymmetric interference channel with a relay in improving the achievable rates in [59]. C2 → ∞ and the sources employ general rate-splitting. Proof 18 Appendix A. Nt2 = c2 (ξ1p + √ √ 2 2 2 ξ2 ) P + a P + ( a α P + c ξ PR )2 + 1. and IC-CR [53] with C1 . ach ∪ ach gives an extended achievable region.6f) are obtained by using the knowledge of (W2p . hence can be attributed to signal-forwarding [55] [72].

Hb is rb × t and x n Y1 . C2 → ∞. R) pair if. a12 ≥ a11 a21 ≥ a22 . yb is rb × n. Noting that for Gaussian IC-OIR. t = 2. the capacity region is known for strong conditions such that. 4. is t × n input matrix and za is ra × n noise matrix. These outer bounds are used to establish: (i) the capacity region of IC-OIR under very strong relay-interference conditions. µ ≥ 1.1 Outer Bounds for the IC-OIR and IC-CR In this section. 4. In [22] (and references therein). we say n T n that in a Gaussian IC-OIR.7) The degradedness deﬁnition is used alternately in the IC literature. where ya is ra × n. In obtaining the capacity region under strong regime. Deﬁnition 5 [33]: A receive vector ya = Ha x + za is said to be degraded with respect to yb = Hb x + zb if there exists a matrix D such that DHb = Ha and such that DDT I. [a21 c1 ] = µ[a22 c2 ]. for the input signal x = [Xn 2 XR ] . Ha = [a22 c2 ] and Hb = [a21 c1 ] with D = µ−1 being a constant. 4. Y2 is a degraded version of Y1n with respect to (S2 . in Sec.67 4. we use the vector degradedness1 conditions deﬁned in [33].5. (ii) the sum capacity of IC-CR under mixed relay-interference conditions.6. In an analogous manner. our degradedness deﬁnition is diﬀerent and used in ordering the destinations in terms of signal reception.a21 = 1 only. we give two outer bounds on the IC-OIR. [10] uses the decodability of the interference signals at the destinations to obtain the corresponding outer bounds. Sec. Ha is ra × t. However. and (iii) the capacity region of a degraded IC-OIR as given in Sec. .4.7. to extend the strong conditions to IC-OIR. ya yb n Y2 . we have ra = rb = 1. while the second bound is derived for C1 . The ﬁrst bound is valid for any C1 ≥ 0 and C2 ≥ 0. 1 (4. a Gaussian interference channel is said to be degraded for a12 . For a regular Gaussian interference channel.

R) pair for an enhanced system if.9) (R1 . ξic ≤1.10b) P2 PR (4.a = 2 2 2 2 2 α2 ic ≤1.10g) + 2a21 c1 α2c ξ2c P2 PR 2 2 2 R1 + R2 ≤ C1 + C2 + C a2 11 (1 − α1c )P1 + a21 (1 − α2c )P2 Proof 19 The proof uses the notion of degradedness in Deﬁnition 5 adopted to the IC-OIR and given in Appendix A. ROIR out.7) particularly in obtaining outer bounds on the sum capacity. Y1n is said to be a degraded version of Y2n with respect to (S2 .15. (4.8) In the following. the following gives an outer bound on the capacity region. R2 ) : P1 PR (4. [a21 c1 ] = µ[a22 c2 ].10a) (4. i=1.68 Similarly. (4. Proposition 22 : For an interference channel with out-of-band reception/in-band transmission relay (IC-OIR) satisfying the conditions.10e) 2 2 R1 ≤ C a2 11 P1 + c1 ξ1c PR + 2a11 c1 α1c ξ1c 2 R1 ≤ C1 + C a2 11 (1 − α1c )P1 2 2 R2 ≤ C a2 22 P2 + c2 ξ2c PR + 2a22 c2 α2c ξ2c 2 R2 ≤ C2 + C a2 22 (1 − α2c )P2 2 2 2 2 R1 + R2 ≤ C2 + C a2 11 P1 + a21 (1 − α2c )P2 + c1 ξ1c PR + 2a11 c1 α1c ξ1c 2 2 R1 + R2 ≤ C a2 11 P1 + a21 P2 + c1 PR + 2a11 c1 α1c ξ1c P1 PR (4. [a12 c2 ] = µ[a11 c1 ].10f) .2 (4.10d) P1 PR (4. µ ≥ 1.10c) (4. where we employ the degradedness condition in (4. µ ≥ 1. we obtain outer bounds for the IC-OIR. . α1c ξ1c +α2c ξ2c ≤1.

it is not necessary for the sources to transmit information to the destinations directly as in IC-OIR model.b is an outer bound on the capacity region of IC-OIR where the union is taken 2 2 2 2 2 2 over all parameters αic ≤ 1. Note that since the relay already knows the sources messages. µ ≥ 1.12) . µ ≥ 1.10g) as 1 → 2 and 2 → 1.12)-(4. W2 non-causally. (4. however. Then for the conditions [a12 c2 ] = µ[a11 c1 ]. (4. The following corollary will be used in proving the sum capacity of IC-CR under mixed relayinterference conditions as given in Proposition 5. Corollary 2 For an interference channel with out-of-band reception/in-band transmission relay (IC-OIR) and an interference channel with a cognitive relay (IC-CR) satisfying the conditions. 2. ξic ≤ 1. or similarly C1 . Denote this region as ROIR out. An analogous model to IC-OIR is the one investigated under interference channel with a cognitive relay (IC-CR) [41] [59] [55] [53]. i = 1. the relay is assumed to have the sources messages W1 . it is possible to show that another outer bound on the capacity region of IC-OIR can be obtained from Proposition 22 by simply switching the indices in (4. α1 c ξ1c + α2c ξ2c ≤ 1. C2 → ∞. [a21 c1 ] = µ[a22 c2 ].69 Remark 18 Using symmetry.b . In this case. all messages are transmitted with the help of the relay.11) ROIR out.

Now. Remark 19 As discussed in the previous remark. the desired destination is assumed to be degraded w. and the outer bound is denoted by RCR out. Similarly.t. R2 ) : a11 a22 a11 P1 + c1 ξ1c P2 + c2 ξ2c P1 + c1 ξ1c PR PR PR 2 R1 ≤ C R2 ≤ C R1 + R2 ≤ C (4. R pair.a = 2 +ξ 2 ≤1 ξ1 c 2c (R1 . 2. using symmetry. we can obtain another outer bound from Corollary 2 by switching the indices which is valid for the conditions [a12 c2 ] = µ[a11 c1 ].13a) (4. or. i = 1. by noticing that for C1 . The rationale of this outer bound is somewhat aligned with the bound in Proposition 22.10g) is maximized for αic = 1. µ ≥ 1.r.15) (4.13c) Proof 20 Follows immediately from Proposition 22. however. RCR out.b . [a22 c2 ] = µ[a21 c1 ]. for the following degradedness conditions. we give another outer bound on the sum-capacity of IC-OIR which is valid for C1 .70 the following gives an outer bound on the capacity region. (4. C2 → ∞. C2 → ∞.10a)-(4.13b) + a21 P2 + c1 ξ2c PR 2 2 2 .14) Y1n is said to be degraded with respect to Y2n . namely for IC-CR. µ ≥ 1. from the perspective of S2 . . the region (4. (4. interfered destination instead.

as in Figure 4.16) 2 (4. when the interfering links are suﬃciently strong such that the destinations are not aﬀected by the interference. Therefore. Proposition 24 The sum capacity of two parallel source-destination pairs sharing a prl 2 2 2 =maxα2 common relay with orthogonal links as shown in Figure 4.5 Very Strong Capacity Region of IC-OIR In [5]. the following gives an outer bound on the sum capacity. in an IC-OIR. Under the so-called very strong interference conditions. √ √ 2 a11 P1 + c1 ξ1c PR R1 + R2 ≤ 2 max C √ √ 2 ≤1 ξ1c +ξ2 1 + a21 P2 + c1 ξ2c PR c + C Proof 21 Appendix A. Carleial shows that in an interference channel when the interfering links are suﬃciently strong. the interference can be decoded ﬁrst and removed without degrading the point-to-point achievable rates.2. the interference channel capacity is equivalent to a pair of point-to-point channels. Note that each parallel relay channel is similar to the model investigated in [25] whose capacity is characterized.ξ1c +ξ2c ≤ .2 is given by Csum 1c ≤1.71 Proposition 23 For an interference channel with out-of-band reception/in-band transmission relay (IC-OIR) and an interference channel with a cognitive relay (IC-CR) satisfying the conditions. a22 P2 + c2 ξ2c PR 2 (4.17a) 4. On the other hand. it is possible to characterize the sum-capacity of this system which further depends on the relay power split among the sources.16. the system becomes equivalent to two parallel source destination pairs sharing a common relay with orthogonal links. [a22 c2 ] = µ[a21 c1 ]. µ ≥ 1.α2c ≤1.

. The 2 2 total relay power satisfy ξ1 c + ξ2c ≤ 1.18d) (Si . where 2 2 R1 ≤ C a2 11 P1 + c1 ξ1c PR + 2a11 c1 α1c ξ1c 2 R1 ≤ C1 + C a2 11 α1c P1 2 2 R2 ≤ C a2 22 P2 + c2 ξ2c PR + 2a22 c2 α2c ξ2c 2 R2 ≤ C2 + C a2 22 α2c P2 2 2 2 2 with α1 c = 1 − α1c . Di ). The following deﬁnition determines the channel conditions in an IC-OIR which guarantee the decodability of the interference at the destinations without reducing the point to point rates. (R1 + R2 ). and α2c = 1 − α2c . ξ2c the optimal allocations in Csum .18a) (4. ξ1c .2: Parallel source-destination pairs with a common orthogonal relay. we denote by α1 c .2. hence transforming an IC-OIR to the system with parallel sourcedestination pairs with a common relay as in Figure 4. on the other hand.18b) P2 PR (4. i = 1. This is an equivalent system for an IC-OIR under very strong interference. gives the sum capacity of IC-OIR under very strong relay-interference conditions. We denote these conditions as very strong relay-interference conditions.72 Z1 W1 S1 X1 R R a11 Power: ξ1c2PR + Y1 D1 a R1 Power: ξ2c2PR aR2 + Z2 W2 S2 X2 a22 Y2 D2 Figure 4. 2 source-destination pair. ∗ ∗ ∗ ∗ prl Now. with the relay allocating ξic PR power for each P1 PR (4.18c) (4. α2c . 2 Proof 22 Follows directly from [25]. The proposition.

19d) give us the very strong interference conditions of a Gaussian interference channel [5]. for αic = ξic = 0. Proposition 25 In an IC-OIR satisfying the very strong relay-interference condiOIR prl .18d) is clearly maximized for αic = ξic = 0. tions. the capacity region of Cvsi given in (4. Similarly. a12 ≥ a12 a21 a21 1 ∗ √ (a11 α1 c P1 √ ≥ a11 γD2 1 ∗ √ ≥ (a22 α2 c ∗ α2c P2 √ ≥ a22 γD1 ∗ α1 c ∗ P1 + c1 ξ1 c √ ∗ PR ) γD2 − c2 ξ1 c √ ∗ PR ) γD1 − c1 ξ2 c PR (4. i = 1. hence giving the capacity region. and the region becomes the capacity region of a Gaussian interference channel under very strong interference regime. 2.19a)-(4.18a)-(4. . the conditions (4. i = 1.19b) ∗ P2 + c2 ξ2 c PR (4. ν ∈ [0. the sum capacity is given by Cvsi = Csum Proof 23 Appendix A. Also. Remark 20 It is possible to obtain the capacity region of IC-OIR under very strong relay-interference conditions following similar steps used in obtaining the sum capacity.19c) (4.19d) √ √ ∗ ∗ ∗ 2 where γDi = 1 + a2 ii αic Pi + (aii αic Pi + ci ξic PR ) . OIR Remark 21 For C1 = C2 = 0.19a) (4. 1]. which transform the system into the system in Figure. Note that it is possible to characterize the whole capacity region of the system shown in Figure 4.17.2 by obtaining the optimal source and relay power allocations that maximize R1 + νR2 .73 Deﬁnition 6: An IC-OIR is in very strong relay-interference regime if. one can obtain the corresponding channel conditions for IC-OIR evaluated at these optimal allocations. 2.

Note that since the relay is cognitive. for the conditions of Proposition 23). Proposition 22.74 4. namely conditions in Proposition 23 and conditions in Proposition 22 adopted for D2 . on the other hand. Proposition 26 The sum-capacity of an interference channel with a cognitive relay (IC-CR) operating under mixed interference channel conditions is given by. all source signals are .20b) The following proposition gives us the sum capacity of IC-CR under mixed relay interference conditions. R1 + R2 ≤ + min C C max 2 2 a22 P2 + c2 ξ2c . C2 → ∞. gives an upper bound to the sum-rate of the IC- CR where the interference is treated as noise at one of the destinations (D1 . provides an outer bound which is obtained by considering decodability of the interference (again at D1 for the conditions of Proposition 22).20a) (4.17a). [a12 c2 ] = µ[a11 c1 ] [a22 c2 ] = µ[a21 c1 ]. Combining the conditions of the two propositions. µ ≥ 1. Deﬁnition 7: An interference channel with a cognitive relay is in mixed relay-interference regime if. C1 . leads us to the channel conditions which can be called as mixed relay-interference regime.C PR 2 √ √ 2 a11 P1 + c1 ξ1c PR √ √ 1 + a21 P2 + c1 ξ2c PR ξ1c +ξ2c ≤1 2 √ √ 2 a12 P1 + c2 ξ1c PR √ √ 1 + a22 P2 + c2 ξ2c PR 2 (4. (4.6 Sum-Capacity of IC-CR under Mixed RelayInterference Conditions Proposition 23.21) Proof 24 Achievability: Achievability follows directly from the general achievable region in Proposition 21. (4.

22) and (4. the decodability of the interfering signal at the destinations leads to the determination of the capacity region of the channel [10]. whereas S1 transmits the private message. only. αs = 0 where s ∈ {1c. we obtain the sum-rate (4. where a degraded IC-CR is deﬁned Sec. . with R1 + R2 = R1c + R2p .b and valid for (4.5a)-(4. only.17a) give us the rate in (4.20a). the following proposition gives the capacity region. i. R1 + R2 ≤ C a22 P2 + c2 ξ2c PR 2 ξ1c and ξ2c ξ2p .b can be written equivalently as. the conditions in (4. Then. Moreover.7 Capacity Region of a Degraded IC-CR In this section. 4. so that we have α1c = 1 and α2p = 1. W1c . 2c.75 transmitted with the help of the relay. in this particular achievable scheme. 1p. Another outer bound which is valid for (4. the minimum of (4.20b) follows from Proposition 23 and given in (4. S1 transmits the common message. For a degraded IC-CR. When a cognitive relay is incorporated into the system. we assume ξ1c Converse: The ﬁrst outer bound follows from Corollary 2 and Remark 19. which is denoted by RCR out.2. W1p .6a)(4.6a)-(4. hence the sum-capacity.5c) similarly give necessary conditions for the decodability of the interfering signals at the destinations. no dirty-paper coding is employed at the relay.6f). Also.21). and ξ1p = ξ2p = 0. 2 4. √ √ 2 a12 P1 + c2 ξ1c PR +C √ √ 1 + a22 P2 + c2 ξ2c PR 2 (4. Moreover. since RCR out.e. In a regular Gaussian interference channel. we establish the capacity region of a degraded IC-CR.17a). from (4. for notational convenience. Again.21). 2p} in (4.6f).22) 2 2 with ξ1 c + ξ2c ≤ 1.

i.23d).8 Illustration of the Results In this section. Wic . i = 1. Rout. followed by the relay beamforming with the messages.76 Proposition 27 : The capacity region of a degraded IC-CR is given by. with the parameters set to αic 1. we illustrate the results in Section 4. We ﬁrst illustrate the capacity region under strong relay-interference conditions. 2.23a) (4. This particular achievable scheme corresponds to the sources trans- mitting common messages only. R2 ) : P1 PR P2 PR P1 PR (4.b as discussed in the Remark 19 of Corollary 2. . 2 4.3-4. C CR = 2 +ξ 2 ≤1 ξ1 c 2c (R1 . CR Converse: The converse uses the outer bound given in Corollary 2.5 through some nu- merical examples. 2 2 2 and ξ1 c + ξ2c ≤ 1.23a)-(4.23c) P1 PR (4.23d).a and CR the bound Rout.e. ξip = ξip = 0.23a)-(4. From Proposition 21.23d) (4. Proof 25 Achievability: The achievability follows directly from the general achievable region in Proposition 21. Taking the minimum of these two outer bounds we obtain the expression in (4.23b) 2 2 R1 ≤ C a2 11 P1 + c1 ξ1c PR + 2a11 c1 ξ1c 2 2 R2 ≤ C a2 22 P2 + c2 ξ2c PR + 2a22 c2 ξ2c 2 2 R1 + R2 ≤ C a 2 11 P1 + a21 P2 + c1 PR + 2a11 c1 ξ1c + 2a21 c1 ξ2c P2 PR 2 2 R1 + R2 ≤ C a 2 12 P1 + a22 P2 + c2 PR + 2a12 c2 ξ1c + 2a22 c2 ξ2c P2 PR . i = 1. we obtain the rate region in (4.

i = 1. In Figure 4.6 0. a11 = a22 = 1.2 C1R.4 0. and P1 = P2 = PR = 10.2 0 0 0. a21 = a12 = 2.4 1.77 2 1. C1 = C2 = 0.4 0.2 0. however with the sources transmitting common messages only. (Wic . the relay starts helping both source-destination pairs with partial decode-and-forward transmission.e.2 1 0. The C1 .3.8 1 C1R.6a)(4.8 1. we consider a symmetric IC-OIR with a11 = a22 = 1.C2R=0. The achievable region follows from Proposition 21.6 C1R.2.2 1. respectively. c1 = c2 = 1. If we .8 0.C2R ³ 2 R2 (bits/ch use) 1.8 2 R1 (bits/ch use) Figure 4. Wic ) and αip = αip = 0. For C1 = C2 = 0. and the capacity region is increased compared with C1 = C2 = 0.4 1. The system parameters are P1 = P2 = PR = 10. whose capacity region is given by the corresponding compound multiple-access channel. we obtain a Gaussian interference channel operating under strong interference. i. a12 = a21 = 2. c1 = c2 = 1. (4. C2 values.3: Achievable region of a symmetric IC-OIR with diﬀerent C1 . As the source-to-relay link capacities are increased to C1 = C2 = 0.2 and C1 = C2 ≥ 2.C2R=0 1.6 0. C2 → ∞ curve also corresponds to the IC-CR model.6 1. 2. The ﬁgure shows the achievable region of this IC-OIR for source-to-relay capacities of C1 = C2 = 0.6f).

i = 1. C2 → ∞ and therefore investigate an IC-CR channel. (Wic .5 2 Figure 4. i.e. C2 .5 0 0. Hence. . we see that for large enough C1 and C2 values. the capacity region remains the same.5 Rate (bits/ch use) 2 Private only Common only Private+dpc Rate splitting+dpc 1.78 2.5 1 0. In Figure 4. further increase C1 . In fact. various transmission techniques used in characterizing the achievable region in Proposition 21 are demonstrated in terms of their maximum sum-rate for a symmetric IC-CR. The system parameters are P1 = P2 = PR = 10.6f). i = 1. a11 = a22 = 1.6a)-(4.4: Achievable sum-rate of a symmetric IC-CR for diﬀerent transmission techniques. for C1 = C2 ≥ 2. instead. we let C1 . IC-OIR is equivalent to IC-CR. 2. this corresponds to the sources transmitting no message splits directly. which are clearly maximized for αic = αip = 0. 2. Wip ). all message splits are conveyed via the relay.5 µ 1 1. c1 = c2 = 1. we observe that for C1 = C2 ≥ 2. The system parameters are P1 = P2 = PR = 10.4. the rate expressions with no C1 and C2 become bottleneck. Physically. µ = a21 /a22 = a12 /a11 . from the achievable region (4. In the ﬁgure.

private only corresponds to αic = αic = αip = 0. a11 = a22 = 1. hence D1 is corrupted by the interference from S2 . c1 = 1. c1 = c2 = 1.5: Achievable sum-rate of a one-sided IC-CR for diﬀerent transmission techniques. common only corresponds to αip = αic = αip = 0. 2. a12 = c2 = 0. ξip = ξip = 0. i = 1.79 9 8. The system parameters are P1 = P2 = PR = 10.5 6 5.5 7 6. In order to understand the role of relaying in an interference limited system more clearly. i = 1. P1 = P2 = PR = 10.5 Rate (bits/channel use) 8 7. we also consider a one-sided interference channel. and also µ = a21 /a22 = a12 /a11 which essentially determine the relative strength of the interference in the system. i = 1. a12 = c2 = 0. a11 = a22 = 1. and rate splitting along with dirty-paper coding corresponds to fully optimizing the achievable region in Proposition 21. 2. µ = a21 /a22 . with a cognitive relay helping the interfered destination only. We have. private+dpc corresponds to αic = αic = αip = 0. ξic = ξip = 0. 2. In the ﬁgure. and the relay . c1 = 1. a11 = a22 = 1. ξic = 0.5 5 Private only Rate splitting Rate splitting+dpc 0 5 10 15 µ 20 25 30 Figure 4.

4 −0. i = 1.4. the eﬀect of dirty-paper transmission at the relay is seen to improve the sum-rate for moderate interference link ratios particularly. Similar to Fig. The system parameters are P1 = P2 = PR = 10. µ = a21 /a22 .80 1 0.5. has link to D1 only. 2.6 −0. Note that. Fig.6 Power allocations 0. 2.2 0 −0. a12 = c2 = 0.5 investigates maximum sumrate of this network for various transmission techniques. and including all power constraints. a11 = a22 = 1. In comparison with the one sided IC .2 −0. denoted by rate splitting+dpc.8 −1 0 5 10 15 20 ξ1p ξ1phat ξ2c ξ2p 25 30 a /a Figure 4. For this ﬁgure we include the power allocations such as αic = αic = αip = 0. i = 1. c1 = 1. 4. we do not need rate splitting at this node) has substantial impact in improving the sum-rate. ξip = 0.4 0. also. 4. Therefore.8 0. that is denoted by private only. the only aim of the relay is to help S1 − D1 pair. denoted by rate splitting.6: Optimal relay power allocations that maximize achievable sum-rate of the one-sided IC-CR in Fig. 4. αic = αip = 0. ξic = ξip = 0. We observe that rate-splitting at S2 (since S1 does not have link to D2 . We set P1 = P2 = PR = 10.

An interesting property observed for very strong interference regime is also observed from .2.5252.6 shows the optimal relay power allocations corresponding to the best achievable scheme in Fig. which leads to the parameters set to αic = αip = 0. which is specialized for cognitive relay.19d) are given by. a12 = a21 = 4. we observe that presence of the relay necessities rate splitting for a large range of interference levels.19a)-(4. the relay only uses a small amount of power to forward S1 ’a signal transmitted over the IC. 4. for 0 ≤ µ ≤ 14. In the ﬁgure. As the interference ratio increases. (4. Fig. and is performed by negative beamforming with S2 ’s private message split at the relay. Interference cancelation.18d).7 considers a symmetric IC-CR with the system parameters P1 = P2 = PR = 10. i. and ξ2p is the relay power allocation for beamforming with the private information of S2 . and c1 = c2 = 0. the outer bound matches with the achievable sum-rate. The ﬁgure also shows that for a12 = a21 ≥ 5. Achievable sum-rate is obtained from Proposition 21. the relay’s functioning can be categorized as signal forwarding and interference cancelation. C2 → ∞. ξ1p is the relay power allocation used in transmitting dirtypaper coded signal to D1 . 2. Interference forwarding takes places by relaying the common information of S2 .6f). 4. the conditions (4. is observed to be as crucial as signal forwarding in these regimes. which is not transmitted by S1 to its destination directly using dirty-paper coding.81 without the relay. For the given system parameters.e. Signal forwarding is mostly performed by transmitting fresh information of S1 . a11 = a22 = 1. i = 1. the relay switches its functioning to interference forwarding .18a-4. ξ2c is the relay power allocated in beamforming with the common information of S2 .5. where private information transmission is optimal for µ < 1 and transmitting common information is optimal for µ ≥ 1. The outer bound is also shown which is given in (4.e. Figure 4. on the other hand. Note that in these regimes. as well as signal forwarding. C1 . i. ξ1p corresponds to the relay power allocation for S1 signal. whereas signal forwarding is done by improving the strength of S1 ’s signal.6a)-(4. We observe that for weak to moderate interference ratios.

Figure 4.82 3 2. while S2 transmits private. a11 = a22 = 1. The system parameters are P1 = P2 = PR = 10.25 bits/ch use.8 1. Interestingly. a12 = a21 = 0.75 bits/ch use.7.6f) and outer bound (4. S2 common message only. the achievable sumrate is given by R1 + R2 = 2.4 2. for no interference case.18a-4. 4.8 Achievable Sum−Rate Outer Bound Rate (bits/ch use) 2. i.8 shows two scenarios. each destination is able to decode the interference fully.6 0 1 2 3 µ 4 5 6 7 8 Figure 4. hence observes interference-free signals from the sources and the relay. for no interference case. the destinations still suﬀer by the interference occurred by the relay’s broadcast signal. we study the mixed relay-interference regime. and S1 transmits private message only.6 2. S1 transmits common message only. According to Proposition . Finally. Note that. the achievable sum-rate and the capacity is R1 + R2 = 2.5 bits/ch use. however for very-strong interference.e. interference helps the IC-CR to improve its sum-rate around 0. whereas for a12 = a21 ≥ 5.7: Achievable sum-rate (4.2 2 1. and c1 = c2 = 0.18d) of a symmetric IC-CR with respect to µ = a21 /a22 = a12 /a11 .6a)-(4. The reason behind this reason is the fact that.2 Fig.

83

3.5 3 2.5 2 1.5 S1 common msg. only, S2 priv. msg. only 1 0.5 0 S1 priv. msg. only, S common msg. only 2 Sum Capacity

Rate (bits/ch use)

0

0.5

µ

1

1.5

2

Figure 4.8: Achievable sum-rate and sum capacity of an IC-CR operating under mixed relay-interference regime. The system parameters are P1 = P2 = PR = 10, a11 = 1, c1 = 1, a21 = 1, µ = a22 /a21 = a12 /a11 = c2 /c1 . 26, the former is optimal (in terms of sum rate) when µ > 1, while the latter is optimal for µ < 1. These results are also conﬁrmed by Figure 4.8.

4.9

Chapter Summary

Interference is a main paradigm in communication systems and often un-

avoidable. Understanding the role and beneﬁt due to cooperation/relaying in interferencelimited systems is therefore essential. In this chapter, we have investigated an interference channel with a relay where the relay is connected to the sources via ﬁnite capacity links. Similar to our previous work, we characterized relevant relaying functions for the model such as interference forwarding, interference cancelation and signal relaying by considering the ﬁnite-capacity links from the sources to the relay. Basi-

84 cally, in-band transmission of the relay and the sources lead to additional interference cancelation opportunity, which is not possible for the relay operating out-of-band to the sources. Also, interference forwarding is shown to be fundamental as the interference at the destinations become larger, and the optimality is shown for very-strong relay-interference conditions. Finite capacity source-relay links entail additional message-splitting at the sources to maximize the achievable region. However, we showed that for large enough ﬁnite-capacities, the system becomes equivalent to a cognitive relay scenario such that the relay has a-priori messages of the sources. This substantially decreases the system complexity by reducing the message splitting. This model is open to various optimality conclusions such as strong or noisy capacity regimes. Also, it may be interesting to understand the situations where the relaying operations discussed lead to the optimal performances.

85

Chapter 5 Interference Channel with an In-band Reception/In-band Transmission Relay

5.1

Introduction

In this chapter, we investigate another model for relaying in interference-

limited systems with the emphasis on single band allocation for the whole network. As opposed to the previously analyzed models IC-OBR and IC-OIR, in this case, the relay receives and transmits in the communication band of the multiple sourcedestination pairs. The in-band reception/tranmission relaying operation can be example of cooperation in an ad-hoc network, where the destinations are prone to the interference as a result of simultaneous transmission at the transmitter nodes. To uncover the beneﬁts of relaying, we assume a Gaussian channel with a full-duplex relay that employs decode-forward (DF) type strategy. This chapter is organized as follows. In Section 5.2, the system model is given. We present an achievable rate region for the Gaussian interference relay channel using rate splitting in Section 5.3. In Section 5.4, the orthogonal transmission is considered. Section 5.5 investigates sum-rate maximization. Numerical results and discussions are presented in Section 5.6 and we conclude the chapter in Section 5.7.

In [5]. . Here subscript R denotes the relay.n represents zero mean. V = [V1 . 2. unit variance Gaussian noise with N (0. respectively. 5.n = where a12 . .e.86 5.n Y1.n √ √ = a12 X1. i = 1. The inputs to the channel at time n are denoted as Xi.n + X2. 2. nm. .n √ √ = X1.n . . D1 . bi is the channel gain between Si and R and ci is the channel gain between R and Di .3 Rate Splitting at the Sources Our achievable region for the interference relay channel follows the rate splitting technique of Carleial at the sources [5] and incorporates the relay. R. 2.n + nD2. . The bold letters deﬁne vectors of length N . Also. We assume that the relay is full-duplex and employs decode-and-forward type strategy [17]. i.n Y2. . and C (x) = 1 2 5.n + c2 XR. i = 1. The transmission is performed over N channel uses and each transmitting node has a power constraint Pi .n and Yi.n . i = 1.n + b2 X2.1. . 1) which is independent for diﬀerent m and n. and the outputs as YR. Vn .n . it was shown that dividing the message index set as common and private indices and transmitting the superposition of the “subcodewords” formed by these splits increases .n + nR. D2 and S2 . 2 and XR. i = 1. 2 simultaneously utilizing the relay. In the sequel we use the standard deﬁnitions of achievable rates and probability of errors [17]. 2 wishes to communicate with Di .n + nD1. i = 1. . Terminal Si . i = 1.2 System Model We consider a Gaussian interference relay channel with two source-destination pairs as shown in Fig.n + a21 X2.n b1 X1. YR.n + c1 XR. a21 are the channel gain between the nodes S1 . VN ] log2 (1 + x). The relation between the channel inputs and outputs can be written as follows.

The relay on the other hand decodes both common and private message indices from the sources and transmits them after properly allocating its power among the common and private messages of S1 and S2 . Similarly. the following conditions describe an achievable rate region. . t ∈ [1c. . US1 . Pk = P2 if √ √ t ∈ [2c. . U2c . the following propositions give an achievable region of Gaussian interference relay channel. wip ∈ {1. 2c. Pk = P1 if t ∈ [1c. US ) 2 2 (5. YR |U1c . .1a) (5. 1p].87 Z1 X1 b1 ZR 1 a12 YR a21 1 R XR c2 c1 + Y1 D1 + b2 X2 + Z2 Y2 D2 Figure 5. Ri i∈S c ≤ I (XS . 2N Ric }.1b) (5. XR ) c c ≤ I (XS1 . 1p. Under this strategy. U1p . US2 .1c) Ri i∈S1 Ri i∈S2 with Xt = αt Pk /PR Ut . YD2 |XS . for the interference relay channel. YD1 |XS . i = 1. 2 such that wic is decoded at both destinations whereas wip is decoded only at Di . US ) 1 1 c c ≤ I (XS2 . . we allow the sources to form common and private index sets as wic ∈ {1. XS . . 2p] and XR = ξ1c U1c + ξ2c U2c + ξ1p U1p + ξ2p U2p . 2N Rip }. . U2p . . The random variables √ ˜t + αt X . Proposition 28 For the Gaussian interference relay channel.1: Gaussian Interference Relay Channel with two source-destination pairs the achievable region of the interference channel for some conditions. 2p].

2f) (5. 2p}. X2c .2d) (5. j ∈ S } and XS denotes the inputs for the complement of S . the following describe the set of achievable rates R10 R1c R1c R1c 1 ≤ C (2b1 α1c P1 ) 2 S1 R α1c.2c) (1 + a21 )α1c. X2p ∼ N (0.tot P1 + c1 ξ1c PR + ρ1 c ≤C S2 R 1 + a21 α1p.2g) R1p R1p R1c + R1p S1 R S2 R 2c1 ξ1c PR + ρ1 c + ρ1c + S2 R 1 + a21 α1p. U1p . XS denotes {Xj .2a) (5.2b) (5. The summation indices are.18. 1p}. P2 ). U2c . that is {Xj . The power allocation variables satisfy α1c + α1c + α1p + α1p = 1.tot P1 + c1 ξ1p PR + ρ1 p 1 ≤ C (2b1 α1c P1 + 2b1 α1p P1 ) 2 . S2 ⊆ c {1c. S ⊆ {1c. b1 = b2 . 2c. X1p ∼ N (0. 2c. 2p}.tot P1 + c1 ξ1c PR + ρ1 c S2 R 1 + a21 α1p. U1c .tot P1 + c1 ξ1p PR + ρ1 p ≤C 1 ≤ C 2 S2 R a21 α1c. c1 = c2 and a12 = a21 . Proof 26 Appendix A. j ∈ S c }. P1 ).tot P1 + c1 ξ1p PR + ρ1 p ≤C S2 R 1 + a21 α1p.tot P1 + c1 ξ1p PR + ρ1 p (5. X1c .2e) (5.tot P1 S2 R 1 + a21 α1p. 2c. 1p.88 have the distributions. α2c + α2c + α2p + α2p = 1 and ξ1c + ξ1p + ξ2c + ξ2p = 1. U2p ∼ N (0. Proposition 29 For the symmetric interference relay channel where P1 = P2 . PR ).tot P1 + c1 ξ1p PR + ρ1 p 1 ≤ C (2α1p b1 P1 ) 2 S1 R α1p. S1 ⊆ {1c.tot P1 + c1 ξ1p PR + ρ1 p (5.

α1p.89 S1 R S1 R P1 + 1 c P + ρ1 c + ρ1p 2 1 R S2 R 1 + a21 α1p. we analyze orthogonal transmission in terms of time division for the interference relay channel as well.tot P1 + α1p. .tot P1 + c1 ξ1p PR + ρ1 p S1 R S2 R S1 R 2c1 ξ1c PR + c1 ξ1p PR + ρ1 c + ρ1c + ρ1p (5.3d) R1c + 2R1p 2R1c + R1p 2R1c + R1p ≤ C (b1 α1c P1 + 2b1 α1p P1 ) ≤ C (2b1 α1c P1 + b1 α1p P1 ) ≤C + (1 + a21 )α1c.3c) (5. Proof 27 Due to symmetry. orthogonal transmission at the sources outperforms rate splitting in terms of sum rate for moderate interference [5].tot P1 + c1 ξ1p PR + ρ1 p S2 R 1 + a21 α1p. 5.3e) √ iR where α1c. We consider the case where in the ﬁrst time slot.3b) (5.4 Orthogonal Transmission For the interference channel even though rate splitting increases the achiev- able rate region for most channel conditions.3a) a21 α1c.tot P1 S2 R 1 + a21 α1p. S1 and R cooperatively transmit to D1 and in the second time slot. Also. 2. ρiR 1c = 2 α1c ξ1c ai1 c1 Pi PR and ρ1p = 2 α1p ξ1p ai1 c1 Pi PR for i = 1. the relay helps S1 and S2 equally.2a)-(5. Motivated by this.tot = α1p + α1p .tot P1 + α1p. α1c = α2c .tot = α1c + α1c .tot P1 + c1 ξ1p PR + ρ1 p (5.tot P1 + c1 ξ1p PR + ρ1 p 1 cξ P 2 1 1c R S2 R S1 R + ρ1 c + ρ1p S2 R 1 + a21 α1p.3e). thus ξ1c = ξ2c and ξ1p = ξ2p . α1p = α2p . α1p = α2p .tot P1 + c1 ξ1p PR + ρ1 p R1c + R1p R1c + R1p ≤C ≤C + (5. the power allocation variables of the sources should be α1c = α2c . One can show that the rate bounds in Proposition 1 can be written as (5. S2 and R cooperatively transmit to D2 .tot P1 S2 R 1 + a21 α1p.

4) where R1c .4. Proof 28 At each time slot.ξ1c .5 Sum-Rate Maximization The achievable sum-rate bounds of the symmetric interference relay channel can be obtained from Proposition 29. R) transmission to D1 . ∗ Rsum = α1c . it is easy to show that each rate expression inside the minimum term of (5. Using [17]. i = 1. the optimal sum-rate will be. τ ∈ (0.α1p .ξ2 max (R1 + R2 ) (5. i = 1. 2 and ξ1 + ξ2 = 1. 5. the power allocations are αi + αi = 1.5) . C R2 ≤ (1 − τ ) min C 1−τ 1−τ √ √ where ρ1 = 2 α1 ξ1 c1 P1 PR . for the orthogonal transmission scheme given in Sec. Note that in this case. we have the point-to-point relay channel. Thus the rates obtained by time sharing between the orthogonal relay channels is achievable.90 Proposition 30 In a time division scheme where the sources S1 and S2 transmit orthogonally and R helps each source in its respective orthogonal channel. the following describe the set of achievable rates for the interference relay channel. 5. 1) is the time duration R1 ≤ τ min C allocated for the (S1 .ξ1 . On the other hand.4) is achievable.α2 .α1 . R1p are given in Proposition 29 with α1c + α1c + α1p + α1p = 1 and ξ1c + ξ1p = 1/2. ρ2 = 2 α2 ξ2 c2 P2 PR . ∗ Rsum = τ. 2 is the power allocation at the source Si for the transmission of new information and ξi is the power allocation at the relay for Si . The maximum sum-rate of the symmetric interference relay channel is given by.α1c . αi . b1 α1 P1 P1 + c1 ξ1 PR + ρ1 . C τ τ P2 + c2 ξ2 PR + ρ2 b2 α2 P2 .ξ1p max 2(R1c + R1p ) (5.4) and (5.α1p .

For d > 0.6.2 and 5. 5. R} and m ∈ {R. rate splitting is equivalent to time division. A √ γ path-loss model is considered such that a . the optimal achievable sum-rate is limited by the source-relay multiple-access channel which has the same maximum sum-rate as time division [17]. In Figs 5.91 where R1 and R2 are the rates given in Proposition 30. D1 .6. Thus. and γ = 2 is the path-loss √ exponent. Contrary to the case without relay. Hence for d > 0. Note that without the relay. S2 . c1 = c2 and a12 = a21 .3 shows the optimal power allocations for the achievable sum-rate as a function of the relay distance.7. d.6. the advantage of the relay vanishes which is due to the decode-and-forward nature of the relay. the transmission from the sources to the relay becomes the bottleneck.m where d . When the relay is added to the system. rate splitting still outperforms time division up to d = 0. the interfering channel gain a21 = 1/8 would correspond to the weak interference region and rate-splitting with only private messages outperforms orthogonal transmission [5]. Fig 5.3 we study the eﬀect of relay locations on total maximum sum rate and power allocation. Obtaining optimal power allocations seems formidable. For d > 0.m = d− . The relay is equidistant from the sources and destinations. The node powers are equal.6 Discussion of the Results For the performance comparison. The ﬁgure also shows the best known achievable rate without the relay.4) requires evaluation of the multiple sum-rate bounds given in Appendix. we consider a symmetric network model such that P1 = P2 . We assume that dS1 D1 = dS2 D2 = 1 and dS1 D2 = dS2 D1 = 8. the next section provides numerical results. and d denotes the distance between the sources and the relay. thus P1 = P2 = 15. the sources . b1 = b2 . D2 }.m is the distance between nodes ∈ {S1 . however for a fair comparison the relay power is split among the sources. P1 = P2 = PR = 10. Note that obtaining the optimal power allocations in (5.

the sum-rates are in fact monotonically non-increasing functions of ξ1p . decoding of these messages at the relay improves the achievable rates of the common messages from sources to the relay since the interfering eﬀect of the private messages is removed. This can be attributed to the relay’s equal contribution to the signals received at the destinations. that is the relay never forwards common information. Moreover. In Fig. the sum rate bounds obtained from (5. We note that all numerical results show ξ1p = 0. it can be seen that rate splitting outperforms time division when a21 < 0. hence DF is not limiting. as in the interference channel. c1 = 1. even for small a21 . For a21 < 1.6. The nodes have powers P1 = P2 = PR = 10. which leads to optimal ξ1p value such that ξ1p = 0. as opposed to the weak interference channel. It should be noted that even though the relay does not transmit private information. This leads the optimal ξ1p to be ξ1p = 0 when these bounds are active. 5.4 and a21 > 1.2f) do not dominate the sum-rate. Thus. the sources do not transmit private information.5 shows that the sources never transmit private messages only. Moreover the relay never helps the sources in transmission of private messages and allocates all of its power for the transmission of sources common information such that ξ1p = 0 and ξ1c = 1/2. 5.2. i. Considering the achievable sum-rate obtained from Proposition 29. for a21 > 1. However.5 depict the maximum sum-rate of the symmetric system for ﬁxed source-relay and relay-destination link gains whereas the interference to direct link gain a21 is changed by varying a21 . Our numerical results illustrate that for a21 > 1 optimal allocation requires α1p = 0. so that ξ1p = 0.4 and 5. We conjecture this result holds for any symmetric Gaussian interference relay channel. respectively.e. the sum-rate bounds obtained from these expressions also preserve monotonically non-increasing behavior with ξ1p .4. . Fig. α1c > 0 as well as α1p > 0.92 allocate power to common information in addition to the private information. Figures 5. We have b1 = 100.

3 0.3 Rate (bits/ch use) 2. the results show that the relay allocates all of its power to transmit common messages. P1 = P2 = PR = 10.6 0.2 2. .2 0. the sources continue to send common messages along with private even when the interference is weak.7 0. 5.8 0. We argue that the presence of the relay aﬀects the transmission and power allocation strategies of the sources and the notions of interference levels. When the relay is not present.8 1.6 Relay location. Using rate-splitting idea of Carleial [5] and a decode-and-forward relay.4 0.4 2.5 0.9 Rate Splitting Time Division No Relay Figure 5.9 1. (weak/medium/strong) compared to the interference channel.93 2. We analyze optimal power allocation among the common and private messages at the sources and relay to maximize the sum-rate.1 2 1. unlike the interference channel.5 2. we have P1 = P2 = 15. we derive an achievable rate region for the Gaussian interference relay channel. d 0.2: Total optimized sum-rate (bits/channel use) of the symmetric channel with a21 = 1/8.7 1. For the symmetric interference relay channel.7 Chapter Summary We study the interference relay channel where two interfering source-destination pairs are helped by a common relay. On the other hand.1 0.

94 1 0.5 0.8 0.9 Power allocation variables 0.4 0.5 0.3: Optimal power allocation variables of the symmetric channel for rate splitting with a21 = 1/8.9 α1c α1p ξ1p ξ1c Figure 5. d 0.3 0. .6 0.2 0.6 Relay location.7 0.1 0 0.8 0.3 0.4 0.2 0.1 0. P1 = P2 = PR = 10.7 0.

.2.8 1.6 Rate Splitting Time Division No Relay Rate (bits/ch use) 0 0.5 Figure 5.5 2 2.8 2.6 2.2 2 1.4 2.4: Total optimized Rate (bits/channel use) of the symmetric channel as a function of a21 with b1 = 100.95 3 2.5 1 a21 1. When the relay is not present. we have P1 = P2 = 15. c1 = 1. P1 = P2 = PR = 10.

P1 = P2 = PR = 10.96 1 0.8 Power allocation variables 0. c1 = 1.2.2 0.1 0 0 0.5 1 aS 1.9 0. .6 0.3 0.5 /aS 2 D α10 α11 β11 β10 2.4 0.5: Optimal power allocation variables of the symmetric system for rate splitting as a function of a21 with b1 = 100.7 0.5 0.5 3 2 1 D 1 1 Figure 5.

due to broadcasting nature of wireless terminals. However. signal relaying and interference . interference is the fundamental design parameter in the architecture of next generation systems. We examined several models to understand the beneﬁts of relaying to explore various practical multi-user networks. increased number of users in a given network and utilization of the coexistence of various systems can be regarded as the primary components. Recently. Despite the simplicity oﬀered by the orthogonal relay channels. this system reveals the major relaying operations in a general interference limited multi-user network with the emphasis of improving communication rates.97 Chapter 6 Conclusion Next generation wireless systems are required to ensure improved qualityof-service (QoS) requirements due to increasing demands by the users. interference is an unavoidable and foremost phenomenon in the wireless communication networks which contradicts with the demands of these and almost all QoS requirements. we considered a two-user Gaussian interference channel aided by a relay where the relay reception and transmission take place over orthogonal bands. We examined two signiﬁcant relaying operations. namely. we investigated the impact of relaying in interference limited multi-user networks. First. Therefore. Among the QoS requirements. higher communication rates. In this thesis. cooperation/relaying has emerged as a very instrumental tool in providing substantial improvements for mostly point-topoint systems.

we established that signal relaying is optimal along with the substantial reduction in encoding/decoding complexity as a result of interference and relay channel separation. On the other hand. Our analysis showed that signal relaying and interference forwarding are essential for improved performance. Moreover. under the conditions where the relay-to-destination channels form the performance bottleneck with respect to the source-to-relay channels. Our results show that signal relaying. and thus non-separable transmission. in general. Moreover. emphasizing the importance of proposed operations. we showed that these results are generally valid under the assumption of variable bandwidth allocation among the relay channels.98 forwarding. . Next. we explored a diﬀerent model with a particular interest in revealing additional relaying functions relevant in interference limited system. which also brings additional performance improvement. hence giving relay the opportunity to reduce the interfering signal strength. For instance. In the model. where the former enables the destination to obtain a stronger desired signal or additional information. and the latter helps removing part of the interference. the orthogonality feature of the relay reception channels has been demonstrated to provide the opportunity of extra source information transmission via the relay through non-linear techniques. interference cancelation can indeed be the primary and most eﬀective relaying operation under some channel conditions. the optimality of these operations have also been related to the problem of assessing optimality of either separable or non-separable transmission over parallel interference channels. However. We denoted this relaying operation as interference cancelation. is necessary to attain optimal performance. the analysis shows that. joint signal relaying and interference forwarding. or the combination of signal relaying and interference forwarding can attain optimal transmission rates under some channel conditions. This clearly complicates the design. however. the relay is assumed to receive over orthogonal bands with ﬁnite capacities and transmit in the band of the interfering nodes.

Despite the complexity added into the system. It is shown that in-band reception at the relay necessitates enhanced decoding and encoding schemes at the relay to maintain the beneﬁts oﬀered by proposed relaying operations. we studied the generalization of Gaussian relay channel [6] to incorporate multiple source-destination nodes. . our analysis revealed the substantial performance improvement compared with standard approach which is to orthogonalize the channels to combat interference.99 Finally.

YR1 ) ηR 1 n ≤ I (W1 .10) can be obtained by noting that. YR 1 )+n n (A. Y1n . We have the following bound nR1 = H (W1 ) ηR 1 n ηR 1 n n = I (W1 .4) n ηR 1 n ηR 1 n n ≤ I (X1 .3) (A. and therefore n nR1 ≤ I (X1 .4) is n from chain rule. the ﬁrst bound on R1 in (3. . YR 1 ) ≤ n and (A.1 Proof of Proposition 5 We consider outer bounds on R1 and the bounds on R2 can be obtained similarly.1) (A. Y1n . ηR1 n Now. ηR 1 n where (A. the Markovity W1 → X1 → Y1n and conditioning decreases entropy. Y1n ) + h(YR 1 ) − h(ZR1 ) + n n .100 Appendix A A. h(YR 1 ) − ηR1 n ηR 1 n ηR 1 n ηR 1 n 2 h(ZR + ZR 1 ) = h(c1 XR1 1 ) − h(ZR1 ) ≤ nηR1 C (c1 PR1 ).2) (A.3) follows from the Fano inequality H (W1 |Y1n . Y1n ) + nηR1 C (c2 1 PR1 ) + n n . YR 1 ) + H (W1 |Y1 .

Y1n . Y2R ) − h(Z1R ) − h(Z2R ) + n η2R n η1R n η2R n n 1R n ≤ I (X1 . Finally. YR 1 )+n n n n n (A. Y1n |X2 ) + nηR1 C (c2 1 PR1 ) + n n .10) from the following series of inequalities ηR1 n nR1 ≤ I (W1 .13) (A. following (A. W1 → X1 → Y1n and conditioning decreases 2R n 1R n and Y2η entropy and (A.3).7) is from chain rule. ηR 1 n ηR1 n ηR 1 n ηR 1 n 2 Now. Moreover. Y2R |W2 ) − h(Z1R ) − h(Z2R ) + n η1R n n n 1R n ) + h(Y1η . Y2R ) + n η2R n η1R n η2R n n 1R n ≤ I (X1 . the last bound R on R1 in (3. Y1n |X2 ) + h(YR 1 |W2 ) − h(ZR1 ) + n n . we have n n nR1 ≤ I (X1 . Y2R 1 n ing inequality.9) η2R n 1R n ≤ I (W1 .5) (A.101 Also.7) (A. from (A. Y1n ) + h(Y1η R ) + h(Y2R ) − h(Z1R ) − h(Z2R ) + n n 2 ≤ I (X1 .13). (A. since h(YR 1 |W2 ) − h(ZR1 ) ≤ h(YR1 ) − h(ZR1 ) ≤ nηR1 C (c1 PR1 ). Y1n |X2 nR1 ≤ I (X1 R . Y1n |X2 ≤ I (X1 R ) − h(Z1R ) + n n n ) + nη1R C (b2 . Y1n |W2 ) + I (W1 . η2R n ηR 1 n 1R n where (A. Y1n .18) η2R n η1R n η2R n n n 1R n ) + h(Y1η .6) is from the Markov chain W1 → Y1η → YR and data processR .8) (A. Y1η R . we have. Y1n ) + nη1R C (b2 1 P1R ) + nη2R C (b2 P2R ) + n n . Y1n ) + h(Y1η R . YR 1 |Y 1 . YR 1 |W 2 ) + n n n (A. η2R n n n n 1R n ) + I (W1 . W 2 ) + n ηR 1 n ηR 1 n n ≤ I (X1 .12) (A.14) ηR1 n n = I (W1 . we get the bound the other bound in (3. Y2R |Y1 .8) is from independence of Y1η R . W2 ) ηR 1 n ≤ I (W1 .15) (A.10) is obtained as.11) (A.10) (A.17) (A. YR1 . Y1n .16) (A. Y1n |X2 ≤ I (X1 1 P1R ) + n n n . Y1η . Y1n .6) (A. YR 1 |W2 ) + H (W1 |Y1 . nR1 = H (W1 |W2 ) ηR 1 n ηR 1 n n = I (W1 . Y1n |X2 ≤ I (X1 R . W2 ) + n n n (A.

i. to facilitate interference cancellation. from independent Gaussian distributions with zero-mean and . (A. [8]. e. On the other hand. respectively. (see. Moreover. The general outer bound is then obtained by taking the union of all rates n (R1 . which can be proved to coincide with the limiting region in (3.2.19b) Message WiR is transmitted to Di via the OBRC only. W2c ) as explained in n n Sec. Y2R 1 . W2c ). W2p . source Si transmits message Wic to the interfered destination Dj .. we have the transmitted codewords over the IC: n n n X1 (W1 ) = X1 p (W1p ) + X1c (W1c .19a) (A.17) is from the fact that X2 is a function of W2 and R η2R n 1R n independence of Y1η and Z2 R R . x2 ) = n p(xn 1 )p(x2 ).10). Ric + Ric for i = 1. W2c ). Such codewords are generated i. R2 ) satisfying the constraints for all n ≥ 0 and input distributions p(xn 1 . 2 A.102 η2R n ηR 1 n 1R n where (A. W2R . via the OBRC. with power constraints (P1 . WiR ) which are generated i.3.d. (A. i = 1.i. 2. after successfully decoding the messages (W1R . and sent over the IC in n channel uses.15) is from W1 → Y1η → YR R . P2 ). Messages Wip and (Wic . WiR ) are jointly encoded by Si into the ηiR n codewords XiR (Wic .d. the OBR enηRi n codes these messages into the codewords XRi (Wjc . with rate Ric + RiR from inde- pendent Gaussian distributions with zero-mean and power PiR .2 Proof of Proposition 6 Codeword Generation and Encoding: The sources divide their messages as W1 = (W1R . j = i. Pic .d. W1p . 2. W1c .g.3. Remark 1). The messages (Wic .i. WiR ) with rate Rjc + RiR which are also generated i. W1c ) n n n X2 (W2 ) = X2 p (W2p ) + X2c (W2c .Wic ) are encoded into codewords Xip and Xic with rates Rip .16) is due to conditioning η2R n decreases entropy and (A. from independent Gaussian distributions with zero-mean and powers Pip . respectively. W1c . W2c . Overall. W1c ). and W2 = (W2R .

Similarly. Moreover. W2c ) with only 2nR2c codewords (and power Pic ). 2 using the orthogonal source-to-relay links as given in (3.3 Proof of Proposition 8 The converse follows from Proposition 5. WiR ).2).11c) and (3.103 power PRi .14b) are a consequence of the second bounds on both R1 and R2 . Decoding: The destination Di initially decodes the messages (Wjc . W1c ) at destination D2 ) is then performed jointly as over a multiple access channel with three sources of rates R1c = R1c + R1c . The relay decodes the messages (Wic . Namely.11d) which are the point-to-point rates in decoding the messages (WiR . j = i.11f).11a) and (3. W2c ) at destination D1 (and (W2c . since the n destination D1 decodes W2c . Therefore. by treating the private messages as noise. 2. The signals received on the IC are given by (3. while the upper bound on the sum rate (3. 2. WRi ) using the R − Di channel which leads to the achievable rates (3. W1c ) with rate R1c . i = 1. thus with equivalent noise power Ni = a2 ji Pjp + 1 for i.11b).1) with (A. Decoding of the messages (W1c . j = 1. W1p . it is possible to show that the rates in (3. i. W2c . R1p and R2c (and R2c = R2c + R2c . R2p and R1c for D2 ). 2 A. D2 sees an equivalent coden book X1 c (W1c . W2p . j = 1.14c) follows by summing second .11e) and (3. Wic ) are achievable. it thus sees an equivalent codebook X2 c (W2c . the upper bounds on individual rates (3.19). W1c . error events corresponding to erroneous decoding of only message W2c at destination D1 and W1c at destination D2 do not contribute to the probability of error and thus can be neglected.14a) and (3. It is also noted that. i = j hence giving the achievable rates (3. as explained in [51].

W2c ) over the IC which are decoded at both destinations. but also message W2c of rate R2c to D1 in order to facilitate interference cancellation. h(X2 + Z2 ) − n n h(a21 X2 + Z1 ) ≤ n log(1) for a21 ≥ 1. and applying Fourier-Motzkin elimination.2 . The OBRC is used to transmit independent messages W1R .25) (A.22) 2 2 + min η1R C (b2 1 P1R ) + η2R C (b2 P2R ). and the ﬁrst entropy is maximized by i.e.i. where the sources transmit common messages (W1c . From Proposition 2. the claim is proved. C (a2 12 P1 + P2 ) − C (P1 + a21 P2 )}. ηR1 C (c1 PR1 ) n n (A. For achievability.20) 1 1 1 n n n n n n n n ) + h(a12 X1 ) − h(a12 X1 ) − h(Z1 + Z2 ) + Z1 + X2 + Z2 h(X1 n n n (A.22) is from the worst-case noise result [36]. R2 . respectively and accounting for the condition a12 ≥ 1 as R1 + R2 ≤ 1 1 n n n 2 I (X1 . In addition.21) (A. (A.23) (A. The other rates are set to R1c = R1p = R2p = 0. W2R with rates R1R = η1R C (b2 1 P1R ) and R2R = ηR2 C (c2 2 PR2 ). i. we obtain the following achievable region R1 ≤ C (P1 ) + η1R C (b2 1 P1R ) R2 ≤ C (P2 ) + ηR2 C (c2 2 PR 2 ) 2 2 R1 + R2 ≤ C (a2 12 P1 + P2 ) + η1R C (b1 P1R ) + ηR2 C (c2 PR2 ) 2 2 R1 + R2 ≤ C (P1 + a2 21 P2 ) + Rex21 + η1R C (b1 P1R ) + ηR2 C (c2 PR2 ).26) 2 so that for Rex21 ≥ max{0.. S2 transmits also the message W2c to be decoded at D2 .21) is due to the conditions η1R C (b2 1 P1R ) ≤ ηR1 C (c1 PR1 ) and ηR2 C (c2 PR2 ) ≤ n n η2R C (b2 2 P2R ). we use the general result of Proposition 6. (A.24) (A. Gaussian inputs.d. Y1n |X2 ) + I (X2 . Y2n ) + min η1R C (b2 1 P1R ).104 and ﬁrst bounds on R1 . ηR2 C (c2 PR2 ) = 2 + η1R C (b2 1 P1R ) + ηR2 C (c2 PR2 ) 2 2 ≤ C (a2 12 P1 + P2 ) + η1R C (b1 P1R ) + ηR2 C (c2 PR2 ) 2 2 where (A.

15a) and (3.29) is from the worst-case noise 1 n n n n log(1) for a21 ≥ 1.15b) follow immediately from the second bounds on R1 and R2 . and the fact result [36]. h(X2 + Z2 ) − h(a21 X2 + Z1 ) ≤ n2 that ﬁrst entropy is maximized by i. ηR2 C (c2 PR2 )) = 2 + ηR1 C (c2 1 PR1 ) + ηR2 C (c2 PR2 ) 2 2 ≤ C (P1 + a2 21 P2 ) + ηR2 C (c2 PR2 ) + ηR1 C (c1 PR1 ). R2 .15c) is obtained from the summation of ﬁrst and second bound on R1 .27) 1 1 1 n n n n n n n n ) − h(a21 X2 ) h(X1 + a21 X2 + Z1 + Z1 ) + h(X2 + Z2 ) − h(Z2 n n n (A.105 A. while the bound on the sum-rate (3. i. 2 .28) is due to the conditions ηR1 C (c2 1 PR1 ) + ηR2 C (c2 PR2 ) ≤ η1R C (b1 P1R ) + 2 2 η2R C (b2 2 P2R ) and ηR2 C (c2 PR2 ) ≤ η2R C (b2 P2R ). hence gives the sum capacity.. consider the achievable rate region given in the proof of Proposition A. and the condition a12 ≥ 1 as R1 + R2 ≤ 1 1 n n n 2 2 I (X1 .14c). (A.29) is achievable. Y2n |X1 ) + min(η1R C (b2 1 P1R ) + η2R C (b2 P2R ). when the conditions in 2 2 Proposition 9 which can also be written as ηR1 C (c2 1 PR1 )−η1R C (b1 P1R ) ≤ η2R C (b2 P2R )− 2 2 ηR2 C (c2 2 PR2 ) and Rex21 ≤ C (a12 P1 + P2 ) −C (P1 a21 P2 ) are satisﬁed.14a)-(3. respectively.4 Proof of Proposition 9 The converse is again a consequence of Proposition 5. (A. Speciﬁcally. the single rate bounds (3. Clearly. For the achievability. ηR1 C (c1 PR1 )) n n (A.d.29) 2 + min(η2R C (b2 2 P2R ).28) (A.3 ((A.i. Y1n ) + I (X2 . 2 2 where (A.26)) (3.e. Gaussian inputs.23)-(A.

106 A. The other rates are set to zero R1c = R1c = R2p = 0. W2c ) as well as private message via OBR. ηR2 C (c2 PR2 )) = 2 + η1R C (b2 1 P1R ) + ηR2 C (c2 PR2 ) ≤ C (P1 ) + C 2 + η1R C (b2 1 P1R ) + ηR2 C (c2 PR2 ) 2 2 where (A. Achievability follows directly from Proposition 6 by letting transmitter S1 transmit private message only. 2 2 R1 + R2 ≤ C (P1 + a2 21 P2 ) + Rex21 + η1R C (b1 P1R ) + ηR2 C (c2 PR2 ) (A.e. for Rex21 ≥ C (P1 ) + C (A. W2R . hence gives the sum capacity.2 P2 1+a2 12 P1 − C (P1 + a2 21 P2 ).e.32) is from the worst-case noise result of [36].34) is achievable.32) P2 1 + a2 12 P1 2 2 + min(η1R C (b2 1 P1R ) + η2R C (b2 P2R ). Then. Then. i.31) (A.33) (A.10) such that. i.13). using Fourier-Motzkin elimination.31) is due to the conditions ηR1 C (c2 1 PR1 ) ≥ η1R C (b1 P1R ) and η2R C (b2 P2R ) ≥ n n ηR2 C (c2 2 PR2 ). whereas user S2 transmits common information both on the IC and OBR (W2c .. (A. . R1 + R2 ≤ 1 1 n n n 2 I (X1 . W1p over the IC and W1R over the OBR.34) R1 + R2 ≤ C (P1 ) + C P2 1 + a2 12 P1 2 + η1R C (b2 1 P1R ) + ηR2 C (c2 PR2 ) where Rex21 is given in (3.30) 1 1 1 1 n n n n n n n n h(X1 + Z1 ) − h(Z1 ) + h(a12 X1 + X2 + Z2 ) − h(a12 X1 + Z2 ) n n n n (A.5 Proof of Proposition 10 The converse is obtained from Proposition 5 by adding the second constraint on R1 and ﬁrst constraint on R2 in (3. ηR1 C (c1 PR1 )) n n (A. it is possible to show that the following sum-rate is achievable. Y1n |X2 ) + I (X2 . h(X1 + Z1 ) − n n h(a12 X1 + Z2 ) ≤ nC (P1 ) − nC (a2 12 P1 ) for a12 < 1. Y2n ) + min(η1R C (b2 1 P1R ).

10). the achievable sum-rate is given by (A. ηRi ) with constraint 2 i=1 (ηiR + ηRi ) 2 2 the bandwidth allocation.5. 2 For the outer bound. W2R ) over the OBRC (signal relaying) and only common information (W1c .7 Proof of Proposition 12 For the achievable scheme.33).11f).e.34) in Appendix A. Hence. ηR2 C (c2 PR2 )) (A.29) in Appendix A. by sending only message (W1R .107 A. which is true for a21 ≥ 1.33) and (A. we consider a special case of Proposition 6.35) = η . for 0 ≤ Rex21 ≤ C (P1 ) + C η1R C (b2 1 P1R ). consider (A. which is equal to the upper bound (A. ηR2 C (c2 PR2 )} (A. P1 = P2 . using Proposition 5 with a12 = a21 ≥ 1. from (3. ηR1 C (c1 PR1 )) 2 + min η2R C (b2 2 P2R ). ηR1 C (c1 PR1 )} 2 + min{η2R C (b2 2 P2R ). For the achievability.36) . using (3. i. we obtain that the following sum-rate is achievable 2 2 R1 + R2 ≤ C P1 + a2 21 P2 + min η1R C (b1 P1R ).29) for Rex21 = ηR1 C (c2 1 PR1 ) − A. W2c ) over the IC. the optimal allocations are η2R = ηR2 = η/2.6 Proof of Proposition 11 The converse is directly from (A. Optimizing over to be maximized over (ηiR . 2 P2 1+a2 12 P1 − C (a2 21 P2 + P1 ). where sources operate separately over the IC and OBRC. From Proposition 2. so that the optimal achievable sum-rate is R1 + R2 ≤ C P1 + a2 21 P2 + η C (c2 2 PR2 ) . and recalling that C (b2 2 P2R ) ≥ C (b1 P1R ) and C (c2 PR2 ) ≥ ∗ ∗ C (c2 1 PR1 ). we obtain the upper bounds 2 2 2 R1 + R2 ≤ C P1 + a2 21 P2 + min{η1R C (b1 P1R ) + η2R C (b2 P2R ).11a)-(3.6.

ηR1 = 0.45) ηBC n = I (W1 .2 A.108 which should be maximized over (ηiR . 2 C (c2 2 PR2 ) + 2C (c1 PR1 ) η C (c2 2 PR 2 ) + . whereas the second to η2 R = ηR2 = 2 . ηRi ) with 2 i=1 (ηiR + ηRi ) = η . Y1n . 2 C (b2 P 1 (A.41) (A. Y1n . YR |Y1n ) + n ηM AC n ηM AC n n ≤ I (X1 . YR 1 1 ηBC n ≤ I (W1 .42) (A. ηR 2 = 2P R1 ) η ∗ ∗ ∗ η2R .40) is from Fano’s inequality. using C (b2 2 P2R ) ≥ C (b1 P1R ). and W1 → X1 YR .16a): nR1 ≤ H (W1 ) ηBC n ηBC n ) ) + H (W1 |Y1n . Y1n ) + I (W1 . YR |Y1n ) + n 1 ηM AC n n ≤ I (X1 . η1R = 0.8 Proof of Propositions 13 and 14 We start with the bound (3.40) (A.43) (A. YR )+n 1 n n n n n (A.38) (A. Y1n ) + h(YR ) − h(ZR )+n ηM AC n ηM AC n ηM AC n ηM AC n n ≤ I (X1 . Y1n → YR 1 .39) (A. (A. Y1n ) + h(b1 X1 + b2 X2 + ZR ) − h(ZR )+n R R 2 n . 2. Optimizing over 2 (ηiR .44) (A. we obtain R1 + R2 ≤ max C P1 + a2 21 P2 + C P1 + a2 21 P2 2 2η C (c2 1 PR1 )C (c2 PR2 ) . Y1n ) + I (W1 . ηR1 + η2R + ηR2 = η.42) is from the Markov relations W1 → ηBC n ηM AC n n → Y1n .37) becomes equal to the optimal achievable sumrate. It is possible to show that for 2 C (c2 2 PR2 ) ≥ 2C (c1 PR1 ) the outer bound obtained in (A.37) ) 2R where the ﬁrst term in the max corresponds to the choice ηR1 = η2R C (c2 . YR ≤ I (W1 . ηRi ). Y1n ) + ηM AC nC b2 ≤ I (X1 1 P1R + b2 P2R + n n where (A. i = 1.

109 From cut-set bound around S1 .48) (A.50) (A. Y1n .41) modiﬁed for R2 ).52) for some 0 ≤ ξ ≤ 1.54) 1 + c2 2 ξPR where we have used the fact that Gaussian distribution maximizes the entropy term for a given variance constraint. ηBC n |W 2 ) + n nR1 ≤ I (W1 . without loss of generality. YR .51) becomes n nR2 ≤ I (X2 . YR |Y2n ) 2 ηBC n ηBC n ηBC n n ≤ I (X2 . Y1n |X2 = I (X1 |Y 1 .47) Using similar steps we obtain the corresponding bounds on R2 (3. Then. Y1n |X2 ≤ I (X1 1 c2 ηBC n n n n ) + I (W1 . Now.16e) as n 2 nR2 ≤ I (X2 . (A.16d) and (3. Y2n ) + ηBC n ηBC n log(2πe(1 + c2 log(2πe(1 + c2 2 PR )) − 2 ξPR )) (A.16f).51) where (A. consider (A. Y1n |X2 ) + I (X1 . YR |Y1n . 2 (A.57) . Y2n ) + h(YR ) − h(c2 XR + ZR |Y2n . W2 ) ≤ h(c2 XR +ZR ) ≤ h(c2 XR h(ZR 2 2 2 ηBC n log(2πe(1+c2 2 PR )). we obtain the bound (3.56) n ηBC n n n ) + I (W1 . W2 ) = 2 ηBC n log(2πe(1 + c2 2 ξPR )). one can assume ηBC n h(YR |Y2n . W2 ) + n n .40) for the bound on R1 given in (3. Now. 2 Hence. ηBC n ηBC n ηBC n ηBC n ηBC n )≤ +ZR |Y2n . Y2n ) + I (W2 . Y1n |X2 ) + ηM AC nC b2 ≤ I (X1 1 P1R .51) is from conditioning decreases entropy. Y2n ) + ηM AC nC b2 1 P1R + b2 P2R n n nR2 ≤ I (X2 . (A. YR |X2 ) R R n n . Y2n |X1 ) + ηM AC nC b2 2 P2R . W2 ) 2 2 (A. W 2 ) + n c1 1 (A.16f). ηBC n n nR2 ≤ I (X2 .16b) as ηM AC n ηM AC n ηM AC n n n nR1 ≤ I (X1 . Y2n ) + ηBC nC (A. (A.53) 2 2 c2 2 ξPR n = I (X2 .49) We now focus on the remaining two bound (3. we have (from (A.16c) and (3.16c).55) (A. For (3. YR 1 n (A. consider the following.46) (A.

we have. W2 ) |Y1n . W2 ) = h(YR 2 ηBC n = log(2πe(1 + c2 2 ξPR )) 2 (A.62) (A.59) (A.142) is due to conditioning decreases entropy and independence of W1 and n n W2 .62) is true since X2 . recalling (A.61) (A.W h Z BC |Y1 2) + 2 ηBC n ( R . (A. Y1n |X2 ) + h( c2 ηBC n n c2 ηBC n Y R 1 |Y 1 . W2 ) = h(YR h(YR 2 2 ηBC n n n |X1 + Z2 .59) is from the fact that a21 X2 is a function of W2 . a12 X1 + 1 ηBC n n n n n Z2 → X1 + Z2 for the fact that a12 ≤ 1. (A. . W2 ) = h(YR 2 ηBC n n n |a12 X1 + Z2 .65) (A.57) and considering the inequality (A. c2 1 is an iid. (A. Since the capacity region of BC depends on the conditional marginal distriηBC n butions and noting that c1 ≥ c2 . W 2 ) − h ( Z R )+n c1 c1 1 n (A. using (A.63).64). we now have 2 ηBC n 2 BC |Y n .58). (A.63) from (A. Y1n |X2 ) + ηBC nC c2 1 ξPR + n n . Gaussian noise with variance 1 − Inequality. and Z2 . (A. we can write YR = 2 ηBC n c2 ηBC n Y + ZR c1 R 1 ηBC n where ZR c2 2 .W ) h(YR2 2 1 η n From the conditional Entropy Power ≥ 2 ηBC n 2 h c2 ηBC n n .58) Also.57) is since scaling does not change the mutual information.W Y |Y1 2 c1 R1 η n 2 n .63) n where (A. W 2 c1 R 1 ηBC n c2 2 log 2πe c2 ξP + R 2 2 c2 1 So that.60) follows from ηBC n n n n . . n n nR1 ≤ I (X1 . and noticing ηBC n that h ZR |Y1n . W 2 → YR is 2 a function of W2 .52). Z1 the independence of ZR . and (A. ηBC n ηBC n n n |X1 + Z1 . W2 = ηBC n 2 log 2πe(1 − ≤ c2 2 ) c2 1 .64) h c2 ηBC n n Y |Y 1 .60) (A. we obtain. Then.66) n n ≤ I (X1 . (A. for the condition a12 ≤ 1.56) is from Markovity W1 → X1 → Y1n and X2 is a function of W2 . we get.61) is due to the Markov chain. (A.110 where (A. W2 ) ≤ h(YR 2 ηBC n |Y2n . (A.

9 Proof of Propositions 15 and 18 We derive ﬁrst the outer bound of Proposition 15. (A.70) (A. c2 → ∞. (A. Y1n |X2 ) + nηBC C (c2 1 PR ). For Proposition 9. W2 ) ≤ I (X1 . so that we obtain an equivalent model as shown in Fig.111 Z1 X1 X1R b1R b1R 1 + + a12 ZR Y1 D1 YR b2R a21 ZR X2R X2 b2R + YR D2 Y2 1 + Z2 Figure A.69) is from the Markov chain W1 → X1 → n Y1n and X2 is a function of W2 . which recovers (3. A.70) is since conditioning decreases entropy and (A.2 A. we obtain the bound (3. Y1n |X2 ) + h(YR 1 1 n n ≤ I (X1 . Assume c1 . Y1n |X2 ) + I (W1 .67) (A.16c) and completes the proof. YR 1 ηBC n n n |Y1n .71) is from the fact that Gaussian distribution maximizes the entropy for given variance constraints. we obtain the rate for R2 .68) (A.68) is from Fano’s inequality.71) n where (A. The equivalent model . YR 1 ηBC n ηBC n n n ) ) − h(ZR ≤ I (X1 .1.1: Equivalent model for IC-OBR Type II channel for c1 . c2 → ∞. Y1n .69) (A. (A. Using similar steps.17a) as nR1 ≤ H (W1 |W2 ) ηBC n |W2 ) + npen ≤ I (W1 .

ηBC C (c1 ξPR ) + ηBC C c2 2 ξPR 1 + c2 2 ξPR (A.112 is in fact equivalent to a 2 × 2 MIMO interference channel matrices.10 Proof of Proposition 16 The converse follows from (3.16f) as R1 + R2 ≤ 1 1 n n n 2 I (X1 . and H4 = . For this channel. H2 = 21 . For 2 ηM AC C (b2 1 P1R ) ≥ ηBC C (c1 PR ).73) = C (P1 ) + C + ηBC C P2 1 + a2 12 P1 c2 2 ξPR 1 + c2 2 ξPR 2 + min ηM AC C (b2 1 P1R ). and hence . the assumed conditions a12 ≥ 1 and a21 ≥ 1 imply the strong interference regime H† 2 H2 † H† 4 H4 and H3 H3 H† 1 H1 . Notice that such equivalence is due to the fact that 0 b1 0 b2 noise correlations are immaterial in terms of the capacity region. The achievable rate for Proposition 14 follows from standard arguments assuming Wyner-Ziv compression at the relay with Gaussian test channels (see.74) where (A.18a)-(3.73) is from the worst-case noise result of [36] applied for a12 ≤ 1.. [42]).18d). ηBC C (c1 ξPR ) (A.16c).g.16b). whose channel 1 0 a 0 .72) 1 + c2 2 ξPR 2 ≤ C (a2 12 P1 + P2 ) + C (P1 ) − C (a12 P1 ) 2 + min ηM AC C (b2 1 P1R ). ηBC C (c1 ξPR )} n n c2 2 ξPR + ηBC C (A.74) is maximized for ξ = 1 since c1 ≥ c2 . so that the capacity region can be found from [74] as given by (3. Y2n ) + min{ηM AC C (b2 1 P1R ). (A. e. H3 = following the notation in [74] are given by H1 = 0 b1 0 b2 a 0 1 0 12 . and (3. (3. Y1n |X2 ) + I (X2 .2 A.

[42]).77) is equal to the outer bound (A.75) 2 ∗ Instead. for the 2 2 conditions ηM AC C (b2 1 P1R ) < ηBC C (c1 PR ) and ηM AC C (b2 P2R ) ≥ ηBC C c2 2 ξ PR ∗ 1+c2 2 ξ PR ∗ ∗ .76). over the IC.75) is achiev- able.113 the outer bound becomes. with the power split of ξ ∗ PR allocated at the relay for the transmission of W1R (and ξ PR for W2R ). we use a special case of the coding scheme (A. (A. (A. On the other hand. denote ξ as the optimal parameter that maximizes (A. for ηM AC C (b2 1 P1R ) < ηBC C (c1 PR ).74) so that we get R1 + R2 ≤ C (P1 ) + C + ηBC C where ξ ∗ + ξ = 1. R2 = R2c + R2R . S1 transmits private information only via a Gaussian coden n book X1 (W1 ) = X1 p (W1p ). W2R ) are sent over ∗ P2 1 + a2 12 P1 ∗ 2 ∗ + min ηM AC C (b2 1 P1R ). R1 + R2 ≤ C (P1 ) + C P2 1 + a2 12 P1 2 + min ηM AC C (b2 1 P1R ). ηBC C c2 2 ξPR 1 + c2 2 ξPR . we obtain an achievable sum-rate. setting R1 = R1p + R1R . The proof then follows similarly to Appendix E.76) the OBRC by using standard Gaussian codebooks and MAC decoding at the relay. Speciﬁcally. and superposition coding at the relay.19) in which. R1 + R2 ≤ C (P1 ) + C P2 1 + a2 12 P1 + ηBC C (c2 1 PR ). and S2 transmits common information using a Gaussian n n codebook X2 (W2 ) = X2 c (W2c ). ηBC C (c1 ξ PR ) c2 2 ξ PR ∗ 1 + c2 2 ξ PR (A. For the achievability. the achievable sum-rate (A. e.2 . for ξ = 1 and ηM AC C (b2 1 P1R ) ≥ ηBC C (c1 PR ). the outer bound (A.g. Only private messages (W1R . by accounting for the capacity regions of MAC and BC Gaussian channels (see.77) 2 Now. ηBC C (c1 ξPR ) + min ηM AC C (b2 2 P2R ). hence we obtain the sum capacity..

Then. i.2 A.10.18a).e. Source S2 transn n (W2 ) = X2 mits common messages (W2c . the following conditions are easily seen to provide an achievable . the achievable sum-rate is equal to the outer bound (3. X1 (W1 ) = X1 p (W1p ) over the IC and indepen- dent private message over the OBRC W1R via Gaussian codebooks. i = 1.11 Proof of Proposition 17 For a symmetric IC-OBR Type-II. C (P + a2 P ) +min 2ηM AC C (b2 Ps ).12 Proof of Proposition 19 The achievable region is obtained similarly to Appendix A. 2) and independent messages (W1R . R1R = R2R and R1 + R2 = 2(R1c + R1R ).79) (A.. W2c )).80) (A.10. for a ≤ √ 1 + P and c2 PR ≥ (1 + 2b2 Ps ) ηM AC ηBC (A. ηM AC C (2b2 Ps ). the following rate region can be easily shown to be achievable R1c ≤ C (P ) R1c + R2c ≤ C (P + a2 P ) R1R ≤ ηM AC C (b2 Ps ) R1R + R2R ≤ ηM AC C (2b2 Ps ) R1R + R2R ≤ ηBC C (c2 PR ) We now set R1c = R2c . Then. so that R1 +R2 ≤ min 2C (P ).81) (A. we have the following achievable region n when the sources transmit common messages over the IC (Xin (Wi ) = Xic (Wic ). as discussed in Appendix A.78) (A. and the private message W2R is transmitted via the OBR along with W2c (interference forwarding). Source S1 n n transmits private message only. W2R ) via OBRC.114 A. Then. W2c ) over the IC (X2 c (W2c .82) − 1. hence giving the sum capacity. ηBC C (c2 PR ) .

for b2 .2 A. and the achievable sum-rate R1 + R2 obtained by the summation of (A. the achievable region becomes R1 ≤ C (P1 ) + ηM AC C (b2 1 P1R ) R2 ≤ C P2 1 + a2 12 P1 + ηBC C c2 2 PR (A.88) (A. with the fact that R1 = R1p + R1R .13 Proof of Proposition 20 n n )= . thus concluding the proof.87) (A.115 region R1p ≤ C (P1 ) R2c + R1p ≤ C (P1 + a2 21 P2 ) R2c ≤ C P2 1 + a2 12 P1 (A. which is obtained from Proposition 9 for a12 < 1.85) (A. The achievability follows from . c1 → ∞.91) and (A.92) is equal to the outer bound (A.84) (A. as b2 . the achievable region in Proposition 15 can be obtained. R2c = R2c + R2c . Now. For the outer bound. c1 → ∞.86) (A.90) R1R ≤ ηM AC C (b2 1 P1R ) R2c + R2R ≤ ηM AC C (b2 2 P2R ) 2 R1R + R2c + R2R ≤ ηM AC C (b2 1 P1R + b2 P2R ) R2c + R1R ≤ ηBC C (c2 1 ξPR ) R2R ≤ ηBC C c2 2 ξPR 1 + c2 2 ξPR Using Fourier-Motzkin elimination method. and R2 = R2c + R2R . Y1n |X2 The converse follows directly from Proposition 2 by noticing that I (X1 n n ) = C (P ) under the given assumptions.89) (A. Y2n |X1 I (X2 .74) Proposition A.91) (A. c1 → ∞. we use the bound on IC-OBR Type-II given in (A.74).92) since the overall region is maximized for ξ = 0 for b2 . Again.83) (A. the outer bound is maximized for ξ = 0.10.

Then the input signals for the IC can be written as (omitting dependency on the message splits). Gaussian distribu(Wic ).i.95) (A.d. i = 1. Xip (Wip ). Wip . each source Si transmits messages (Wic . W1p ) V2 = (W2c . i = 1.94) To maintain power constraint at each source. Moreover. αip Pi . W2p . ηM AC 2 log 1 2 + b2 Ps }. encoded into codewords Xs with rates Rs . the allocations should satisfy. Messages Ws . Wic . ip}. i = 1.14 Proof of Proposition 21 Codebook generation and encoding: Each source performs message splitting as discussed in Sec. Wi = (Wic . αic Pi and αip Pi .e. 2. i = 1.96) With the assumption that the relay decodes these splits with arbitrarily . that is. Furthermore. i = 1. Wip . W1p . Xip 2 2 2 2 tions with zero mean and variances αic Pi .4.2. to the relay via the ﬁnite capacity links in n uses of these links. Wip . 2. W2p ) (A. ic. all codewords are independent. s ∈ {ic. respectively. using V1 = (W1c . 2. Wip ). 2 2 2 2 αic + αic + αip + αip ≤ 1. i=1. 2 A. 2. Wip ).e. i. 4. the IC) and the second term follows from achievable rates on the two-way relay channel in [48] [47]. are generated using i. where the ﬁrst term (C (P )) is the capacity over the two-way channel (i. one can achieve a symmetric rate R ≤ C (P )+min{ηBC C (c2 PR ) . The IC codewords Xic n n n Xic (Wip ). respectively. are then n n (Wic ).93) (A. n n n n n X1 = X1 c + X1c + X1p + X1p n n n n n X2 = X2 c + X2c + X2p + X2p (A.116 the discussion above since on the equivalent channel described before Proposition 20. ip. 2..

W1cr .95). the interference due to X1p is treated as noise. in encoding of n nI (U1p . In forming the codebook. can be obtained however in encoding of X1p and X2p .e. (A. Overall. i. by following same steps in encoding of ξ1p c2 α 1p ach . X2pr (W2pr ) as noise and similarly for . The codebook generation follows exactly same steps for X2p . After ﬁnding this sequence. ξ1 p PR + σ1p PSD1 ) where PSD1 is power of SD 1 and σ1p is the Costa parameter. W2p ). W2c ) + XRp (W1p . the allocations should hold 2 2 2 2 2 2 ξ1 c + ξ2c + ξ1p + ξ2p + ξ1p + ξ2p ≤ 1. the encoder searches for the sequence (U1 p − σ1p SD1 ) n that is nearly orthogonal to SD 1 . we generate 2 2 2 N (0.Y1 ) n X1 number of codewords denoted as U1 p such that U1p ∼ p .96). W2p ) + XR p (W1p . P2 n XRp = ξ1p α1p ξ2p PR n X1 p + α2p P1 PR n X2 p. these sequences are uniformly placed into 2nR1p bins. W2cd .117 small probability of error. PR n )X2 p P2 ach . 2. On the other hand. at each destination. using the ﬁnite-capacity links. W1pd .97) n and XR p is n n n the relay signal used to transmit the message splits (W1p . it’s transmitted signal can be written as. P2 (A. For example. i = 1. n n n n XR (V1 . To form X1p . W2p ) n where XRc = ξ1c α1c ξ2c PR n X1 c + α2c P1 PR n X2 c. at D1 . Note that the achievable region discussed in Remark 1. n n + a21 X2 p and SD2 = (a12 + Decoding: The relay immediately recovers corresponding splits from each source. the relay performs Costa encoding ξ2p n [12] by considering the interference signal SD 1 = (a21 + c1 α2p PR n )X2 p P2 + c1 X2p as the state known non-causally at the transmitter. the codeword to be n n n transmitted is obtained by X1 p = U1p − σ1p SD1 . (A. W2cr . i = 1. ξ2p n the states are given by SD 1 = (a21 + c1 α2p PR n )X1 p P1 n + a12 X1 p + c2 X1p . To obtain n n the codeword to be transmitted. W1pr ) are decoded jointly by treating n n the signals Xin p (Wip ). 2 and X2pd (W2pd ). we assume joint decoding of the desired message splits. the splits (W1cd . except the fact that the non-causally known state at D2 1p n is SD 2 = (a12 + c2 α1p ξ PR n )X1 p P1 n + a12 X1 p . As given in [12]. V2 ) = XRc (W1c . to satisfy the relay power constraint. with XR p = X1p + X2p with E[Xi2p ] ≤ ξi2p PR . then.

2. with Vi being a function of Wi .98) (A.t → V1 → XR.t PR. D2 simply treats X1p as noise as well in decoding W1p .t ≤ P1 . 2 ξ1 c PR 1 = n 1 n n 2 β1 c. Therefore the following Markov chains.t 2 Var(XR.t }. at Di . whereas XR.t PR.t } satisfy 1/n n t=1 (A. and is a function of (V1 . 2. the signal X1p is treated as noise.118 D2 . X1.t ≤ 1.t t=1 n 2 β2 c.t )PR.t and X2. the messages transmitted from Si to R.t ≤ PR .t t=1 (A.99) Var(X1.t )P1.t |V2 ) = (1 − ρ2 2c. The decoding of Wip .t ≤ 1 and ρ2 ic. for i = 1.t ≤ P2 and 1/n n t=1 2 PR.t → V2 → XR. 2. i = 1. 2 Var(XR. Moreover. we have βic.t )P2.t (A. i = 1.15 Proof of Proposition 22 We deﬁne as Vi ∈ Vi . We use the following deﬁnitions. V2 ). Hence.t hold. PR.t |V1 ) = (1 − ρ2 1c.t 2 2 where Pi.102) (A. i = 1.t |V1 ) = (1 − β1 c. 2 A.100) (A.t is the relay signal transmitted to both destinations at time t. the message splits W1p and W2p . 2 follows from Costa’s decoding [12]. are decoded at their desired destination without observing any interference from the remaining message splits.t = E {Xi. since Var(X ) ≥ 0. We also let. However.t Var(X2. 1/n n t=1 P2.t )PR.t = E {XR.103) 2 ξ2 c PR = . since in the n n n encoding of X2 p at the relay.101) P1.t |V2 ) = (1 − β2 c.

The following lemma.111) (A.t |V2 ) + n . Y1n |V2 .113) (A.t + c1 XR.t β1c. X2 ) P2. X2 )+n n ≤ h(Y1n |V2 .t ] ≤ ρ2c. i = 1.106) Then using Lemma 1 for the above Markov chains we obtain the correlations as..105) 2 α2 c P2 = ρ2 2c.114) n n n n = I (W1 . X2 )+n n (A. E [X1 ] ≤ ∞ and E [X2 ] ≤ ∞.t β2c.t t=1 n (A. then E [X1 X2 ] ≤ 2 E [X1 ] − Var(X1 |U ) 2 E [X2 ] − Var(X2 |U ) (A.t XR. E[X1.t P2.112) (A. nR1 = H (W1 ) n = H (W1 |V2 . Lemma 1 [46]: Given a Markov triple X1 → U → X2 with X1 and X2 2 2 with ﬁnite second moments.t We start with the constraint on R1 . (A. X2 ) + H (W1 |Y1n .e.119 1 = n 1 n n 2 α1 c P1 ρ2 1c.t P1. E[X2.t P1. 2.t + Z1. i.t PR. is essential in proving the outer bounds.t − (1 − ρ2 1c.t XR.t .115) (A. X2 )+n n n n n n n = h(a11 X1 + c1 XR + Z1 |V2 .116) ≤ t=1 h(a11 X1.109) (A.t ] ≤ P1.t − (1 − β1 c.t t=1 2 2 to be used throughout the proof and hence ξic ≤ 1 and αic ≤ 1.t 2 PR. Y1n |V2 .t (A. V2 .t PR.108) = ρ1c. X2 ) n ≤ I (W1 .t )P1.110) (A.t )PR.104) (A. which demonstrates relations among correlations of Markov chain elements.t and similarly.107) (A.

116) is from chain rule and removing conditioning does not reduce entropy. Cov(X1.t )PR.t P1.t t=1 n 2 ≤ C a2 11 P1 + c1 1 PR − n n 2 β2 c.t |V2 ) + n(A.t + c1 Var(XR.117) + 2a11 c1 Cov(X1.t β1c.t XR. and since E [X1.t ] = 0.t n (A. Moreover.t |V2 ) = E [X1. In (A.t β1c. (A.t XR.t i=1 + 2a11 c1 1 n n 1 ρ2 1c.t n (A.t by (A.t t=1 P1.t XR.99) and use further the equalities such that.t PR. we have.119) = 2 C a2 n 11 P1.t ) + c1 EV2 [Var(XR.t + 2a11 c1 ρ1c.t and V2 .t )PR.t |V2 = v2 ]E [XR.t |V2 ) + 2a11 c1 Cov(X1.120 n where (A.118) 2 C a2 11 Var(X1.121) + a2 11 P1 + 1 c2 1 n n (1 − t=1 2 β2 c.t |V2 = v2 )] t=1 n + 2a11 c1 EV2 [Cov(X1.t P1.t + Z1.t P1.t |V2 = v2 ] which follows from the independence of X1.t XR.t + c1 XR.123) where (A.t β1c.113) is from Fano’s in- equality and (A. Continuing.108).t 1 + 2a11 c1 n n ρ1c.t PR.119) is from Jensen’s inequality. R1 1 ≤ n ≤ 1 n n EV2 h(a11 X1.t ] ≤ ρ1c.t XR.t ] − E [X1.t + n PR.t PR.t XR.t |V2 = v2 ) ≤ 1 n 1 n n +n n (A.t PR.111) is from independence of W1 and V2 .t + c1 (1 − β2c.t |V2 = v2 )] + n n (A.t |V2 = v2 ] − E [X1.t |V2 = v2 ] =E [X1.118) is from the fact that entropy is maximized by Gaussian inputs for given variance constraints.t |V2 = v2 ) t=1 n (A.122) 2 β1 c. (A. we obtain Cov(X1.t XR.t + c1 Var(XR.122). we incorporate (A.t XR.t |V2 = v2 ) + n t=1 n 2 EV2 C a2 11 P1.t |V2 ) = EV2 E [X1. X2 .122) . (A.t t=1 n t=1 + n (A.120) t=1 1 ≤ n ≤ C n 2 2 C a2 11 P1.t ]EV2 E [XR.

131) is from the deﬁnition XR = fR (V1 . X2 n n ) ) + H (W1 |Y1n . X2 ) − h(Z1 )+n n n n (A.10a).102) and (A.123) is from Cauchy-Schwartz inequality. Y1n |V1 . V1 . Finally incorporating (A.124) is from independence of W1 and V2 . X2 = I (W1 .100) with Jensen’s inequality as done in (A.121 is from Jensen’s inequality. V2 .128) n n ≤ nC1 + I (W1 . V1 |V2 . with n → 0.t P1. X1 . Y1n |V1 . V 2 .124) (A. we obtain the bound on R2 given in (4.t |V1 ) − nh(Z1. X2 ) + I (W1 . with H (V1 ) ≤ nC1 . X2 .10c). . Y1n . X2 n ≤ I (W1 .130) (A. we obtain the bound in (4. we get the expression (A.126) (A. V2 ) and (A. V 1 . and n from the fact that H (V1 |W1 . V2 . Y1n |V1 .125) (A. V2 .104). X2 ) + I (W1 .130) is from n conditioning decreases entropy.134) n n n n n ≤ nC1 + h(a11 X1 + Z1 |XR .131) (A. X2 )+n n (A.t + Z1. V 1 . V2 .129) n n n ≤ nC1 + h(Y1n |V1 .t + n t=1 n ≤ nC1 + n n where (A. V1 . n ) nR1 = H (W1 |V2 . Another bound on R1 can be obtained as follows. (A. V1 |V2 . Using (A. since V1 is a function of W1 . V2 . (A. V2 .133) is from chain rule and the fact that conditioning decreases entropy. X2 ) = 0. Using similar steps. (A. Y1n .t ) + n 2 C a2 11 1 − ρ1c. X2 ) + n (A.126) is from Fano’s inn equality. X2 )+n n n (A. X2 ) − h(Y1n |W1 .134). X2 ) − H (V1 |W1 . X 2 )+n n n n ≤ nC1 + h(Y1n |XR . (A. V1 |V2 .129) is from independence of V1 and V2 .132) (A. V2 . V2 .127) n n ≤ I (W1 . X2 ) − h(Z1 )+n n ≤ nC1 + t=1 n h(a11 X1. V 2 . X2 )+n n n n = H (V1 |V2 .119). (A.133) (A. X2 .

we obtain the other bound for R2 given in (4.t Y1n for the system in (4. we consider the outer bound on the sum rate R1 + R2 .r.134) as. using deﬁnition 1 to obtain degradedness conditions of Y2n w.142) n n n n n n n n ) |W 1 .135) (A.139) is from the independence of W2 and (W1 . XR ) + 2n n (A. Y1n ) + I (W2 .141) n n ≤ h(Y1n ) − h(Y1n |W1 . .122 We further proceed from (A. Y1n ) + I (W2 .t P1. Y2n |W1 .t t=1 n + n 2 = C1 + C a2 11 (1 − α1c )P1 + where (A. X1 ) + H (W1 |Y1n ) + H (W2 |Y2n .137) follows by using (A.104). X1 ) n n n − h(Y2n |W1 . W (A. X1 ) (A. X 1 + Z2 + c2 XR ) + h(a22 X2 |W 1 . Using similar steps. Next.136) is from Jensen’s inequality. (A. X1 ) n ≤ I (W1 .136) (A. X1 ) + h(Y2n |W1 .1a) and (4. X1 ) + 2n n (A. R1 1 ≤ C1 + n ≤ C1 + C n 2 C a2 11 1 − ρ1c. [a21 c1 ] = µ[a22 c2 ]. X 2 . µ ≥ 1.138) (A. Now. X1 n a function of inequality and (A. and (A.139) n n = I (W1 .140) 1 .10b). n n(R1 + R2 ) = H (W1 ) + H (W2 |W1 . X 1 + Z1 + c1 XR = h(Y1n ) − h(a21 X2 n ) + 2n − h(Z2 n (A.t + t=1 n (A. For n → 0.1b). we get.143) n ).141) is from Fano’s where (A. Hence.137) a2 11 P1 − 1 a2 11 n n ρ2 1c.t P1.10d). X1 . Y2n |W1 .142) is from conditioning decreases entropy and X1 W1 . we obtain the bound (4. W2 .

123). (A.109) satisfy the following.148) is obtained by using (A.t +2 n (A.t (A.t ≤ 1.117)-(A. 2 2 2 ρ2 1c.t t=1 t=1 2 β1 c.t PR.146) where (A.144) + h(a22 X2 + c2 XR + Z2 |W 1 . using Jensen’s inequality. since jointly Gaussian rv’s. X1 ) 1 n n n n n ) + 2n (A.t t=1 t=1 2 β2 c. Moreover.t + t=1 n 1 a2 21 n n P2.148) 2 2 = C a2 11 P1 + a21 P2 + c1 PR + 2a11 c1 α1c ξ1c P1 PR (A. we use the correlations given in (A.t 1 + 2a21 c1 n n n ρ2 2c.147).t t=1 1 P1. where in (A. X 1 ) − log(µ2 ) − h(Z2 n 2 n ≤ h(Y1n ) − h(Z1 ) + 2n n (A.109).t t=1 n P2.123 Incorporating the condition (4. the correlations (A.105). n n n n n(R1 + R2 ) ≤ h(Y1n ) − h(a22 X2 + c2 XR + Z1 /µ|W1 .t β1c.147) +2 n 2 2 ≤ C a2 11 P1 + a21 P2 + c1 PR + 2a11 c1 ρ2 1c.t P2.12) into (A.145) is a consequence of worst-case result in [36] which holds for µ ≥ 1.149) + 2a21 c1 α2c ξ2c P2 PR + 2 n.102)-(A.t β2c. (A.145) n ≤ i=1 h(a11 X1i + c1 XRi + a21 X2i + Z1i ) − h(Z1i ) + 2n n (A.t P1.t β1c.148) is from Cauchy-Schwartz inequality.t + ρ2c.t PR.t β2c. X1i .t PR. (A. and (A. Similar to (A.t + 2a21 c1 n 1 n n ρ2c.150) .146) becomes (omitting the intermediate steps).t + t=1 1 c2 1 n n PR. X2i .108) and (A.143) we get. R1 + R2 ≤ C 1 a2 11 n n P1.t t=1 n 1 + 2a11 c1 n ρ1c.147).t PR.108) and (A. and XRi maximize the sum-rate in (A.

X1 . V2 ) − h(Z1 )+2 n (A. X1 . V2 ) n − h(Z2 )+2 n (A. V 1 .153) ≤ I (W1 . V2 ) + h(Y2n |W1 . for i = j = 1.155) n . we get. V1 .t ) − nh(Z1. X 1 . V2 ) + 2 n n (A.155) is from I (Wi . V1 . Y1n . and (A.156) is from conditioning decreases entropy.157) n n = nC1 + nC2 + h(Y1n |V1 . V2 ) − log(µ2 ) − h(Z2 )+2 (A.154) (A. X1 . X2 . Similar to (A. n n n n n(R1 + R2 ) ≤ nC1 + nC2 + h(Y1n |V1 . (A. Y2n .145). V 2 ) + 2 n (A. X1 ) n )+2 ≤ I (W1 . V2 ) − h(a22 X2 + c2 XR + Z1 /µ|W1 . V2 |W1 . X1 n (A. X1 .159) n where (A.156) (A. V1 ) + I (W2 . V2 ) 1 n n n n n + h(a22 X2 + c2 XR + Z2 |W 1 . V2 ) n n n − h(Y2n |X1 . W 2 . X1 .t |V1 . X1 . V 2 ) n n n n n + h(a22 X2 + c2 XR + Z2 |W 1 . Y2n |X1 n n ≤ nC1 + nC2 + h(Y1n |V1 .124 which lead to 2 2 2 2 α1 c ξ1c + α2c ξ2c ≤ 1. V2 ) + h(Y2n |X1 . Y2n .t ) + 2 n (A. V2 ) + nC2 + I (W2 . V1 .ea |W1 . V2 ) + 2 ≤ nC1 + I (W1 . V2 ) − h(Y1n |W1 .158) n n n n = nC1 + nC2 + h(Y1n |V1 . for n → 0. Y1n |V1 ) + I (W2 . X 1 .151) Therefore.160) n 2 n ≤ nC1 + nC2 + h(Y1n |V1 .10f). XR. n n(R1 + R2 ) = H (W1 ) + H (W2 |W1 . V 1 .152) (A. V 1 . V 1 . Vi ) ≤ H (Vi ) ≤ nCi and independence of Wi and Vj . V2 .153) is from Fano’s in- equality. V2 ) − h(Y1n |W1 . V2 ) − h(a21 X2 + c1 XR + Z1 |W1 . V2 ) n + I (W2 . V1 . Y1n |V1 . V2 ) − h(Z2 )+2 n (A. V1 . we obtain the bound in (4.162) . (A. using [a21 c1 ] = µ[a22 c2 ] with µ ≥ 1.161) n ≤ nC1 + nC2 + t=1 h(Y1. X1 . V1 ) + I (W1 . 2.152) is from independence of W2 and W1 . V1 . (A. XR . We give another bound on R1 + R2 as follows.

t |V2 ) + 2n (A. For n → 0. V2 |W1 . (A.167) from where (A. X1 ) + 2n n (A.171) n ≤ nC2 + h(Y1n |V2 ) + 2n n n ≤ nC2 + t=1 h(Y1.t + Z1.162).104) and (A. Y2n . V2 |W1 .125 where (A.170) is from Fano’s inequality.161) is from worst-case result of [36]. V2 |W1 . X1 . and from XR = fR (V1 . we obtain the rate in (4. (A. X1 n ) ≤ H (V2 ) ≤ nC2 . n n(R1 + R2 ) = H (W1 |V2 ) + H (W2 |W1 . W2 ) + 2n n (A.163) is from conditioning decreases entropy. X1 ) n + I (W2 . Y1n |V2 ) + I (W2 .165) is from Jensen’s inequality and by using (A.166) is from independence of W1 and V2 and W2 . V 2 .168) n ≤ h(Y1n |V2 ) − h(Y1n |W1 . (A. and (A. (W1 . (A. X1 .t ) + (A.172) n ). V 2 ) + 2n n (A.164) is from Jensen’s inequality (following similar steps used in showing (A. Y2n |W1 . X1 .105). X1 .165) where (A.10g).166) (A. Next.164) t=1 n 2 2 2 ≤ C1 + C2 + C a2 11 (1 − α1c )P1 + a21 (1 − α2c )P2 + 2 (A. R1 + R2 1 ≤ C1 + C2 + n ≤ C1 + C2 + 1 n n h(a11 X1.t + a21 X2. V2 ) n + h(Y2n |W1 . X1 .t |V1 . V2 ) n − h(Y1n |W1 . Proceeding from (A.162) is from chain rule and n conditioning decreases entropy. X1 .163) 2n t=1 n 2 2 2 C a2 11 (1 − ρ1ri )P1i + a21 (1 − ρ2ri )P2i + 2 n (A. V2 ) − h(Z1. consider the last bound on the sum-rate given below. V2 ) + I (W2 .169) n ≤ nC2 + h(Y1n |V2 ) − h(Y1n |W1 . X1 ) n ≤ I (W1 . V2 ) + nC2 + h(Y2n |W1 . V2 ) + 2n n (A.169) is from I (W2 . V2 ).170) (A.167) n = h(Y1n |V2 ) − h(Y1n |W1 .134)) and (A.

12). (A. A.t XR.t ] + a21 E [X2.t |V2 ) + 2n n (A. From (A.105).t )PR.102).t PR.10e). (A. (A. For 2 n → 0.104) and (A.109).175) is also from the Jensen’s inequality and incorporating the correlations given in (A.16 Proof of Proposition 23 In this proof.t |V2 ] + c1 E [XR.171) follows similar steps leading (A. which leads to Cov(X2i XRi |V2 ) = 0.175) P1 PR (A. .174) 1 a2 21 n ≤ C2 + C + 2a11 c1 1 n a2 11 P1 n + n (1 − t=1 ρ2 2c.173) 2 2 2 2 2 C a2 11 E [X1.t |V2 ] t=1 + 2a11 c1 Cov(X1.98)-(A.t |V2 ) + 2a21 c1 Cov(X2. we use the deﬁnitions in (A. we have.t )P2. we obtain the outer bound (4.t + 2n n + 1 c2 1 n 2 (1 − β2 c.t |V2 ) + 2n t=1 n n (A.t (A.145) using worst-case result in [36] for µ ≥ 1. We have Cov(X2i XRi |V2 ) = EV2 E [X2i XRi |V2 = v2 ] − E [X2i |V2 = v2 ]E [XRi |V2 = v2 ] .126 conditioning decreases entropy and non-negativity of entropy.176) is obtained by CauchySchwartz inequality as well as using (A.98)-(A. then EV2 E [X2i XRi |V2 = v2 ] = E [X2i |V2 = v2 ]E [XRi |V2 = v2 ].174) follows from Jensen’s inequality. Finally.t t=1 P1.176) 2 2 2 2 ≤ C2 + C a2 11 P1 + a21 (1 − α2c )P2 + c1 ξ1c PR + 2a11 c1 α1r ξ1c + 2n n where (A. and using the Markovity X2i → V2 → XRi holds.t β1c. (A. that is true for the condition (4.109).171).t t=1 n ρ1c. R1 + R2 ≤ C2 + ≤ C2 + 1 n 1 n n h(Y1.t XR.

KX2 matrix of XR given X1 2 and 1 R .179) n n ≤ h(Y1n ) − h(Y1n |W1 .181) n where (A. [36]. KX n |X n is the conditional covariance where KX2 2 1 R n n n n X n |X n is the conditional correlation matrix of X .177) is from independence of W2 and X1 . Y2n |X1 ) + H (W2 |Y2n . W2 . (A. by incorporating the conditions in (4. i. XR ) + 2n n (A.16). we know that τ is maximized by Gaussian inputs. n n(R1 + R2 ) = H (W1 ) + H (W2 |X1 ) (A. We let. From Proposition 8. (A.127 Consider the following bound on rate-sum. X 2 . we continue as. X(A.182) with µ < 1. Y1n ) + I (W2 .181) is due to cognitive relay n n with W1 → X1 → XR .178) is from Fano’s inequality.178) 1) n ≤ I (W1 . Y2n |X1 ) + 2n n (A. Y1n ) + H (W1 |Y1n ) + I (W2 .177) n n = I (W1 . n n n n n(R1 + R2 ) ≤ h(Y1n ) − h(µa22 X2 + µc2 XR + Z1 |X1 ) n n n n n + h(a22 X2 + c2 XR + Z2 |X1 ) − h(Z2 ) + 2n n (A. hence τ ≤ 1 2 n + c KX n |X n + 2a22 c2 KX n X n |X n + I log a2 2 22 KX2 1 2 1 R R 2 1 2 2 2 n + µ c KX n |X n + 2µ a22 c2 KX n X n |X n + I − log µ2 a2 2 22 KX2 1 2 1 R R 2 (A. X1 ) + h(Y2n |X1 ) n n n − h(Y2n |X1 .e.. Now.180) n n n n ≤ h(Y1n ) − h(a21 X2 + c1 XR + Z1 |X1 ) n n n n n + h(a22 X2 + c2 XR + Z2 |X1 ) − h(Z2 ) + 2n n (A.183) n n is the covariance matrix of X .180) is from conditioning decreases entropy and (A. n n n n n n n n τ = h(a22 X2 + c2 XR + Z2 |X1 ) − h(µa22 X2 + µc2 XR + Z1 |X1 ).

128

n XR , and I is the n × n identity matrix. Note that the corresponding power constraints

**lead to the trace constraints,
**

n } ≤ nP2 tr{KX2

(A.184)

n 2 (1 − β1 c,t )PR,t

n |X n } ≤ tr{KXR 1

(A.185) (A.186)

t=1 n

tr{K

n X n |X n X2 1 R

}≤

t=1

β2c,t

P2,t PR,t

n ) = E [X2,t XR,t ] where the (A.185) and (A.186) are due to the fact that Cov(X2,t XR,t |X1

**and follow from (A.109) with optimal ρ2c,t = 1, due to cognitive relay.
**

2 n + c KX n |X n + 2a22 c2 KX n X n |X n and For simplicity, denote KXtn = a2 2 22 KX2 1 2 1 R R n denote λXt ,t as the tth eigenvalue of KXtn , t = 1, ..., n. Note that tr{KX } is ﬁnite t

**due to the ﬁnite individual trace constraints given above. We argue that (as in
**

n [67]), for a given trace constraint on KX , τ is maximized for equal eigenvalues, i.e. t n n n λt = λu , u = t = 1, ..., n and therefore iid. inputs X1 , X2 and XR maximize 2 21 (A.183). Then, optimal eigenvalue is given by λ∗ t = a22 P2 + c2 n 1 2a22 c2 n n t=1 n 2 t=1 (1 − β1c,t )PR,t

+

β2c,t

**P2,t PR,t , and τ follows,
**

2 ∗ τ ≤ nC λ∗ t − nC µ λt .

(A.187)

**Also note that, since µ < 1, τ is an increasing function of λ∗ t , and we may further upper bound τ by increasing λ∗ t , such as, τ ≤ nC a2 22 P2 + 1 c2 2
**

n

n

(1 −

t=1 n

2 β1 c,t )PR,t

+ 2a22 c2

P2

1 n

n 2 β2 c,t PR,t t=1

− nC

µ2 a 2 22 P2

+

1 µ2 c 2 2

n

(1 −

t=1

2 β1 c,t )PR,t

+ 2µ a22 c2

2

P2

1 n

n 2 β2 c,t PR,t . t=1

(A.188) where (A.188) is due to Cauchy-Schwartz inequality.

129 Now, for incorporating the notation in (A.102) and (A.103) (however with a slight change of notation as ξ1c ξ1c and ξ2c ξ2c ), we obtain that, P2 PR P2 PR (A.189)

2 2 τ ≤ nC a 2 22 P2 + c2 (1 − ξ1c )PR + 2a22 c2 ξ2c 2 2 − nC a 2 21 P2 + c1 (1 − ξ1c )PR + 2a21 c1 ξ2c

Therefore, the overall outer bound on the sum-rate becomes, R1 + R2 ≤ 1 1 n h(Y1n ) + τ − h(Z2 )+2 n n n n 1 1 n ≤ h(Y1i ) + τ − h(Z2 )+2 n t=1 n ≤ C (a11 P1 + c1 ξ1c (A.190)

n

(A.191) P2 + c1 ξ2c P2 PR P2 PR +2

n

PR )2 + (a21

PR )2

2 2 + C a2 22 P2 + c2 (1 − ξ1c )PR + 2a22 c2 ξ2c 2 2 − C a2 21 P2 + c1 (1 − ξ1c )PR + 2a21 c1 ξ2c

(A.192)

where (A.191) is from chain rule and conditioning decreases entropy, (A.192) is using Jensen’s inequality along with using the notations in (A.102) and (A.103).

2 2 Now it is easy to check that (A.192) is an increasing function of ξ1 c + ξ2c ≤ 1, 2 2 hence incorporating ξ1 c = 1 − ξ2c and since n

→ 0, we obtain the outer bound (4.17a).

2

A.17

Proof of Proposition 25

Achievability: The essential part in the achievability is to obtain the condi-

tions such that destinations can decode the interference without limiting the interferencefree capacity region of the system. Since in our achievability scheme, the destinations are able to decode the interference, we assume αs = ξs = 0 where s ∈ {1p, 1p, 2p, 2p, 2p, 1p} in (4.6a)-(4.6f). On the other hand, in IC-OIR, since the sources transmit their signals both via relay and directly to the destinations, i.e. (W1c , W1c ) messages for S1 , we need two-fold conditions to obtain interference-free region.

130 Now consider the achievable region in (4.6a)-(4.6f). We modify this achievable scheme to involve sequential-decoding as opposed to simultaneous decoding used to obtain (4.6a)-(4.6f). Then, we have the following achievable rates, R1c ≤ C1 R1c ≤ C R1c ≤ C √ √ (A.193) (A.194) (A.195) (A.196)

(a12 α1c P1 + c2 ξ1c PR )2 √ √ 2 1 + a2 22 α2c P2 + (a22 α2c P2 + c2 ξ2c PR ) 2 a2 12 α1c P1 √ √ 2 1 + a2 22 α2c P2 + (a22 α2c P2 + c2 ξ2c PR ) P1 + c1 ξ1c PR )2

R1c ≤ C (a11 α1c

2 R1c ≤ C a2 11 α1c P1

(A.197) √ √ (A.198)

2

R2c ≤ C2 R2c ≤ C R2c ≤ C

P2 + c1 ξ2c PR ) √ √ (a11 α1c P1 + c1 ξ1c PR )2 1+ 2 a2 21 α2c P2 √ √ 2 1 + a2 11 α1c P1 + (a11 α1c P1 + c1 ξ1c PR ) P2 + c2 ξ2c PR )2

(a21 α2c 2 a11 α1c P1 +

(A.199) (A.200)

**R2c ≤ C (a22 α2c
**

2 R2c ≤ C a2 22 α2c P2

(A.201) (A.202)

Then, for (A.194)≥ (A.196), (A.195)≥ (A.197) and (A.199)≥ (A.201), (A.200)≥ (A.202), we obtain the very strong relay-interference conditions for all parameters of αic , αic , ξic , i = 1, 2, and the new rate region, with Ri = Ric + Ric , i = 1, 2, gives us

∗ ∗ ∗ ∗ the region in (4.18a)-(4.18d). Note that α1 c , α2c , ξ1c , ξ2c are the optimal power alloca-

tions giving us the sum capacity. Finally, evaluating the very strong relay-interference conditions for these optimal parameters, we obtain the conditions in (4.19a)-(4.19d). Converse: The converse follows directly from the Proposition 24 with the

prl serves as an outer bound on the sum capacity of IC-OIR. fact that Csum

2

Let αt and αt . Pk = P2 for t ∈ [2c.n ) where U1c . x2c . x1p and x2p using distributions n n n p(x1p. 1p.d.s with N (0. 2c. p(u2p. t ∈ [1c.n |u1c. i. 2c.n ). 2c. for each index s1c . 1p. u2p (s2p ) where st = {1.i. t ∈ [1c. P1 ). where wt ∈ 1. w2c ). 2N R2p i. the relay on the other hand transmits the signal xR = ξ1c u1c (s1c ) + √ (b) (b−1) (b) (b) (b) (b) and wt is the estiξ2c u2c (s2c ) + ξ1p u1p (s1p ) + ξ2p u2p (s2p ) where st = wt mate of wt at the relay. The source transmission corresponding to index t at block √ (b) (b) (b) (b) b is xt = αt x ˜t (st . The codewords of each codebook are denoted as u1c (s1c ). U2p are i. x1p (s1p . wt ) + αt Pk /PR ut (st ).n ). the sources transmit the previous message indices as well as current message indices. 2p]. X1p are i.d with N (0.d. u1p (s1p ). . N n=1 p(u2c. Each sequence is labeled as x1c (s1c .n ). 1p.i. 1p. 2N Rt .v. X2p are i. respectively we generate 2N R1c . p(x1c. .18 Proof of Proposition 28 The sources transmit B message indices using superposition block Markov encoding in B+1 blocks. and X2c . 2N R2c . BRt /(B + 1) → Rt . U1p . P2 ). 2c. Then.i. PR ). The codewords are generated using distributions N n=1 N n=1 p(u1c. be the power allocation coeﬃcients used for transmitting new message indices and previous message indices. 1p]. 2p] and st = wt (b) (b) (b−1) . respectively such that α1c + α1c + α1p + α1p = 1 and α2c +α2c +α2p +α2p = 1. such that ξ1c + ξ1p + ξ2c + ξ2p = 1. 2p] with Pk = P1 for t ∈ [1c. . 2p] denote the message indices. both the sources and the relay transmit a superposition of codewords for each (b) . The power allocation variables ξt . 1p. p(x2c. 2c. the achievable rates are are not reduced. w1c ). Gaussian r.n |u2c. 2p] perform proper power scaling of the cooperative signals ut (sb t ). Codebook Generation: We ﬁrst generate randomly four independent codebooks. s2p . . N n=1 p(u1p. x2c (s2c . . 1p. t ∈ [1c.n |un (s1p )). . .i. Encoding: In order to exploit cooperation with the relay. Overall.131 A. w1p ) and x2p (s2p .d. s2c . w2p ). 2N R1p .n |un (s2p )) where X1c . sequences x1c . t ∈ [1c. u2c (s2c ). Note that the cooperative codeword ut (st ) is properly scaled to satisfy the power constraints of the sources. s1p .n (s2c )). U2c . 2p]. √ (b) (b) For index t. t ∈ [1c.n (s1c )).e. Since B → ∞. . 2c. 2N Rt }. n p(x2p. 2p]. t ∈ [1c. Each block consists of N channel uses. with N (0.

at blocks b = 1. S2 ⊆ {1c. YD1 |XS . US ) 1 1 c c ≤ I (XS2 . P (wt = wt ) = . i. . U2p . Thus D1 considers the codeword with message w22 as noise. B + 1. t ∈ [1c. 2c. At the last block. . U2c . 1p. At the end of b=1. Note that D1 decodes the messages w1c . Exact probability of error calculations follow directly from [11] and hence omitted here.204) (A. 2 .e.203) complement set {Xj . Ri i∈S1 c c ≤ I (XS1 . The decoding proceeds backwards and assuming the messages are decoded correctly in block b + 1. 1p. w1p ) + α1p u1p (s1p ). US ) 2 2 (A. wt (b) (b−1) = wt (b−1) . i. w1c ) + α1c u1c (s1c )+ α1p x1p (s1p . Reliable decoding of wt . US1 . √ √ √ (b) (b) (b) (b) (b) (b) x2 (b) = α2c x2c (s2c . no new information is transmitted. w2c . wt can be decoded with arbitrarily small probability of error if (A. XS i∈S c is satisﬁed. U1p . the system is a multiple-access channel. 2c. t = [1c. assuming the messages transmitted in b − 1 are decoded correctly. 2c. 2p] corresponding to the ﬁrst and last blocks. if. B . . the messages in block b are decoded with low probability of error if (A.205) Ri i∈S2 where S1 ⊆ {1c. Decoding: Relay decodes the messages after each received block b. S ⊆ {1c. √ √ (b) (b) (b) (b) xR (b) = ξ1c u1c (s1c ) + ξ2c u2c (s2c ) + ξ1p u1p (s1p )+ ξ2p u2p (s2p ). j ∈ S c }. is received. w2c . In this case the corresponding messages can be decoded with low probability of error if. Note that st (1) = 1 and wt (B +1) = 1. YR |U1c . . Proceeding forward. respectively. 1p}. from the relay’s point of view. 2p}. YD2 |XS . w2c )+ α2c u2c (s2c ) + α2p x2p (s2p . . 2c. w2p . √ √ √ (b) (b) (b) (b) (b) (b) x1 (b) = α1c x1c (s1c . The destinations start decoding after the last block.132 t corresponding to. 2c. . b = B + 1. c . . such as D2 considers w1p message. XR ) Ri ≤ I (XS . 2p] is possible. . 1p.204) is satisﬁed. at the end of block b = 2. w2p ) + α2p u2p (s2p ).e. US2 . w1p and D2 decodes the messages w1c . where YR is the relay output.203) is satisﬁed. 2p} and XS denotes the (A.

. vol. Statist. 25. 5. Nov.133 Bibliography [1] C. vol. 1974. no. pp. no. and Prob. Armenian S.D. Jan. 60-70. 1978. CA. 1980.R. Inform.584. pp. IEEE Trans. 1. E. pp.” IEEE Trans. “Two-way communication channels”. 6. Salehi.” in Proc. 3. “Three-terminal communication channels. “Interference channels”. 1979. Ahlswede. Carleial. 120-154. Han and K. Theory. [9] T. E. Cover and A. Ahlswede. 1. . 24. vol. vol. Inf. June 1960. 49-60. Fourth Berkeley Symp. Oct. “Multi-way communication channels. pp. no.” Advances in Applied Probability. dissertation. no. “Information ﬂow in relay networks. [4] R. 1971. 1971. IEEE Trans.. 1980 [8] T. Sep. “A new achievable rate region for the interference channel. “The capacity region of a channel with two senders and two receivers. Cover. no. Theory. pp. [6] T. “Capacity theorems for the relay channel”. [2] E. [3] R.” Ph. on Math. Sept. “Multiple access channels with arbitrarily correlated sources”. IEEE International Symposium on Information Theory. Kobayashi. Stanford.S. IEEE Trans. Proc. California. El Gamal.. Tsahkadsor. Stanford Univ. vol. Theory. Inform. 1981. vol. 2. El Gamal and M. 5. 27. van der Meulen. A. Shannon. Aref. [7] M. 572. Jan.. Berkeley. 648-657. Inform. Theory. [5] A. C. R. 26.” Annals Probabil.

41. 3. “The discrete memoryless multiple-access channel with cribbing encoders”. pp. Sept. “Informationtheoretical Results for the Discrete Memoryless Multiple Access Channel. ” IEEE Trans.” IEEE Trans. no. no. E. J. Theory. [14] M.” IEEE Trans. Willems and E.” Doctor in de Wetenschappen Proefschrift dissertation. Cover and J. Telatar. 33. “Capacity of multiple-antenna Gaussian channels. [17] T. vol. H. A. J. vol. [19] I. Inform. 33. Sep. “Writing on dirty paper. pp. 5. Belgium. 1985. Inform. no. Theory. 31. 10. vol. [13] F. p. 1999. Inf. vol. [11] F. Katholieke Univ. 6. ”The capacity region of the discrete memoryless interference channel with strong interference”. A. M. 1987. Theory. 710-711. Inform. H. “On the Gaussian interference channel. 31. Costa. Thomas. 1996. pp. p. New York: Wiley. Costa. 1983.134 [10] H. “Feedback can at most double Gaussian multiple access channel capacity”. Oct. Gamal. 439-441. H. 27. Leuven.” Europian Transactions on Telecommunications. van der Meulen. IEEE Trans. Theory. no. 5. IEEE Trans. vol. Thomas. 585. vol. Inform. Oct. [12] M. [16] J. Sept. pp. Sato.” Bell Laboratories Technical Journal. 607-615. 786-788. M. 711-716. “Layered space-time architecture for wireless communication in a fading environment when using multi-element antennas. . [18] G. Theory. E.” IEEE Trans. vol. May 1985. Theory. 1981. Nov. 27. pp. Elements of Information Theory. Costa. 1991. [15] M.Willems. 313-327. Inform. M. Nov. Leuven. 5. M. Foschini. 1982. pp. no. 1987. “The capacity of the Gaussian interference channel under strong interference.

5. Gupta. E. no. [26] G. W. IEEE Trans. 11. Theory. “Review of rate regions for interference channels. vol. 3062-3080. Theory. 2006. 12. Wu. IEEE International Symposium on Information Theory. Switzerland. 2005. Steinberg. vol. El Gamal. 51. “Toward an information theory of large networks: An achievable rate region. and G. IEEE Trans.” IEEE Trans. Kramer. Theory. R. D. Erkip and B. S. “Integrating heterogeneous wireless technologies: a cellular aided mobile ad hoc network (CAMA)”. no. Y. [27] Y. [24] B. Nov. no. “Capacity of a class of relay channels with orthogonal components”. Bhargava. 9. Lu and W. Theory. June 2004. Inform. Zurich. Aug. Sept. no. M. Aazhang. Inform. no. C. Gastpar and P. 2004. May 2005. Inform. no. pp. no. 51. 51. Journal on Mobile Networks and Applications. vol.1356. N. 49. vol. pp. “On achievable rate regions for the Gaussian interference channel”.” IEEE Trans. Aug. N. “Cooperative strategies and capacity theorems for relay networks”. vol. 1927-1938. Laneman. Sason. Tse. pp. Int’l Zurich Seminar on Communications. pp. Wornell. 2004. [25] A. “Achievable rates for the broadcast channel with states known at the transmitter” in Proc. [22] I. pp. Australia. 50. Feb. 1345 . [28] G.135 [20] P.. Adelaide. Zahedi. Kramer. Kumar. 4. Comm. vol. Theory.IEEE Trans. X. 393-408. S. . 18771894. 5. vol. May 2005. “User cooperation diversity-Part I: System description. 3037-3063. 50. Dec. pp. 1815-1817. 8. Gupta and P. [23] J. “Cooperative diversity in wireless networks: Eﬃcient protocols and outage behavior. pp. Sendonaris. 2003 [21] A. Inform. 2003. Inf. Shamai.”Proc. 6. Wang.” IEEE Trans.

136 [29] D. Tuninetti, “The interference channel with generalized feedback (IFC-GF)”, in Proc. Information Theory and Applications Workshop, UCSD, California, Feb. 2006. [30] N. Devroye, P. Mitran, and V. Tarokh,“Achievable rates in cognitive radio channels,” IEEE Trans. Inform. Theory, vol. 52, no.5, pp. 1813-1827, May 2006. [31] A. Jovicic, P. Viswanath, “Cognitive radio: An information-theoretic perspective,” Proc. IEEE International Symposium on Information Theory, Seattle, USA, July 2006. [32] H. Weingarten, Y. Steinberg and S. Shamai, “The capacity region of the Gaussian multiple-input multiple-output broadcast channel”, IEEE Trans. Inform. Theory, vol. 52, no. 9, pp. 3936-3964, Sept. 2006. [33] H. Weingarten, T. Liu and S. Shamai, “The capacity region of degraded multiple input multiple output compound broadcast channel”, submitted to IEEE Trans. Inform. Theory, Jan. 2007. [34] O. Sahin, E. Erkip, “Dynamic resource allocation for multi source-destination relay networks”, Proc. 41st Annual Conference on Information Sciences and Systems, Baltimore, MD, Feb. 2007. [35] Y. Liang and V.V. Veeravalli, “Cooperative relay broadcast channels”, IEEE Trans. Inform. Theory, vol. 53, no. 5, pp. 900-928, May 2007. [36] T. Liu and P. Viswanath, “An extremal inequality motivated by multi terminal information theoretic problems,”, IEEE Trans. Inform. Theory, vol. 53, no. 5, pp. 1839-1851, May 2007. [37] T.M. Cover and Y.H.Kim, “Capacity of a class of deterministic relay channels”,in Proc. IEEE International Symposium on Information Theory, Nice, France, June 2007.

137 [38] B. Nazer and M. Gastpar, “The case for structured random codes in network communication theorems”, Proc. Information Theory Workshop, Lake Tahoe, California, Sept. 2007. [39] Y. Liang and G. Kramer, “Rate Regions for Relay Broadcast Channels”, IEEE Trans. Inform. Theory, vol. 53, no. 10, pp. 3517-3535, Oct. 2007. [40] W. Wei, S. Vishwanath, A. Arapostathis,“ Capacity of a class of cognitive radio channels: Interference channels with degraded message sets,” IEEE Trans. Inform. Theory, vol. 53, no. 11, pp. 4391-4399, Nov. 2007. [41] O. Sahin and E. Erkip, “On achievable rates for interference relay channel with interference cancellation,” Proc. 41st Annual Asilomar Conference on Signals, Systems, and Computers, Paciﬁc Grove, CA, Nov. 2007. [42] G. Kramer, Topics in Multi-User Information Theory, Foundations and Trends in Communications and Information Theory, vol. 4, no. 4-5, pp. 265-444, 2007. [43] O. Sahin and E. Erkip, “Achievable rates for the Gaussian interference relay channel,”Proc. Global Telecommunications Conference, Washington, DC, Nov. 2007. [44] S. Annapureddy and V. Veeravalli, “Sum capacity of the Gaussian interference channel in the low interference regime,” in Proc. Information Theory and Applications Workshop, UCSD, California, Feb. 2008. [45] G. Kramer, Topics in Multi-User Information Theory, now Publishers, 2008. [46] M. Wigger, Cooperation on the multiple-access channel, PhD thesis, ETH Zurich, Sept. 2008. [47] W. Nam, S.Y. Chung and Y.H. Lee, “Capacity bounds for two-way relay channel”, Proc. Int’l Zurich Seminar on Communications, Zurich, Switzerland, March 2008.

138 [48] M. P. Wilson, K. Narayanan, H. Pﬁster and A. Sprintson, “Joint physical layer coding and network coding for bi-directional relaying”, submitted to IEEE Trans. Inform. Theory, April 2008. [49] R. Dabora, I. Maric and A. Goldsmith, “Relay strategies for interferenceforwarding”, in Proc. IEEE Information Theory Workshop, Porto, Portugal, May 2008. [50] A. Sanderovich, S. Shamai, Y. Steinberg and G. Kramer, “Communication via decentralized processing”, IEEE Trans. Inform. Theory, vol. 54, no. 7, pp. 30083023 , July 2008. [51] H. Chong, M. Motani, H. Garg and H. El Gamal, “On the Han-Kobayashi region for the interference channel”, IEEE Trans. on Inform. Theory, vol. 54, no. 7, pp. 3188-3195, July 2008. [52] I. Maric, R. Dabora and A. Goldsmith, “On the capacity of the interference channel with a relay,” in Proc. IEEE International Symposium on Information Theory, Toronto, Canada, July 2008. [53] S. Sridharan, S. Vishwanath, S.A. Jafar and S. Shamai, “On the capacity of cognitive relay assisted Gaussian interference channel” in Proc. IEEE International Symposium on Information Theory, Toronto, Canada, July 2008. [54] S. I. Bross, A. Lapidoth and M. A. Wigger, “The Gaussian MAC with conferencing encoders,” in Proc. IEEE International Symposium on Information Theory, Toronto, Canada, July 2008. [55] I. Maric, R. Dabora and A. Goldsmith, “On the capacity of the interference channel with a relay,” in Proc. IEEE International Symposium on Information Theory, Toronto, Canada, July 2008.

CA. Vishwanath. New Orleans. “Cognitive relaying with one-sided interference”. 5534-5562. July 2008. . CA. “Encoding against an interferer’s codebook. in Proc. IEEE Global Telecommunications Conference. Etkin. Zhou. and Computers. Sridharan. Global Telecommunications Conference. Systems.” submitted to IEEE Trans. Theory. Maric and A. Proc. [58] I. Maric. Maric. vol. Paciﬁc Grove. Wang. Sept. “The gateway channel: Outage analysis.” Proc. N. IEEE Trans. Sahin and E. Liu and A. I. N. Abouelseoud and A.” Proc. LA. Shamai. 2008. Goldsmith. Nov. “On the capacity of cognitive relay assisted Gaussian interference channel” in Proc. Dec. [57] W. 2008. 2008. [61] I. LA. Yu and L. Erkip. Nov. Canada. S. Inform. 12. [63] M. 54.” in Proc. Goldsmith. Monticello. R. IEEE International Symposium on Information Theory. Jafar and S. no. 2008. pp.H. 42nd Annual Asilomar Conference on Signals. “Interference forwarding in multiuser networks. Toronto. Nov. S. Control and Computing. 42nd Annual Asilomar Conference on Signals. Tse and H. D. [60] R. “Gaussian Z-interference channel with a relay link: achievability region and asymptotic sum capacity. 2008. IL. Dabora and A.A. Allerton Conference on Communications. 2008. “Gaussian interference channel capacity to within one bit”. Theory. Goldsmith.139 [56] S. on Inform. [62] R. [59] O. Systems. and Computers. Paciﬁc Grove. Sept. Dec. C. 2008. Nosratinia. Dabora.“Generalized relaying in the presence of interference”. New Orleans.

Chen. 2009. Kramer and H. V. A. Feb. Mohasseb. Feb. [68] D. Seoul. Feb. Erkip and O. S. South Korea. Shamai. Inform. Theory. Inform.140 [64] X. “Capacity bounds for the Gaussian interference channel”. pp. [67] A.V. [70] X. Poor. June 2009. vol. Kramer. 620-643. Sahin. E. Chen. A. Information Theory and Applications Workshop. “Interference channel with a relay: models. June 2009. “The MIMO wireless switch: Relaying can increase the multiplexing gain”. Feb. Seoul. Chung. in Proc. 689-699. Khandani. Choi and S. “A new outer bound and noisy-interference sum-rate capacity for the Gaussian interference channels. “On the separability of parallel Gaussian interference channels”. G¨ und¨ uz. March 2009. Yener. [66] O. [71] S. no. Proc. UCSD. 55. 2009. H. Inform. IEEE International Symposium on Information Theory. G. bounds”. Shang. California. submitted to IEEE Trans. in Proc. . Simeone. 2009. Theory. pp. H. G. no. in Proc.” IEEE Trans. relaying strategies. K. South Korea. O. IEEE International Symposium on Information Theory. Simeone. IEEE Trans. on Inform. 2009.V. 2. Theory. A. [69] H. W. 55. “Multiple multicasts with the help of a relay”. Shang. vol. Gamal and G. B. Theory. Poor and S. E. Kramer and B. “Noisy-interference sum-rate capacity of parallel Gaussian interference channels”. Poor. 2. IEEE International Symposium on Information Theory. June 2009. Y. Goldsmith and H. Seoul. Goldsmith. submitted to IEEE Trans. Gunduz. [65] D. “Multi-way relay channel”. Ghozlan. Motahari and A. South Korea.

Simeone and E. to appear Proc. IEEE Trans. . on Inform. Simeone and E. Erkip and H. Sahin. [73] L.A. Tian and A. Chen. Erkip. on Inform. Theory. V. “Interference channel aided by an infrastructure relay. June 2009. Theory. “Capacity regions and sum-rate capacities of vector Gaussian interference channels”. Inform. [75] V. IEEE Trans. R. No. B. [76] V. Honolulu. Shang. 8. Jafar. to be submitted to IEEE Trans. pp. on Inform. Sankar. 55.141 [72] O. 2009. “The Gaussian interference relay channel with a potent relay”. Theory. Poor. Korea. Cadambe and S. G. ” in Proc. Aug. V. submitted to IEEE Trans. Seoul. Cadambe and S. submitted to IEEE Trans. Jafar. Sahin. O. 54. E. 3425-3441. IEEE Global Telecommunications Conference. Shang. “Interference alignment and degrees of freedom of the K -user interference channel”. Theory. “Interference channel with an out-of-band relay”. “Ergodic fading interference channels: sum-capacity and separability”. 3983-3990. X. Dec. June 2009. vol. Poor. vol. 9. [77] Y. pp. on Inform. [78] O. no. [74] X. Erkip. July 2009. A. 2009. Theory. IEEE International Symposium on Information Theory. “Parallel Gaussian interference channels are not always separable”.R. O. Kramer and H. 2009. Yener. Hawaii. Sep.

Austin. “Cooperative interference management: The role of cognitive relaying”.142 List of Publications 1. Systems. Information Theory and Applications Workshop. on Inform. Seoul. Simeone and E. O. Croix. O. 5. E. in Proc. Paciﬁc Grove. Proc. “Interference channel with a half-duplex out-ofband relay”. in Proc. IEEE Trans. IEEE Communication Theory Workshop (invited talk). Erkip. “Interference channel with a relay: models. “Gaussian interference channel aided by a relay with out-of-band reception and in-band transmission”. Sahin. June 2009. relaying strategies. Erkip. Erkip. IEEE International Symposium on Information Theory. ” in Proc. Sahin and E. 7. O. Simeone. O. and Computers. Erkip. UCSD. O. June 2010. May 2008. on Communications. 3. TX. IEEE International Symposium on Information Theory. USVI. O. Sahin. O. O. Theory. Sahin. 4. 42nd Annual Asilomar Conference on Signals. Sahin. Erkip and O. Simeone and E. bounds”. St. E. “Interference channel with an out-of-band relay. To appear. O. 6. Korea. . “Interference channel aided by an infrastructure relay. ” under review. California. under review. 2009. CA. Nov. Simeone.Erkip. “Cognitive relaying with one-sided interference”. Feb. Sahin. 2008. 2. O. Erkip. Sahin and E. O. IEEE Trans. Simeone and E.

11. O. Sahin. Global Telecommunications Conference. Feb. Erkip. Paciﬁc Grove. DC.” in Proc. 41st Annual Conference on Information Sciences and Systems. in Proc. Goodman. “On achievable rates for interference relay channel with interference cancellation. 2007. J. May 2006. and Signal Processing. Baltimore. IEEE International Conference on Acoustics. Nov. September 2005. Sahin. Ghosh. Sahin and E. E. Sahin and E. Workshop on Cooperative Communications. Speech. Polytechnic University. Erkip. Nov. “Iterative power control for wireless multimedia communications”. O. E. 2007. Systems. Washington. CA. Erkip and D. O. and Computers. Sahin. 9. 2007. 41st Annual Asilomar Conference on Signals. Toulouse. “Achievable rates for the Gaussian interference relay channel. “Dynamic resource allocation for multi source-destination relay networks”. Erkip.143 8. 12. O. Brooklyn.”Proc. in Proc. 10. MD. Yang and M. O. “An eﬃcient relaying scheme for multiple users in spectral eﬃcient cooperative networks”. France. .