Rate-Splitting Multiple Access for 6G

Prof. Bruno Clerckx

Communications and Signal Processing Group

Department of Electrical and Electronic Engineering
Imperial College London

Thanks to my PhD students, postdocs and many collaborators at TU

Eindhoven,
University Erlangen-Nuremberg, Ruhr University Bochum, TU
Braunschweig,
University of Bremen, TU Darmstadt, Princeton, HKU, Xidian, NUDT, TU
Berlin,
Eurecom, UCL, KCL, Edinburgh, Surrey, York, PUC-Rio, Pusan,
Southeast,
Imperial College London

2
Outline

1 (Existing) Multiple Access Techniques

2 Rate Splitting Multiple Access (RSMA)

3 Applications/Interplay of RSMA

4 Physical Layer Design

5 Conclusions

3
Outline

1 (Existing) Multiple Access Techniques

2 Rate Splitting Multiple Access (RSMA)

3 Applications/Interplay of RSMA

4 Physical Layer Design

5 Conclusions

4
Multiple Access Techniques
• Wireless communication systems are multi-user

• How to serve multiple users?

Broadcast Channel - Multiple Access Channel -

Downlink Uplink

5
Multiple Access Techniques
• Transmitter(s) send(s) independent streams to receiver(s).
𝒉𝒉1 User-1 𝒉𝒉1 User-1
⋮ ⋮
𝒉𝒉𝑘𝑘 𝒉𝒉𝑘𝑘
Transmitter ⋮ User-k Receiver ⋮ User-k
⋮ ⋮
𝒉𝒉𝐾𝐾 𝒉𝒉𝐾𝐾
M antennas User-K M antennas User-K
K single-antenna users K single-antenna users

• Maximizing Rate:
per-user, (weighted) sum-rate,
min rate, etc
• Under high SNR (P)
𝑅𝑅𝑘𝑘 ≈ 𝑑𝑑𝑘𝑘 log 2 𝑃𝑃 + 𝒪𝒪(1)
� 𝑅𝑅𝑘𝑘 ≈ 𝑑𝑑𝑠𝑠 log 2 𝑃𝑃 + 𝒪𝒪(1)
𝑘𝑘

DoF

6
Reminder from Information Theory
• Transmitter sends independent streams to multiple receivers.
𝒉𝒉1 User-1
⋮
𝒉𝒉𝑘𝑘
Transmitter ⋮ User-k
⋮
𝒉𝒉𝐾𝐾
M antennas User-K
K single-antenna users
• SISO BC (𝑀𝑀 = 1):
- Users can be ordered according to strength
- Capacity achieving strategies:
☆ Superposition coding with successive interference cancellation
(SC-SIC) [Cover IEEE TIT 1972]
☆ Dirty paper coding (DPC) [Costa IEEE TIT 1983]
• MISO and MIMO BC (𝑀𝑀 > 1) :
- Users cannot be ordered Assumptions:
- SC-SIC (NOMA) leads to performance loss Perfect CSI
- Capacity achieving strategies:
☆ DPC [Weingarten, Y. Steinberg, S.S. Shamai IEEE TIT 2006]
7
 Existing Multiple Access Techniques

OMA SDMA NOMA

Orthogonality Non-orthogonality

2G 2G

3G
4G, 5G

OMA: eliminate multi-user interference by allocating orthogonal radio resources.

8
 Existing Multiple Access Techniques

OMA SDMA NOMA

Orthogonality Non-orthogonality

Space division multiple access (SDMA):

• Multiplex users in spatial domain using multi-user linear precoding
(MU-LP)
• Used in MU/massive/mmw/network-MIMO

4G, 5G
together with OFDMA

SDMA: manage multi-user interference by spatial precoding at the transmitter

and fully treating interference as noise at receivers. 9
 Existing Multiple Access Techniques

OMA SDMA NOMA

Orthogonality Non-orthogonality

Power-domain non-orthogonal multiple access (NOMA):

• Use power domain to break the orthogonality
• Apply superposition coding (SC) at the transmitter and successive
interference cancellation (SIC) at each receiver

Study item,
no work
item in 5G
NOMA: manage multi-user interference by fully decoding interference at
receivers.
10
 Existing Multiple Access Techniques

OMA SDMA NOMA

Orthogonality Non-orthogonality

Multi-Antenna NOMA with G (<K) groups:

One user fully decodes the message(s) of other user(s)

Careful of the inefficient use of multi-antenna and SIC receivers! [1]
[1] B. Clerckx, Y. Mao, R. Schober, E. Jorswieck, D. J. Love, J. Yuan, L. Hanzo, G. Ye Li, E. G. Larsson, and G. Caire. "Is NOMA
Efficient in Multi-Antenna Networks? A Critical Look at Next Generation Multiple Access Techniques." IEEE OJ-COMS 2021. 11
Outline

1 (Existing) Multiple Access Techniques

2 Rate Splitting Multiple Access (RSMA)

3 Applications/Interplay of RSMA

4 Physical Layer Design

5 Conclusions

12
 Rate Splitting Multiple Access

Orthogonal vs. Non-Orthogonal: not the problem

• SDMA (4G/5G) is non-orthogonal: users interfere

Real problem: how is interference managed

• OMA: no interference
• SDMA: treat interference as noise
• NOMA: fully decode interference

Lessons from Information Theory: [Etkin, Tse, Wang, IEEE TIT 2008]
treat as noise partially treat as noise, partially decode fully decode
weak strong
Re-think Multiple Access [1]
Weak Medium Strong

SDMA, 4G, 5G

NOMA

RSMA

[1] Y. Mao, O. Dizdar, B. Clerckx, R. Schober, P. Popovski, H. V. Poor, “Rate-Splitting Multiple Access: Fundamentals, Survey, and Future
13
Research Trends,” arXiv:2201.03192
Rate Splitting Multiple Access
Rate Splitting Multiple Access (RSMA) a general, flexible and robust multiple access [1]

Builds upon rate splitting (RS) concept originally introduced for SISO interference
channel [Carleial IEEE TIT 1978, Han Kobayashi IEEE TIT 1981]
treat as noise partially treat as noise, partially decode fully decode
SDMA RSMA NOMA

RSMA partially decodes interference and partially treats interference as noise.

[1] Y. Mao, B. Clerckx, and V.O.K. Li, "Rate-splitting multiple access for downlink communication systems: bridging, generalizing, and
14
outperforming SDMA and NOMA," EURASIP Journal on Wireless Communications and Networking, 133 (2018). [Best paper Award 2022]
Downlink RSMA: Two-User Example
𝑔𝑔𝑐𝑐,1
𝑦𝑦1 𝑠𝑠̂ 𝑐𝑐 �𝑐𝑐 �𝑐𝑐, 1
Decode 𝑊𝑊 𝑊𝑊

Split
𝑊𝑊𝑐𝑐,1 ×

combiner
Schedule two users
𝑥𝑥1

Message
Message splitter

Linear precoder

Combine
𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐 SIC Encode
𝑊𝑊1
𝑊𝑊𝑐𝑐,2 + �1
𝑊𝑊

Encoder
MIMO + Channel Precode
-

channel 𝑔𝑔1
𝑊𝑊2 𝑊𝑊𝑝𝑝,1 𝑠𝑠1 𝑠𝑠̂ 1 � 𝑝𝑝, 1
𝑊𝑊
× Decode
𝑊𝑊𝑝𝑝,2 𝑠𝑠2 𝑥𝑥2
User-1
𝑦𝑦2 �2
Transmitter 𝑊𝑊
User-2

Transmitter Receiver
• Message splitting:
split split • Both users first decode 𝑠𝑠𝑐𝑐 by
𝑊𝑊1 𝑊𝑊𝑐𝑐,1 , 𝑊𝑊𝑝𝑝,1 and 𝑊𝑊2 𝑊𝑊𝑐𝑐,2 , 𝑊𝑊𝑝𝑝,2 treating 𝑠𝑠1 and 𝑠𝑠2 as noise.
• Creating the common message:
combine
𝑊𝑊𝑐𝑐,1 , 𝑊𝑊𝑐𝑐,2 𝑊𝑊𝑐𝑐 • Both users perform SIC and then
• Independent encoding: decode 𝑠𝑠1 and 𝑠𝑠2 , respectively.
encode encode encode
𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐 , 𝑊𝑊𝑝𝑝,1 𝑠𝑠1 , 𝑊𝑊𝑝𝑝,2 𝑠𝑠2
• Transmit signal: •• Rate
Rate of
of user-k
user-k isis split:
split:
𝐱𝐱 = 𝐩𝐩𝑐𝑐 s𝑐𝑐 + 𝐩𝐩1 s1 + 𝐩𝐩2 s2 rate of 𝑠𝑠𝑘𝑘𝑘𝑘 +part of the rate of 𝑠𝑠𝑐𝑐𝑐𝑐
• From 2 messages, we generate 3 streams! 15
SDMA and NOMA: subsets of RSMA
𝑔𝑔𝑐𝑐,1
𝑦𝑦1 𝑠𝑠̂ 𝑐𝑐 �𝑐𝑐 �𝑐𝑐, 1
Decode 𝑊𝑊 𝑊𝑊

Split
𝑊𝑊𝑐𝑐,1 ×

combiner
Schedule two users
𝑥𝑥1

Message
Message splitter

Linear precoder

Combine
𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐 SIC Encode
𝑊𝑊1
𝑊𝑊𝑐𝑐,2 + �1
𝑊𝑊

Encoder
MIMO + Channel Precode
-

channel 𝑔𝑔1
𝑊𝑊2 𝑊𝑊𝑝𝑝,1 𝑠𝑠1 𝑠𝑠̂ 1 � 𝑝𝑝, 1
𝑊𝑊
× Decode
𝑊𝑊𝑝𝑝,2 𝑠𝑠2 𝑥𝑥2
User-1
𝑦𝑦2 �2
Transmitter 𝑊𝑊
User-2

SDMA based on MU-LP NOMA based on SC-SIC

• Allocate no power to 𝑠𝑠𝑐𝑐 • Allocate no power to 𝑠𝑠2
• Independent encoding: • Independent encoding:
encode encode encode encode
𝑊𝑊1 𝑠𝑠1 , 𝑊𝑊2 𝑠𝑠2 𝑊𝑊1 𝑠𝑠1 , 𝑊𝑊2 𝑠𝑠𝑐𝑐
• Transmit signal: • Transmit signal:
𝐱𝐱 = 𝐩𝐩1 𝑠𝑠1 + 𝐩𝐩2 𝑠𝑠2 𝐱𝐱 = 𝐩𝐩𝑐𝑐 𝑠𝑠𝑐𝑐 + 𝐩𝐩1 𝑠𝑠1

Common stream ≠ Multicast stream

Clarification: The common stream in RSMA is decoded by all users but it is not
necessarily intended for all users. NOMA has common streams as well! 16
Mapping of Messages to Streams
𝑔𝑔𝑐𝑐,1
𝑦𝑦1 𝑠𝑠̂ 𝑐𝑐 �𝑐𝑐 �𝑐𝑐, 1
Decode 𝑊𝑊 𝑊𝑊

Split
𝑊𝑊𝑐𝑐,1 ×

combiner
Schedule two users
𝑥𝑥1

Message
Message splitter

Linear precoder

Combine
𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐 SIC Encode
𝑊𝑊1
𝑊𝑊𝑐𝑐,2 + �1
𝑊𝑊

Encoder
MIMO + Channel Precode
-

channel 𝑔𝑔1
𝑊𝑊2 𝑊𝑊𝑝𝑝,1 𝑠𝑠1 𝑠𝑠̂ 1 � 𝑝𝑝, 1
𝑊𝑊
× Decode
𝑊𝑊𝑝𝑝,2 𝑠𝑠2 𝑥𝑥2
User-1
𝑦𝑦2 �2
Transmitter 𝑊𝑊
User-2

RS is a more general framework: RS ⊃ SDMA/OMA/NOMA/Multicast

B. Clerckx, Y. Mao, R. Schober, and H. V. Poor, “Rate-splitting unifying SDMA, OMA, NOMA, and multicasting in MISO broadcast 17
channel: A simple two-user rate analysis,” IEEE Wireless Communications Letters, vol. 9, no. 3, pp. 349–353, 2020.
 Rate-Splitting Example
How to split user messages? [1]
𝑔𝑔𝑐𝑐,1
𝑦𝑦1 𝑠𝑠̂ 𝑐𝑐 �𝑐𝑐 �𝑐𝑐, 1
Decode 𝑊𝑊 𝑊𝑊

Split
𝑊𝑊𝑐𝑐,1 ×

combiner
Schedule two users

𝑥𝑥1

Message
Message splitter

Linear precoder

Combine
𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐 SIC Encode
𝑊𝑊1
𝑊𝑊𝑐𝑐,2 + �1
𝑊𝑊

Encoder
MIMO + - Channel Precode
channel 𝑔𝑔1
𝑊𝑊2 𝑊𝑊𝑝𝑝,1 𝑠𝑠1 𝑠𝑠̂ 1 � 𝑝𝑝, 1
𝑊𝑊
× Decode
𝑊𝑊𝑝𝑝,2 𝑠𝑠2 𝑥𝑥2
User-1
𝑦𝑦2 �2
Transmitter 𝑊𝑊
User-2

User-1 User-2
• 4-bit user message: • 3-bit user message:
𝑊𝑊1 =(𝑎𝑎1 𝑎𝑎2 𝑎𝑎3 𝑎𝑎4 ) ∈ 𝒲𝒲1 𝑊𝑊1 =(𝑏𝑏1 𝑏𝑏2 𝑏𝑏3 ) ∈ 𝒲𝒲2
• Message set |𝒲𝒲1 | = 16: • Message set:
𝒲𝒲1 = {0000, 0001, 0010, … , 1111} 𝒲𝒲2 = {000, 001, 010, … , 111}
• Message split: • Message split:
𝑊𝑊𝑐𝑐,1 =(𝑎𝑎1 𝑎𝑎2 ), 𝑊𝑊𝑝𝑝,1 =(𝑎𝑎3 𝑎𝑎4 ) 𝑊𝑊𝑐𝑐,2 =(𝑏𝑏1 ), 𝑊𝑊𝑝𝑝,2 =(𝑏𝑏2 𝑏𝑏3 )

[1] B. Clerckx, et al. "Is NOMA Efficient in • Common message

Multi-Antenna Networks? A Critical Look
at Next Generation Multiple Access 𝑊𝑊𝑐𝑐 = (𝑊𝑊𝑐𝑐,1 𝑊𝑊𝑐𝑐,2 ) = (𝑎𝑎1 𝑎𝑎2 𝑏𝑏1 ) 18
Techniques." IEEE OJ-COMS 2021.
 K-user Downlink RSMA: Linear Precoding
1-layer RS for MISO [1,2]:
𝐱𝐱 = 𝐩𝐩𝑐𝑐 𝑠𝑠𝑐𝑐 + ∑𝐾𝐾
𝑘𝑘=1 𝐩𝐩𝑘𝑘 𝑠𝑠𝑘𝑘

• Only one SIC required at each receiver

• No user ordering/grouping at the transmitter
• MU-LP/SDMA subset of 1-layer RS
• NOMA not a subset of 1-layer RS (for 𝐾𝐾 > 2)
𝑦𝑦1
�
𝑊𝑊 1
User-1
𝑊𝑊𝑐𝑐,1

…
𝑥𝑥1 𝑔𝑔𝑐𝑐,𝑘𝑘
combiner
Message

𝑦𝑦𝑘𝑘 𝑠𝑠𝑐𝑐̂ Decode �

𝑊𝑊
…

𝑊𝑊𝑐𝑐,𝑘𝑘 𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐 𝑐𝑐

Linear precoder 𝑥𝑥2 ×
Schedule K users

Split
Message splitter

𝑊𝑊1
SIC
…

Encode � 𝑐𝑐, 𝑘𝑘
Encoder

𝑊𝑊𝑐𝑐,𝐾𝐾 𝑊𝑊
� 𝑘𝑘

Combine
+ 𝑊𝑊
…

MIMO
channel + - Channel Precode
𝑊𝑊𝑘𝑘 𝑊𝑊𝑝𝑝,1 𝑠𝑠1
…

𝑔𝑔𝑘𝑘
𝑠𝑠̂ 𝑘𝑘 � 𝑝𝑝, 𝑘𝑘
𝑊𝑊
× Decode
…

𝑊𝑊𝑝𝑝,𝑘𝑘 𝑠𝑠𝑘𝑘
…

𝑊𝑊𝐾𝐾 𝑥𝑥𝑀𝑀 User-k

…

𝑊𝑊𝑝𝑝,𝐾𝐾 𝑠𝑠𝐾𝐾 …
𝑦𝑦𝐾𝐾 �
Transmitter 𝑊𝑊
User-K 𝐾𝐾

[1] B. Clerckx, et al. , "Rate splitting for MIMO wireless networks: a promising PHY-layer strategy for LTE evolution," in IEEE Comm. Mag., 2016.
19
[2] H. Joudeh and B. Clerckx, "Sum-rate maximization for linearly precoded downlink multiuser MISO systems with partial CSIT: A rate-splitting
approach," in IEEE TCOM, 2016.
 K-user Downlink RSMA: Linear Precoding
1-layer RS for MIMO [1]:
𝐱𝐱 = 𝐏𝐏𝑐𝑐 𝐬𝐬𝑐𝑐 + ∑𝐾𝐾
𝑘𝑘=1 𝐏𝐏𝑘𝑘 𝐬𝐬𝑘𝑘

Vectors of messages and streams

[1] A. Mishra, Y. Mao, O. Dizdar, and B. Clerckx, “Rate-Splitting Multiple Access for Downlink Multiuser MIMO: Precoder Optimization and PHY-
Layer Design,” IEEE Trans. on Commun., vol. 70, no. 2, pp. 874-890, Feb. 2022. 20
K-user Downlink RSMA: Linear Precoding

Generalized RS framework for 3-user [1]:

Framework extendable to K-user

SDMA and NOMA subsets of RSMA
[1] Y. Mao, B. Clerckx, and V.O.K. Li, "Rate-splitting multiple access for downlink communication systems: bridging, generalizing,
21
and outperforming SDMA and NOMA," EURASIP Journal on Wireless Communications and Networking, 133 (2018).
K-user Downlink RSMA: Non-Linear Precoding

Dirty Paper Coded Rate-Splitting (DPC-RS) [1]:

CSI acquisition 𝑦𝑦1 �1

𝑊𝑊
User-1
𝑊𝑊𝑐𝑐,1

…
𝑥𝑥1 𝑔𝑔𝑘𝑘,𝑐𝑐
combiner

precoder
Message

Encoder
𝑦𝑦𝑘𝑘

Superposition Coding
…

� 𝑐𝑐

Linear
𝑊𝑊𝑐𝑐,𝑘𝑘 𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐 𝐩𝐩𝑐𝑐 𝑠𝑠𝑐𝑐 𝑠𝑠̂𝑐𝑐 Decode 𝑊𝑊
Schedule K users

×
Message splitter

𝑊𝑊1

Split
…
…

Combine
𝑊𝑊𝑐𝑐.𝐾𝐾 MIMO SIC Encode � 𝑐𝑐, 𝑘𝑘
𝑊𝑊
channel � 𝑘𝑘
…

+ 𝑊𝑊
𝑥𝑥𝑛𝑛 Precode
𝑊𝑊𝑘𝑘 𝑊𝑊𝑝𝑝,1 𝑠𝑠1 𝐩𝐩1 𝑠𝑠1 + - Channel
𝑔𝑔𝑘𝑘
� 𝑝𝑝, 𝑘𝑘
Precoder
𝑠𝑠̂ 𝑘𝑘 𝑊𝑊
Encoder
…

…
…

𝑊𝑊𝑝𝑝,𝑘𝑘 𝑠𝑠𝑘𝑘 𝐩𝐩𝑘𝑘 𝑠𝑠𝑘𝑘 × Decode

…

DPC

…
𝑊𝑊𝐾𝐾 User-k
𝑥𝑥𝑀𝑀
…

…
…

𝑊𝑊𝑝𝑝,𝐾𝐾 𝑠𝑠𝐾𝐾 𝐩𝐩𝐾𝐾 𝑠𝑠𝐾𝐾

…
𝑦𝑦𝐾𝐾 � 𝐾𝐾
𝑊𝑊
DPC
Transmitter User-K
CSI acquisition
CSI acquisition

Encode the private parts using DPC DPC-RS boils down to DPC if CSIT is perfect

[1] Y. Mao and B. Clerckx, "Beyond dirty paper coding for multi-antenna broadcast channel with partial CSIT: A rate-splitting 22
approach," in IEEE Transactions on Communications, vol. 68, no. 11, pp. 6775-6791, Nov. 2020.
Downlink RSMA

Relation among existing strategies and RSMA framework

Including MU-MIMO

[1] Y. Mao, B. Clerckx, and V.O.K. Li, "Rate-splitting multiple access for downlink communication systems: bridging, generalizing, and
outperforming SDMA and NOMA," EURASIP Journal on Wireless Communications and Networking, 133 (2018).
[2] Y. Mao, O. Dizdar, B. Clerckx, R. Schober, P. Popovski, H. V. Poor, “Rate-Splitting Multiple Access: Fundamentals, Survey, and
Future Research Trends,” arXiv:2201.03192 23
Uplink RSMA

Superposition coding

[1] B. Rimoldi and R. Urbanke, “A rate-splitting approach to

the Gaussian multiple-access channel,” IEEE Trans. Inf.
Theory, vol. 42, no. 2, pp. 364–375, Mar. 1996.
[2] Y. Mao, O. Dizdar, B. Clerckx, R. Schober, P. Popovski, H.
V. Poor, “Rate-Splitting Multiple Access: Fundamentals,
Survey, and Future Research Trends,” arXiv:2201.03192 24
DoF Enhancement of RSMA

SDMA
RSMA
DoF=2.5
DoF=2
DoF=1.5

DoF=1

To notice: NOMA worse DoF than

SDMA despite higher Rx complexity
2
CSIT error power: 𝔼𝔼 � 𝑘𝑘
𝒉𝒉 ~𝑃𝑃−𝛼𝛼
Slope (Sum-DoF) and rate increase! Cannot reach any better DoF!

RSMA achieves the optimal DoF region for both perfect and imperfect CSIT [1]
[1] E. Piovano and B. Clerckx, "Optimal DoF region of the K-user MISO BC with partial CSIT," in IEEE Communications Letters,
25
vol. 21, no. 11, pp. 2368-2371, Nov. 2017.
DoF vs. Number of SIC Layers
• Sum DoF/multiplexing gain 𝑑𝑑𝑠𝑠 :
• MMF DoF 𝑑𝑑mmf :
• 𝐾𝐾 = 6, 𝑀𝑀 = 4 [1]
∑𝑘𝑘 𝑅𝑅𝑘𝑘 ≈ 𝑑𝑑𝑠𝑠 log 2 𝑃𝑃 + 𝒪𝒪(1)
min 𝑅𝑅𝑘𝑘 ≈ 𝑑𝑑mmf log 2 𝑃𝑃 + 𝒪𝒪 1
𝑘𝑘

To notice: NOMA worse sum

DoF than SDMA despite higher
Rx complexity

1-layer RS achieves larger sum/MMF multiplexing gains than MU-LP and

NOMA with a single layer of SIC!
[1] B. Clerckx, Y. Mao, R. Schober, E. Jorswieck, D. J. Love, J. Yuan, L. Hanzo, G. Ye Li, E. G. Larsson, and G. Caire. "Is NOMA
Efficient in Multi-Antenna Networks? A Critical Look at Next Generation Multiple Access Techniques." IEEE OJ-COMS 2021. 26
DoF vs. Number of Users
• MMF DoF 𝑑𝑑mmf : min 𝑅𝑅𝑘𝑘 ≈ 𝑑𝑑mmf log 2 𝑃𝑃 + 𝒪𝒪 1
𝑘𝑘
• 𝑀𝑀 = 6 1 :

1-layer RS is significantly more efficient than NOMA since RS with only one-SIC
layer can support a larger number of users than NOMA with many SIC layers.
[1] B. Clerckx, Y. Mao, R. Schober, E. Jorswieck, D. J. Love, J. Yuan, L. Hanzo, G. Ye Li, E. G. Larsson, and G. Caire. "Is NOMA 27
Efficient in Multi-Antenna Networks? A Critical Look at Next Generation Multiple Access Techniques." IEEE OJ-COMS 2021.
Precoder Optimization
1-layer RS system model:
𝐱𝐱 = 𝐩𝐩𝑐𝑐 𝑠𝑠𝑐𝑐 + ∑𝐾𝐾
𝑘𝑘=1 𝐩𝐩𝑘𝑘 𝑠𝑠𝑘𝑘
• Precoding matrix: 𝐏𝐏 = 𝐩𝐩𝑐𝑐 , 𝐩𝐩1 , … , 𝐩𝐩𝐾𝐾 , 𝐏𝐏𝑝𝑝 = [𝐩𝐩1 , … , 𝐩𝐩𝐾𝐾 ]
• Average power constraint: 𝐩𝐩𝑐𝑐 2
+ ∑𝐾𝐾
𝑘𝑘=1 𝐩𝐩𝑘𝑘
2
≤ 𝑃𝑃
• Rate of user-k:
𝑅𝑅𝑘𝑘,𝑡𝑡𝑡𝑡𝑡𝑡 = 𝐶𝐶𝑘𝑘 + 𝑅𝑅𝑝𝑝,𝑘𝑘 , where ∑𝐾𝐾
𝑘𝑘=1 𝐶𝐶𝑘𝑘 = min 𝑅𝑅𝑐𝑐,1 , … , 𝑅𝑅𝑐𝑐,𝐾𝐾 .

• Rate of decoding the intended common and private streams at user-k:

2 2
𝐡𝐡𝐻𝐻
𝑘𝑘 𝐩𝐩𝑐𝑐 𝐡𝐡𝐻𝐻
𝑘𝑘 𝐩𝐩𝑘𝑘
𝑅𝑅𝑐𝑐,𝑘𝑘 = log 2 1 + 2 , 𝑅𝑅𝑝𝑝,𝑘𝑘 = log 2 1 + 2
∑𝐾𝐾 𝐻𝐻
𝑗𝑗=1 𝐡𝐡𝑘𝑘 𝐩𝐩𝑗𝑗 + 𝜎𝜎𝑘𝑘2 ∑𝐾𝐾 𝐻𝐻
𝑗𝑗=1,𝑗𝑗≠𝑘𝑘 𝐡𝐡𝑘𝑘 𝐩𝐩𝑗𝑗 + 𝜎𝜎𝑘𝑘2

𝑦𝑦1 �1
𝑊𝑊
User-1
𝑊𝑊𝑐𝑐,1

…
𝑥𝑥1
combiner

𝑔𝑔𝑐𝑐,𝑘𝑘
Message

𝑦𝑦𝑘𝑘 𝑠𝑠𝑐𝑐̂ �𝑐𝑐

Decode 𝑊𝑊
…

𝑊𝑊𝑐𝑐,𝑘𝑘 𝑊𝑊𝑐𝑐 𝑠𝑠𝑐𝑐

Linear precoder

𝑥𝑥2 ×
Schedule K users

Message splitter

Split
𝑊𝑊1
SIC
Encoder

Encode �𝑐𝑐, 𝑘𝑘
…

𝑊𝑊𝑐𝑐,𝐾𝐾 𝑊𝑊

Combine
+ � 𝑘𝑘
𝑊𝑊
…

MIMO Channel Precode

+-
𝑊𝑊𝑘𝑘 𝑊𝑊𝑝𝑝,1 𝑠𝑠1 channel
𝑔𝑔𝑘𝑘
…

𝑠𝑠̂ 𝑘𝑘 � 𝑝𝑝, 𝑘𝑘
𝑊𝑊
× Decode
…

𝑊𝑊𝑝𝑝,𝑘𝑘 𝑠𝑠𝑘𝑘
…

𝑊𝑊𝐾𝐾 𝑥𝑥𝑀𝑀 User-k

…

𝑊𝑊𝑝𝑝,𝐾𝐾 𝑠𝑠𝐾𝐾
…

Transmitter 𝑦𝑦𝐾𝐾 �𝐾𝐾

𝑊𝑊
User-K
28
Perfect CSIT: Weighted Sum Rate Maximization

1-layer RS

max ∑𝐾𝐾
𝑘𝑘=1 𝑢𝑢𝑘𝑘 𝐶𝐶𝑘𝑘 + 𝑅𝑅𝑝𝑝,𝑘𝑘 𝑢𝑢𝑘𝑘 = 1: Sum Rate
𝐏𝐏, 𝐜𝐜

s.t. ∑𝐾𝐾
𝑘𝑘=1 𝐶𝐶𝑘𝑘 ≤ min 𝑅𝑅𝑐𝑐,1 , … , 𝑅𝑅𝑐𝑐,𝐾𝐾 Common Msg Rate Allocation
(to be decoded by all)
𝐩𝐩𝑐𝑐 2
+ ∑𝐾𝐾
𝑘𝑘=1 𝐩𝐩𝑘𝑘
2
≤ 𝑃𝑃 Transmit Power Constraint
𝐶𝐶𝑘𝑘 + 𝑅𝑅𝑝𝑝,𝑘𝑘 ≥ 𝑅𝑅𝑘𝑘𝑡𝑡𝑡 , ∀𝑘𝑘 QoS Rate Constraint
𝐶𝐶𝑘𝑘 ≥ 0, ∀𝑘𝑘,

where 𝐜𝐜 = [𝐶𝐶1 , … , 𝐶𝐶𝐾𝐾 ],

2
𝐡𝐡𝐻𝐻
𝑘𝑘 𝐩𝐩𝑐𝑐
𝑅𝑅𝑐𝑐,𝑘𝑘 = log 2 1 + 2 ,
∑𝐾𝐾 𝐻𝐻
𝑗𝑗=1 𝐡𝐡𝑘𝑘 𝐩𝐩𝑗𝑗 + 𝜎𝜎𝑘𝑘2
Extension possible to imperfect CSIT (bounded, unbounded)
2
𝐡𝐡𝐻𝐻 and other metrics (max min rate, energy efficiency, etc)
𝑘𝑘 𝐩𝐩𝑘𝑘
𝑅𝑅𝑝𝑝,𝑘𝑘 = log 2 1 + 2
∑𝐾𝐾 𝐻𝐻
𝑗𝑗=1,𝑗𝑗≠𝑘𝑘 𝐡𝐡𝑘𝑘 𝐩𝐩𝑗𝑗 + 𝜎𝜎𝑘𝑘2

[1] H. Joudeh and B. Clerckx, "Sum-rate maximization for linearly precoded downlink multiuser MISO systems with partial CSIT: A
rate-splitting approach," in IEEE Transactions on Communications, vol. 64, no. 11, pp. 4847-4861, Nov. 2016.
[2] Y. Mao, B. Clerckx, and V.O.K. Li, "Rate-splitting multiple access for downlink communication systems: bridging, generalizing,29
and outperforming SDMA and NOMA," EURASIP Journal on Wireless Communications and Networking, 133 (2018).
 Perfect CSIT: Weighted Sum Rate Maximization

Two-user ergodic rate region:𝑀𝑀 = 𝐾𝐾 = 2, SNR = 20 dB, 10 dB average channel gain gap

RSMA bridges and outperforms SC-SIC (NOMA) and MU-LP (SDMA).

RSMA achieves a rate region closer to the capacity region.
[1] Y. Mao, B. Clerckx, and V.O.K. Li, "Rate-splitting multiple access for downlink communication systems: bridging, generalizing, 30
and outperforming SDMA and NOMA," EURASIP Journal on Wireless Communications and Networking, 133 (2018).
 Perfect CSIT: WSR with QoS constraints

10-user weighted sum rate with QoS with 𝑀𝑀 = 2 and perfect CSIT:

1 SIC

To notice:
9 SIC Minor gain of
0 SIC NOMA over
SDMA at huge
Rx cost

Huge gains with RS (1 SIC layer) vs. NOMA (9 SIC layers!)

[1] Y. Mao, B. Clerckx, and V.O.K. Li, "Rate-splitting multiple access for downlink communication systems: bridging, generalizing, 31
and outperforming SDMA and NOMA," EURASIP Journal on Wireless Communications and Networking, 133 (2018).
Imperfect CSIT: Ergodic Rate Region Enhancement

Two-user ergodic rate regions with 𝑀𝑀 = 𝐾𝐾 = 2 , imperfect CSIT (𝛼𝛼 = 0.6), SNR = 20 dB:

To notice: NOMA
worse than SDMA
despite higher Rx
complexity

RS schemes outperform conventional NoRS, i.e.,

SDMA/MU-LP/MU-MIMO, NOMA and DPC!
[1] Y. Mao and B. Clerckx, "Beyond dirty paper coding for multi-antenna broadcast channel with partial CSIT: A rate-splitting
32
approach," in IEEE Transactions on Communications, vol. 68, no. 11, pp. 6775-6791, Nov. 2020.
(Im)perfect CSIT: Max-Min Fairness

Optimization results: 𝑀𝑀 = 4, 𝐾𝐾 = 6, SNR = 20 dB, 10dB path loss difference.

Perfect CSIT Imperfect CSIT

1 SIC

5 SIC
1 SIC

0 SIC

Huge gains with 1-layer RS: rate, fairness and robustness

enhancements with only 1 SIC!
[1] B. Clerckx, Y. Mao, R. Schober, E. Jorswieck, D. J. Love, J. Yuan, L. Hanzo, G. Ye Li, E. G. Larsson, and G. Caire. "Is NOMA
33
Efficient in Multi-Antenna Networks? A Critical Look at Next Generation Multiple Access Techniques." IEEE OJ-COMS 2021.
Imperfect CSIT: Robust Max-Min Fairness

Robust max-min fairness: 𝑀𝑀 = 𝐾𝐾 = 3 , imperfect CSIT

SDMA
RSMA

NoRS saturates due to non-scaling CSIT errors.

RS avoids saturation and performs better across all SNRs.

[1] H. Joudeh and B. Clerckx, "Robust transmission in downlink multiuser MISO systems: A rate-splitting approach," in IEEE 34
Transactions on Signal Processing, vol. 64, no. 23, pp. 6227-6242, 1 Dec.1, 2016.
 RSMA
Multiplex users in spatial and power domains using (non)linearly precoded Rate-Splitting
(RS) with SIC
Properties:
• Partially decode interference and partially treat interference as noise
• Bridge the extremes of NOMA and SDMA (more general and powerful)
Pros:
1. Encompass SDMA and NOMA as special cases
2. RSMA rate ≥ SDMA and NOMA rates
3. Optimal from a DoF perspective in both perfect and imperfect CSIT [1-4]
4. Cope with any user deployments (diversity of channel strengths and
directions), CSIT inaccuracy and network load
5. Lower computational complexity than NOMA for both the transmit
scheduler and the receivers
Cons:
1. Higher encoding complexity than SDMA and NOMA (e.g. K -> K+1 streams)
[3] A. G. Davoodi and S. A. Jafar, “Aligned image sets under channel
[1] E. Piovano and B. Clerckx, “Optimal DoF region of the K-user MISO uncertainty: Settling conjectures on the collapse of degrees of freedom
BC with partial CSIT,” IEEE Commun. Lett., vol. 21, no. 11, pp. 2368– under finite precision CSIT,” IEEE Trans. Inf. Theory, vol. 62, Oct. 2016.
2371, Nov. 2017 [4] A. G. Davoodi and S. Jafar, “Degrees of freedom region of the (M,
[2] C. Hao, B. Rassouli, and B. Clerckx, “Achievable DoF regions of N1, N2) MIMO broadcast channel with partial CSIT: An application of
MIMO networks with imperfect CSIT,” IEEE Trans. Inf. Theory, vol. 63, sum-set inequalities based on aligned image sets,” IEEE Trans.35
Inf.
no. 10, pp. 6587–6606, Oct. 2017. Theory, vol. 66, no. 10, pp. 6256–6279, Oct. 2020
… Fundamentally Robust

Conventional approach New approach

SDMA/MU-LP: 𝐱𝐱 = ∑𝐾𝐾
𝑘𝑘=1 𝐩𝐩𝑘𝑘 s𝑘𝑘 RSMA: 𝐱𝐱 = 𝐩𝐩𝑐𝑐 s𝑐𝑐 + ∑𝐾𝐾
𝑘𝑘=1 𝐩𝐩𝑘𝑘 s𝑘𝑘

BUT perfect CSIT to start with! Motivated by DoF optimality in imperfect CSIT!
4G/5G

This is why RSMA is fundamentally robust 36

Outline

1 (Existing) Multiple Access Techniques

2 Rate Splitting Multiple Access (RSMA)

3 Applications/Interplay of RSMA

4 Physical Layer Design

5 Conclusions

37
Applications/Interplay of RSMA in 6G
 MU-MIMO with Statistical CSIT
 MU-MIMO with Quantized Feedback
[1] A. Mishra, Y. Mao, O. Dizdar, and B. Clerckx,
 FDD Massive MIMO “Rate-Splitting Multiple Access for 6G – Part I:
 TDD Massive MIMO Principles, Applications and Future Works,”
arXiv:2205.02548
 Cell-free Massive MIMO
 Multiuser Millimeter Wave and THz Systems [2] L. Yin, Y. Mao, O. Dizdar, and B. Clerckx,
“Rate-Splitting Multiple Access for 6G – Part II:
 Cooperative Transmission with User Relaying Interplay with Integrated Sensing and
 Physical Layer Security Communications,” arXiv:2205.02462
 Reconfigurable Intelligent Surface (RIS) [3] H. Li, Y. Mao, O. Dizdar, and B. Clerckx,
 Multi-Cell Networks “Rate-Splitting Multiple Access for 6G – Part III:
Interplay with Reconfigurable Intelligent
 Cloud-Radio Access Network (C-RAN) Surfaces,” arXiv:2205.02036
 Overloaded Cellular IoT and massive access
 Joint Communication and Jamming
 Mixed Services: Non-Orthogonal Unicast and Multicast Transmission
 Multigroup Multicast
 Multibeam Satellite Communications
In all those areas,
 Low latency and finite block length
RSMA > SDMA, NOMA, OMA
 Integrated Sensing and Communications
 Machine learning for RSMA and RSMA-aided learning
 Grant-free access
 Network slicing 38
 Cognitive radio, …
FDD Massive MIMO
𝑀𝑀 = 100, 𝐾𝐾 = 12, SNR = 30 dB, imperfect CSIT(𝜏𝜏 2 = 0.4) :

Problems:
• overlap between the eigen-subspaces
• imperfect CSIT

HRS specifically designed for FDD massive MIMO to cope with imperfect
CSIT and overlap between eigen-subspaces [1].

[1] M. Dai, B. Clerckx, D. Gesbert and G. Caire, "A rate splitting strategy for massive MIMO with imperfect CSIT," in IEEE
Transactions on Wireless Communications, vol. 15, no. 7, pp. 4611-4624, July 2016. 39
TDD Massive MIMO: Mobility and Latency
Imperfect CSIT Model [1]:
𝐡𝐡𝑘𝑘 [𝑚𝑚] = 𝜖𝜖 2 𝐡𝐡𝑘𝑘 [𝑚𝑚 − 1] + 1 − 𝜖𝜖 2 𝒆𝒆𝑘𝑘 [𝑚𝑚]
• Link-level simulation with OFDM waveform and 3GPP channel model (𝑀𝑀 =
32, 𝐾𝐾 = 8, 10ms feedback delay):

QoS rate constraint (6 bps/Hz)

7 km/h for SDMA

40 km/h for RSMA With the same QoS
rate constraint, RSMA
supports a much
higher user speed.

RSMA maintains multiuser connectivity in mobility conditions.

[1] O. Dizdar, Y. Mao, and B. Clerckx. "Rate-Splitting Multiple Access to Mitigate the Curse of Mobility in (Massive) MIMO 40
Networks," IEEE Trans. on Commun., Oct 2021.
Cooperative Transmission with User Relaying
Cooperative Rate-Splitting: one user forwards its decoded common message
to another user [1, 2].

CRS outperforms 1-layer RS, MU-LP, NOMA, Cooperative NOMA even in the lower SNR regime!

[1] J. Zhang, B. Clerckx, J. Ge and Y. Mao, "Cooperative Rate Splitting for MISO Broadcast Channel With User Relaying, and Performance
Benefits Over Cooperative NOMA," in IEEE Signal Processing Letters, vol. 26, no. 11, pp. 1678-1682, Nov. 2019.
[2] Y. Mao, B. Clerckx, J. Zhang, V. O. K. Li and M. A. Arafah, "Max-min fairness of K-user cooperative rate-splitting in MISO broadcast channel
41
with user relaying," in IEEE Transactions on Wireless Communications, vol. 19, no. 10, pp. 6362-6376, Oct. 2020.
Cloud-Radio Access Network (C-RAN)
Data Compression [1] Data Sharing [2]

For a given fronthaul capacity constraint, RSMA schemes are more

spectrally and energy efficient than SDMA and NOMA.
[1] D. Yu, J. Kim and S. Park, "An efficient rate-splitting multiple access scheme for the downlink of C-RAN systems," in IEEE Wireless
Communications Letters, vol. 8, no. 6, pp. 1555-1558, Dec. 2019.
[2] A. Alameer Ahmad, H. Dahrouj, A. Chaaban, A. Sezgin and M. Alouini, "Interference mitigation via rate-splitting and common message42
decoding in cloud radio access networks," in IEEE Access, vol. 7, pp. 80350-80365, 2019.
Reconfigurable Intelligent Surfaces
M=K=2, 32 RIS elements
Multi-RIS aided RSMA

Advantage of RIS-aided RSMA:

1) higher spectral efficiency,
2) coverage extension and beam control flexibility,
3) robust to CSI imperfection and user mobility,
4) lower computational and hardware complexity
[1] H. Li, Y. Mao, O. Dizdar, and B. Clerckx, “Rate-Splitting Multiple Access for 6G – Part III: Interplay with Reconfigurable Intelligent
Surfaces,” arXiv:2205.02036
43
[2] A. Bansal, K. Singh, B. Clerckx, C.-P. Li, and M.-S. Alouini, “Rate-Splitting Multiple Access for Intelligent Reflecting Surface aided Multi-
User Communications”
Non-Orthogonal Unicast and Multicast Transmission

Unicast: one-to-one vs. Multicast: one-to-many

• Conventional approach: allocate orthogonal radio resources to unicast and multicast

services.

issue: inefficient bandwidth and transmit power utilization/allocation

• Solution: Non-Orthogonal Unicast and Multicast (NOUM) by superimposing in the power

domain

issue: interference between multicast and unicast and among the unicast messages

Applications of NOUM: Multicast message

Unicast messages
• B5G: scarcity of radio resources and heterogeneity of
applications

• Layered Division Multiplexing (LDM) in digital

television systems
44
Non-Orthogonal Unicast and Multicast Transmission
Transmit: one multicast message 𝑊𝑊0 intended for all users
+ 𝐾𝐾 unicast messages 𝑊𝑊1 , … , 𝑊𝑊𝐾𝐾 intended for different users

Conventional MU-LP-assisted NOUM Proposed 1-layer RS-assisted NOUM [1]

With RS, SIC in NOUM is efficiently exploited for the dual purpose:
• Separate the unicast and multicast streams
• Better manage interference among unicast streams

[1] Y. Mao, B. Clerckx and V. O. K. Li, "Rate-splitting for multi-antenna non-orthogonal unicast and multicast transmission:
Spectral and energy efficiency analysis," in IEEE Transactions on Communications, vol. 67, no. 12, pp. 8754-8770, Dec. 2019. 45
Multibeam Satellite Communications
Follow PHY layer multigroup multicast transmission.
Differences in multibeam satellite:
• Per-feed power constraints • 𝑀𝑀 = 7, 𝐾𝐾 = 14, 𝐺𝐺 = 7, |𝒢𝒢𝑔𝑔 | = 2:
• Satellite channel model with
imperfect CSIT
• Overloaded

Superiority of RS: manage inter-beam interference, CSIT uncertainty, practical per-

feed constraints and overloaded regime.
[1] L. Yin and B. Clerckx, "Rate-Splitting Multiple Access for Multigroup Multicast and Multibeam Satellite Systems," in IEEE
Transactions on Communications, vol. 69, no. 2, pp. 976-990, Feb. 2021.
[2] Z. W. Si, L. Yin, and B. Clerckx, “Rate-Splitting Multiple Access for Multigateway Multibeam Satellite Systems with Feeder 46
Link Interference,” IEEE Trans. on Commun., vol. 70, no.3, pp. 2147-2162, Mar. 2022.
Short-Packet and Low-Latency

WSR with QoS constraints: 𝑀𝑀 = 2, 𝐾𝐾 = 4 , perfect CSIT, 10 dB path loss difference

SDMA
RSMA

RSMA can rely on shorther blocklengths → lower latency [1]

[1] Y. Xu, Y. Mao, O. Dizdar, and B. Clerckx, “Rate-splitting multiple access with finite blocklength for short-packet and low- 47
latency downlink communications,” arXiv:2105.06198.
Integrated Sensing and Communications
How to best use the spectrum for the dual purpose of radar and communication?
Find the strategy that achieves the best trade-off between Rate and MSE/CRB [1-2].
DOA/DOD Radar cross section

Doppler

Better communication-
sensing trade-off achieved
by RSMA

[1] C. Xu, B. Clerckx, S. Chen, Y. Mao, and J. Zhang, “Rate-Splitting Multiple Access for Multi-Antenna Joint Radar and
Communications,” IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 6, pp. 1332 – 1347, Nov. 2021.
[2] L. Yin, Y. Mao, O. Dizdar, and B. Clerckx, “Rate-Splitting Multiple Access for 6G – Part II: Interplay with Integrated Sensing and
48
Communications,” arXiv:2205.02462
Outline

1 (Existing) Multiple Access Techniques

2 Rate Splitting Multiple Access (RSMA)

3 Applications/Interplay of RSMA

4 Physical Layer Design

5 Conclusions

49
Physical Layer Design
Transmitter and receiver structure [1]:

Transmitter: Receiver:
• Message split • Successive Interference
• Channel coding (polar codes) Cancellation (SIC)
• Interleaver • Minimum Mean-Square Error
• Finite constellation modulation (4-QAM, (MMSE) equalization
16-QAM, 64-QAM and 256-QAM), • Soft Decision (SD) polar decoding
• Optimized precoding
• Adaptive Modulation and Coding (AMC)
[1] O. Dizdar, Y. Mao, W. Han and B. Clerckx, "Rate-splitting multiple access for downlink multi-antenna communications: 50
Physical layer design and link-level simulations," PIMRC, London, United Kingdom, 2020.
Link-Level Simulations
Throughput vs. SNR for RSMA, SDMA and NOMA, 𝑀𝑀 = 𝐾𝐾 = 2, 𝛼𝛼 = 0.6, No
QoS constraint:

The gain of RSMA over SDMA in LLS is even larger (relatively speaking) than that
in Shannon bound.
[1] O. Dizdar, Y. Mao, W. Han and B. Clerckx, "Rate-splitting multiple access for downlink multi-antenna communications: 51
Physical layer design and link-level simulations," PIMRC, London, United Kingdom, 2020.
Link-Level Simulations
Multigroup multicast communication [1]:

• K users grouped into G groups • Max-min fairness throughput versus SNR,

• One message for each group 𝑀𝑀 = 4, 𝐾𝐾 = 6, 𝛼𝛼 = 0.6, 3 groups (2 users
per group):

RSMA achieves significant LLS throughput gains over SDMA, which is

consistent with (or even larger than) its gain in Shannon bounds.
[1] L. Yin, O. Dizdar, and B. Clerckx. "Rate-Splitting Multiple Access for Multigroup Multicast Cellular and Satellite
Communications: PHY Layer Design and Link-Level Simulations." IEEE ICC 2021. 52
Link-Level Simulations

Throughput vs. SNR with 2.5ms CSI feedback delay, operating frequency of 3.5GHz:

[1] O. Dizdar, Y.
Mao, and B.
Clerckx. "Rate-
Splitting Multiple
Access to
Mitigate the
Curse of Mobility
(a) 𝑀𝑀 = 4, 𝐾𝐾 = 4, speed= 20 km/h (b) 𝑀𝑀 = 4, 𝐾𝐾 = 4, speed= 30 km/h in (Massive)
MIMO
Networks." IEEE
Trans. on
Commun., Oct
2021.

(c) 𝑀𝑀 = 8, 𝐾𝐾 = 4, speed= 30 km/h (d) 𝑀𝑀 = 32, 𝐾𝐾 = 8, speed= 30 km/h

RSMA is robust to the degrading effects of user mobility/latency

and significantly outperforms 5G NR SDMA. 53
Outline

1 (Existing) Multiple Access Techniques

2 Rate Splitting Multiple Access (RSMA)

3 Applications/Interplay of RSMA

4 Physical Layer Design

5 Conclusions

54
Conclusions

General Observations of RS/RSMA:

• Partially decode interference, partially treat interference as noise

• Robust interference management strategy

• Flexible non-orthogonal transmission strategy

• General and unified multiple access

• Fundamental changes to PHY and MAC layers

55
Conclusions

Performance benefits:
• Spectral and energy efficiency gains
• QoS and fairness enhancements
• Robust to imperfect CSIT (i.e. quantized feedback, pilot contamination,
channel estimation error, mobility, latency, subband feedback,...)
• Reduce feedback overhead
• Robust to hardware impairments (phase noise)
• Cope with any user distribution (disparity of channel directions and
strengths) and network load (underloaded, overloaded)
• Complexity reduction (compared to NOMA)
• Lower latency

56
Future Challenges
A gold mine of research problems for academia and industry [1]:

[1] Y. Mao and B. Clerckx, “5G and Beyond: Fundamentals and Standards.”, Springer Nature Switzerland AG, Switzerland
(chapter: “Multiple Access Techniques”) 57
Pathways to 6G

Pathways to 3GPP B5G/6G [1,2]:

• RS can leverage 5G study/work items

• Missing piece: message split at the transmitter

• Applications to 6G in enhanced eMBB, URLLC, mMTC, and new services (joint

sensing/radar and communications, integrated cellular and satellite communications)
[1] O. Dizdar, Y. Mao, W. Han and B. Clerckx, "Rate-Splitting Multiple Access: A New Frontier for the PHY Layer of 6G," 91st
Vehicular Technology Conference (VTC)-Spring, 2020.
[2] O. Dizdar, Y. Mao, Y. Xu, P. Zhu, and B. Clerckx, “Rate-Splitting Multiple Access for Enhanced URLLC and eMBB in 6G,” IEEE 58
ISWCS 2021.
IEEE ComSoc Special Interest Group on RSMA

Link: https://sites.google.com/view/ieee-comsoc-wtc-sig-rsma/home

59
IEEE Special Issues

https://www.comsoc.org/publications/journals/ieee-jsac/cfp/rate-splitting-future-
wireless-networks

https://www.comsoc.org/publications/journals/ieee-ojcoms/cfp/rate-splitting-and-
multiple-access-techniques-6g 60