Full Text

Title
Design and analysis of detection algorithms for MIMO wireless

communication systems
Advisor(s)
Yuk, TTI; Cheung, SW
Author(s)
Shao, Ziyun.; .
Citation
Issued Date
URL
Rights
2011
http://hdl.handle.net/10722/174460
The author retains all proprietary rights, (such as patent rights)

and the right to use in future works.
Design and Analysis of Detection Algorithms for

MIMO Wireless Communication Systems
by
SHAO, Ziyun
B.Sc.(Eng), M.Sc.(Telecom)
A thesis submitted in partial fulfillment

of the requirements for the degree of
Doctor of Philosophy
at the
Department of Electrical and Electronic Engineering
The University of Hong Kong
in
October 2011
DECLARATION
I hereby declare that this thesis represents my own work, except where due
acknowledge is made, and that it has not been previously included in a thesis,
dissertation or report submitted to this University or to any other institution for a
degree, diploma or other qualifications.
Signed
SHAO, Ziyun
October 2011
To my beloved parents and husband
Abstract of thesis entitled

Design and Analysis of Detection Algorithms for MIMO Wireless
Communication Systems
Submitted by
SHAO Ziyun
for the degree of Doctor of Philosophy
at The University of Hong Kong
in October 2011
The increasing demand for high-mobility and high data rate in wireless
communications results in constraints and problems in the limited radio spectrum,
multipath fading, and delay spread.
The multiple-input multiple-output (MIMO) system has been generally
considered as one of the key technologies for the next generation wireless
communication systems. MIMO systems which utilize multiple antennas in both
the transmit side and the receive side can overcome the abovementioned
challenges since they are able to increase the channel capacity and the spectrum
usage efficiency without the need for additional channel bandwidth.
The detection algorithm is a big bottleneck in MIMO systems. Generally, it is
expected to fulfill two main goals simultaneously: low computational complexity
and good error rate performance. However, the existing detection algorithms are
either too complicated or suffering from very bad error-rate performance.
The purpose of this thesis is to comprehensively investigate the detection
algorithms of MIMO systems, and based on that, to develop new methods which
can reduce the computational complexity while retain good system performance.
Firstly, the background and the principle of MIMO systems and the previous work
on the MIMO decoding algorithms conducted by other researchers are thoroughly
reviewed. Secondly, the geometrical analysis of the signal detection is
investigated, and a geometric decoding algorithm which can offer the optimum
BLER performance is proposed. Thirdly, the semidefinite relaxation (SDR)
detection algorithms are extended to high-order modulation MIMO systems, and a
novel SDR detector for 256-QAM constellations is proposed. The theoretical
analysis on the tightness and the complexity are conducted. It demonstrates that
the proposed SDR detector can offer better BLER performance, while its
complexity is in between those of its two counterparts. Fourthly, we combine the
SDR detection algorithms with the sphere decoding. This is helpful for reducing
the computational complexity of the traditional sphere decoding since shorter
initial radius of the hyper sphere can be obtained. Finally, the novel
lattice-reduction-aided SDR detectors are proposed. They can provide
near-optimum error rate performance and achieve the full diversity gain with very
little computational complexity added compared with the stand-alone SDR
detectors.
Total words: 343
II
ACKNOWLEDGEMENT
First of all, I would like to express my sincere gratitude to my supervisors, Dr.
S. W. Cheung and Dr. T. I. Yuk for their precious guidance and persistent
encouragement throughout my entire PhD study. They taught me academic
knowledge and research skills and enlightened my passion to explore the
unknown scientific world. The thesis would not have been completed without
their supports.
I would also like to thank Mr. Eric W.L. Ng, Ms. Julie Hung and Ms. Lily Lo
in the Department of Electrical and Electronic Engineering for their kind help
during the past few years.
I truly appreciate the friendship of all teammates and friends in HKU for their
kind help, advice, guidance and encouragement, most notably Dr. Z. Zhang, Dr.
M. X. Xiao, Dr. F. Mai, Dr. X. G. Dai, Dr. W. Zhou, Dr. Z. Kong, Dr. K. C.
Leung, Mr. Y. F. Weng, Mr. Y. Y. Sun, Miss M. J. Mao, Miss L. Li, Mr. H. L.
Xiahou, and Mr. Z. B. Ni. In particular, many thanks to Dr. Dai who always had
taken seriously every question I asked him.
Finally, I am most grateful to my parents and my husband. Their selfless love,
continuous supports and encouragements throughout all these years are the most
precious thing to me. Without these, I could never get my work done well.
III
CONTENTS
ABSTRACT
ACKNOWLEDGMENTS
III
CONTENTS
IV
CHAPTER 1
INTRODUCTION
1.1
MIMO Wireless Communication System...........................2
1.2
Detection Problem for MIMO Wireless Communication
Systems..........................................................................................8
1.3
Literature Review................................................................9
1.4
Motivation and Contribution of the Thesis .......................14
1.5
Thesis Outline ...................................................................15
CHAPTER 2
STATE-OF-THE-ART MIMO DETECTION
ALGORITHMS
2.1
Introduction .......................................................................17
2.2
Linear Decoders ................................................................17
2.3
Sphere Decoding ...............................................................19
2.4
Successive Interference Cancellation ...............................25
2.5
Lattice-Reduction Aided Detection...................................27
2.6
Summary ...........................................................................34
CHAPTER 3
GEOMETRIC DETECTION
ALGORITHMS
IV
3.1
Introduction .......................................................................35
3.2
Geometrical Analysis of Signal Decoding for MIMO
Channels ......................................................................................35
3.3
Ellipsoid-searching decoding algorithm ...........................40
3.4
Simulation Results ............................................................48
3.5
Summary ...........................................................................52
CHAPTER 4
MIMO DETECTION ALGORITHMS

BASED ON SEMIDEFINITE
RELAXATION
4.1
Introduction .......................................................................54
4.2
Convex Optimization Problems ........................................54
4.3
Semidefinite Relaxation....................................................59
4.4
Semidefinite
Relaxation
Detection
Algorithms
for
Low-Order Modulation Systems.................................................62

4.5
Semidefinite
Relaxation
Detection
Algorithms
for
High-Order Modulation Systems ................................................74

4.6
SDR-initiated Sphere detector ..........................................89
4.7
Summary ...........................................................................93
CHAPTER 5
LATTICE-REDUCTION-AIDED
SEMIDEFINITE RELAXATION
DETECTION ALGORITHMS
V
5.1
Introduction .......................................................................94
5.2
Lattice Reduction ..............................................................96
5.3
Lattice-Reduction-Aided SDR Detection .......................101
5.4
Simulation Results .......................................................... 110
5.5
Discussion ....................................................................... 116
CHAPTER 6
CONCLUSIONS AND
RECOMMENDATIONS
6.1
Conclusions ..................................................................... 118
6.2
Recommendations ........................................................... 119
LIST OF FIGURES
121
LIST OF TABLES
124
ABBREVIATIONS
125
REFERENCES
127
PUBLICATIONS
138
VI
Chapter 1 Introduction
CHAPTER 1
INTRODUCTION
Wireless communications is the most vital field in digital communications.

The first wireless telegraph was developed by an Italian scientist Marconi, who
used radio waves to transmit telegraph messages without connecting wires over
the Bristol Channel in 1897. But the wireless communications did not provide
mobile services for people until 1960s when the AT&T Bell Labs proposed and
developed the Cellular Radio Network. In the recent two decades, the demand for
higher data rates and better quality has been increasing constantly with the
development
of
the
wireless
communications
technologies.
From
the
second-generation (2G) mobile communications services which provide up to 115

Kbit/s data rates to the third-generation (3G) mobile communications services that
are able to provide peak data rates at 56 Mbit/s, people are expecting that the
speed of the fourth-generation (4G) mobile communications can reach up to
1Gbit/s.
However, the property of high-mobility and high data rate of the wireless
communications systems result in several challenges such as limited radio
spectrum, multipath fading, delay spread and so on. The multiple antenna systems
with multiple antennas in both the transmitter and receiver sides can overcome
these challenges. It can increase the channel capacity and spectrum usage
efficiency without the need of additional channel bandwidth. Such kind of system
is the so-called multiple-input multiple-output (MIMO) wireless communication
system.
The MIMO technology has gone through a long history. It was firstly
proposed for application in wireless communication systems in the 1970s. In 1993,
Indian scientists Paulraj and Kailath introduced the idea of using spatial
multiplexing (SM) in MIMO system [1]. Since 1990s, the researchers in AT&T
Bell Lab have given a huge boost on MIMO technology. In 1995, Telatar showed
that the capacity of the MIMO systems in the fading channel conditions increases
linearly with the number of the transmit antennas and the receive antennas [2]. In
1996, Foschini proposed a diagonally-bell laboratories layered space-time
(D-BLAST) architecture for MIMO systems [3]. In 1998, Golden and other
researchers [4] built the laboratorial platform of MIMO system by using
vertical-bell laboratories layered space-time (V-BLAST) algorithm, where the
spectral efficiencies could reach 20-40bit/s/Hz at the indoor fading rates.
Up till now, MIMO technology has been widely considered as one of the key
technologies of the next generation wireless communication systems [5]. Some
mobile communications standards such as the 3G syetems, long-term-evolution
(LTE) and 4G have included the MIMO technology. The standard of wireless local
area network (WLAN) 802.11n recommends MIMO combined with orthogonal
frequency-division multiplexing (MIMO-OFDM). In many other wireless
communication research fields, such as ultra-wide-band (UWB) system and
cognitive radio (CR), researchers are considering to take MIMO technology into
consideration. Therefore, MIMO system is a promising solution to future wireless
communications and has become a very hot issue in both the academic and the
industrial fields.
1.1 MIMO Wireless Communication System

1.1.1 MIMO System Structure
The general structure of MIMO system can be illustrated in Fig.1.1. It
consists of multiple transmit antennas and multiple receive antennas. With the
space-time coding (STC) [82-83], correlated data is transmitted by different
antennas and the information redundancy can improve the system error rate
performance. This type of MIMO system is the MIMO diversity system which
focuses on the reliability. There is another type of MIMO system called MIMO
multiplexing system, in which, different data streams are transmitted by different
antennas simultaneously so as to increase the transmission data rate.
Figure 1.1 Structure of MIMO system.
1.1.2 Space-Time Coding for MIMO Systems

In MIMO systems, space-time coding is an important method to improve the
spatial diversity and reliability. There are two main kinds of the space-time codes:
the space-time trellis code (STTC) and the space-time block code (STBC) [5]. The
STBC is more popular than the STTC since it has a simpler structure and can offer
better performance. In 1998, Alamouti proposed a kind of STBC called Alamouti

code [6], which is able to provide full diversity gain with low complexity. Fig 1.2
illustrates an example for a 2 2 antennas system, the transmitted signal X is
given by:
x
X= 1
x2
x2*
x1*
(1.1)
The column vectors of the matrix are orthogonal to each other, and sent by
different antennas during each time slot. At the receiver, the received signal is
separated by linear transformation and then decoded by maximum likelihood
decoding.
Figure 1.2 MIMO space-time block code system.
1.1.3 Spatial Multiplexing for MIMO Systems
In MIMO systems, different antennas transmitting the signal in parallel can

offer the multiplexing gain. This multiplexing gain is called spatial multiplexing,
which is unique in MIMO system. Spatial multiplexing can offer a linear increase
in data rate with the number of the transmit antennas and the receive antennas
4
without in need of additional bandwidth and power. In the SM-MIMO system, the
data stream is split into several sub-streams, then modulated and transmitted by
different antennas simultaneously. With no doubt, this process could result in a
gain in data rate. The D-BLAST [3] is the first structure using MIMO
multiplexing. The data streams are multiplexed diagonally and transmitted in the
period of transmitting a block of signal. Then another V-BLAST [4, 7] structure is
proposed, which is more effective. The data streams are transmitted in parallel,
that is, the i th data symbol is transmitted by the i th antenna directly. The
BLAST structure can provide high multiplexing gain, and the capacity of such
system is increased linearly with the number of the transmit antennas and the
receive antennas [2]. In this thesis, we will focus on the SM-MIMO system.
1.1.4 MIMO System Model
The Spatial multiplexing MIMO system is illustrated in Fig.1.3, in which,

NT transmit antennas send the signal vectors to N R receive antennas over a
wireless channel. At the transmit side, the user data stream is partitioned into NT
sub-streams and then sent by different transmit antennas. At the receive side, each
receive antenna receives signal vectors from all the transmit antennas.
Figure 1.3 MIMO spatial multiplexing system.
Thus, the MIMO system is modeled as [5]:

r1 h11
r
2 = h21

rN R hN R 1
h12
h22
hN R 2
h1NT x1 n1

h2 NT x2 n2
+

hN R NT xNT nNT
(1.2)
or
r = Hx + n
(1.3)
where r is the N R -dimensional received signal vector, and x is the

NT -dimensional signal vector in the transmit lattice. H denotes the channel
matrix, with elements hij representing the transfer function from the j-th transmit
antenna to i-th receive antenna. n is the N R -dimensional additive noise. In this
thesis, the signal vector x is assumed to be a statistically independent variable
with zero mean and unit variance x 2 = 1 . Perfect channel knowledge is assumed
to be known to the receiver. In addition, the channel matrix H is assumed to be a
flat fading channel and all the entries in H are complex Gaussian and
independent. The noise is an independently and identically distributed (i.i.d.)
zero-mean Gaussian noise vector with elements having a fixed variance n 2 .

The complex transmission in (1.3) can be equivalently represented in real
matrix form as:
Re ( r ) Re ( H ) -Im ( H ) Re ( x ) Re ( n )
=
+
Im ( r ) Im ( H ) Re ( H ) Im ( x ) Im ( n )
(1.4)
with Re() and Im() being the real and imaginary parts of (), respectively. The
real-valued representation is written as [5]:
r = Hx + n
(1.5)
The dimension of r and x are M R = 2 N R and M T = 2 NT , respectively. H

becomes an M R M T matrix. And the noise n is an M R - dimensional vector.
1.1.5 Channel Capacity

The channel capacity represents the maximum possible data rate that can be
reliably transmitted. In other words, the possible error rate should be acceptably
small [84-86]. In a flat fading channel, the channel capacity can be written as

Q
C = log 2 det I +
NT

(1.6)
HH H , N R < NT
Q= H
H H , NT < N R
(1.7)
where det ( i ) denotes the determinant of a matrix, is an identity matrix,

is the average received signal to noise ratio (SNR), H H is the complex conjugate
transpose of H . Generally, the capacity is roughly linear with the number of the
antennas.
1.2 Detection Problem for MIMO Wireless Communication

Systems
1.2.1 MIMO Detection
Detection is the reverse process of the signal transmission. As mentioned
before, multiple transmit antennas send different signal symbols simultaneously,
the received signal vector is a superposition of all transmitted signal distorted by
the channel matrix and corrupted by the additive Gaussian noise. Although we
assume that the channel information is known to the receiver, the noise is random
and unknown. Thus, detection is aimed at finding the transmitted signal vector
based on the channel matrix and the received signal vector. Generally, the
detection algorithm needs to fulfill two main goals. One is low computational
complexity, and the other is good error rate performance. Unfortunately, these two
goals usually contradict each other: low computational complexity means bad
error rate performance, while good error rate performance results in high
computational complexity. In practice, a tradeoff has to be made between them.
1.2.2 Maximum Likelihood Decoding

The maximum likelihood (ML) decoding [74-75] is the optimum decoding
algorithm as it can provide the best error rate performance. It is formulated as:
x ML = arg min r Hx
(1.8)
It implies that ML calculates the Euclidean distance between the possible transmit
signal vectors and the received signal vector, and then choose the one which is
closest to the received vector as the solution. However, since all the possible
signal vector in the lattice space should be considered, and their Euclidean
distances away from the received vector have to be calculated, the complexity of
ML decoding increases exponentially with the number of transmit antennas and
polynomially with the size of the constellation [5]. Thus, in an SM-MIMO system,
the complexity of ML decoding is non-deterministic polynomial time hard
(NP-hard). This disadvantage makes the system impractical to be implemented.
Thus, detection has become one of the major challenges in MIMO system.
The main objective of designing a detection algorithm is that it provides good
error rate performance with low computational complexity. The key point is how
to balance the performance and complexity.
1.3 Literature Review

In this section, the existing MIMO detection methods will be simply
overviewed. Generally, the detection methods of MIMO system can be classified
into two types: linear detection algorithm and non-linear detection algorithm. ML
decoding is known as a non-linear detection algorithm, which performs the
exhausted search of the lattice space and provides the optimum error rate
performance. However, it is very complicated and impractical. Thus, the
low-complexity linear detectors are proposed.
Linear detection algorithms such as zero-forcing [8] and minimum
mean-square-error [9] estimation are sub-optimum methods in which the received
signal vector is multiplied by a transformation matrix to get an estimation vector,
and after that, the determination is carried out to get the final solution. In
zero-forcing (ZF), the received signal vector is multiplied by the generalized
inverse matrix of the channel matrix and then quantized to get the result. The
performance of ZF is poor since the noise is enlarged by the generalized inverse

matrix. The minimum mean-square-error (MMSE) detector takes the noise
variance into account and minimizes the square error between the transmitted
signal vector and estimated vector. Thus it can provide better performance than ZF.
Although the linear detection algorithm has very low complexity, their error rate
performance is inferior.
The non-linear detection algorithms include sphere decoding (SD), successive
interference cancellation (SIC) detection, lattice reduction aided decoding (LRD),
semidefinite relaxation (SDR) detection, and so on. They are proposed to improve
the error rate performance.
Sphere decoding is firstly introduced to significantly reduce the average
decoding complexity [14], yet achieving the optimum performance as ML
decoding. After that, the SD has been further discussed in various publications
[15]-[17]. SD decoding is also a search-based algorithm like ML decoding. In ML
decoding, the search is conducted among the whole lattice structure, while in SD,
the searching process is confined inside a hyper sphere of certain radius centered
at the received signal point as illustrated in Fig. 1.4. The solution of ML decoding
is likely to locate inside the hyper sphere. To improve the searching efficiency,
some searching strategies are proposed, such as Fincke-Pohst (F-P) [18] and
Schnorr-Euchner (S-E) [19].
F-P sphere decoding is usually thought as the original sphere decoding
algorithm. The search begins at the root then calculates the weights of connected
branches and nodes. In the F-P strategy, the tree is traversed depth-first and from
left to right, which means enumerating the points located within a sphere along a
10
branch until a node is encountered. For all other branches, the same process is
conducted until all nodes within the hyper sphere are discovered.
It has been shown that the S-E enumeration is more computational efficient
than the F-P enumeration [20]. The S-E strategy performs traversing the tree
depth-first too. But it calculates the branch weights and searches them in
increasing order. And after it obtains a lattice point, the radius of the hyper sphere
is reduced to be the distance between the received signal point and the lattice
point. Then the search process is restarted again with the new radius. As a result,
the lattice points visited is less and the searching process becomes faster.
Figure 1.4
Sphere decoding.
There are two major aspects to be improved for the sphere decoding. The first
one is to determine the initial radius of the hyper sphere. If the radius is too large,
there will be too many node weights need to be calculated. In the contrast, if it is
too small, it would be possible that no point is inside the hyper sphere, and the
search should be restarted again with a larger radius. Thus, a proper initial radius
11
of the hyper sphere can reduce the complexity of SD. In [14] and [21], it uses the
noise variance and probability equation to define a proper initial radius. The
MMSE equalizer is also applied to set the initial radius [22]. Another aspect is the
searching strategy. In [23], a statistical pruning method that uses a set of bounds
based on the minimum metric of the current solution is proposed for S-E sphere
decoding. A preprocessing stage and a new ordering are engaged in the searching
method in [24]. With the new ordering, the nodes are expanded according to the
level and offset coefficients. These searching strategies provide higher
computational efficiency. Although SD is able to provide the BLER performance
of ML detection with less complexity, it has been proven that its expected
complexity is still exponential [25]. Thus it becomes impractical when the system
order is high and the SNR is low.
The successive interference cancellation detection [88] has the error rate
performance gain by sacrificing a certain complexity compared with the linear
detections. It detects the signal from the first transmit antenna instead of that of all
the transmit antennas, and then subtract its impact from the received signal vector.
After that, it detects the signal from the second transmit antenna. At this time, the
row of the channel matrix that is corresponding to the first antenna is deleted.
Thus, the dimension of channel matrix becomes M R ( M T 1) and the
dimension of the transmitted signal vector is reduced to M T 1 . The process
continues until all the elements of the signal vector are detected. Because the
detected signal has effect on the later signal to be detected, we need to firstly
detect the reliable signals in order to reduce the error propagations. Usually,
ordering is adopted to improve the performance. The authors in [10] proposed an
12
ordered successive interference cancellation algorithm based on the maximum

signal to noise ratio (SNR). In [7, 11] it detects the signal with the maximum
signal to interference plus noise ratio (SINR) first, and the ordering obeys the
log-likelihood ratio rule in. ZF or MMSE methods are engaged to obtain an
estimate of the value in each dimension of the transmitted signal vector, which are
known as ZF-SIC and MMSE-SIC [12, 13], respectively.
Lattice reduction aided detectors [26]-[29] are developed in order to improve
the error rate performance of MIMO systems. They are combined with the linear
detection such as ZF and MMSE, and so are called LR-ZF and LR-MMSE [29].
In MIMO transmissions, if the column of the channel matrix is less correlated, the
transmit signal from different antennas is more independent and can be detected
more correctly. The lattice reduction technique is to find an optimal lattice basis of
a matrix which is more orthogonal and short compared with the original lattice
basis. In MIMO detection, it transforms the channel matrix H into new matrix
H by an unimodular matrix. The more orthogonal of the matrix H is, the more
significant improvement of the error rate performance will be obtained.
However, lattice reduction is also an N-P hard problem. The most popular
lattice reduction algorithm is the LenstraLenstraLovsz (LLL) algorithm [30]. It
is a polynomial time algorithm that operates iteratively and stops when the lattice
basis is obtained. It has been shown that in LLL algorithm, the average number of
iterations is bounded by a polynomial in the dimension of the lattice [31]. Then in
[32], researchers proposed a novel joint sorting and reduction algorithm that can
offer better diversity order and error rate performance than the traditional LLL at
the cost of a polynomial computation time. A complex LLL reduction algorithm is
13
introduced in [33]. It can reduce the average complexity by almost half of the
conventional LLL and achieve full diversity in LR detection.
Recently, MIMO decoders using semidefinite relaxation approach have
attracted great attention. They are able to provide acceptable BLER performance
and feature polynomial worst-case complexity [34]. Since the ML decoding
problem has the optimum error rate performance, the SDR detection applies the
convex optimization toolbox such as SEDUMI [35] to solve the convex relaxation
of ML optimization problem, which is called semidefinite programming (SDP).
The SDR approach was firstly applied to detect binary phase-shift keying
(BPSK) and four quadrature amplitude modulation (4-QAM) signals [36]-[38]. It
has been shown that the SDR detector for BPSK can achieve full receive diversity
[39]. Then the extensions to different SDR techniques for 16-QAM signals had
been
proposed,
such
as
polynomial-inspired
SDR
(PI-SDR)
[40],
bound-constrained SDR (BC-SDR) [41] and virtually-antipodal SDR (VA-SDR)

[42], all exhibit acceptable BLER performance and relatively low complexity. In
[43], it has been proved that there exists an equivalence among PI-SDR, BC-SDR
and VA-SDR for 16-QAM, all of them provide the same BLER performance.
1.4 Motivation and Contribution of the Thesis

MIMO system is a promising solution to the high data rate and high reliability
of the future wireless communications. It can increase the channel capacity and
spectrum usage efficiency without the need of additional channel bandwidth. The
detection design is one of the major challenges of MIMO systems, since there is
always a tradeoff between the computational complexity and the error rate
14
performance. The thesis aims at investigating detection algorithms to reduce the

complexity while retain good system performance.
The main contributions of the thesis are as follows:
Geometrically analyzing the signal detections of MIMO system, which is
another perspective of reconsidering the principle of the ML decoding. Based on

this, a geometric decoding algorithm is proposed, which can provide the optimum
error rate performance.
Giving the extension of the existing SDR detection algorithms to
high-order modulation MIMO system, and proposing a novel SDR detection

algorithm for 256-QAM MIMO system which could offer better performance than
its existing counterparts. The theoretical analysis on the tightness of the SDR
detection algorithms is also conducted.
Combining the SDR detection algorithms with the sphere decoding,
which can achieve the optimum BLER performance while retain acceptable
computational complexity.
Proposed the lattice-reduction-aided semidefinite relaxation detection. It
is able to achieve the full diversity gain and improve the error rate performance
with a little complexity added.
1.5 Thesis Outline

This thesis is composed of six chapters. All references quoted within this
thesis are listed in the References part. The thesis outline is given as follows:
Chapter 1 is the introduction on the background information of the MIMO
system and literature review of the previous work. The research motivation and
15
aims of this thesis are also reported.

Next in Chapter 2, the state-of-the-art of the MIMO detection algorithms are
investigated.
In Chapter 3, the geometric analysis of the signal detection is given. Then an
optimum ellipsoid-searching decoding algorithm is introduced and simulation
results are given.
After that, convex optimization and the interior point method are investigated
in Chapter 4. The semidefinite relaxation detection algorithms for low-order
modulation and high-order modulation system are given. Then the theoretical
analysis on the tightness and the complexity of SDR are also involved. In addition,
the semidefinite relaxation initiated sphere detector is given. At the end, the
simulation results are given.
Next in Chapter 5, two different lattice-reduction-aided semidefinite
relaxation detection algorithms are investigated with the simulation results.
Finally, the conclusion and recommendation are given in Chapter 6.
16
Chapter 2 State-of-the-Art MIMO Detection Algorithms
CHAPTER 2
STATE-OF-THE-ART MIMO DETECTION
ALGORITHMS
2.1 Introduction
This chapter will be devoted to review the state-of-the-art of the MIMO
detection algorithms. Firstly, the linear decoders which include the zero-forcing
decoder and the minimum mean-square-error decoder are introduced. Linear
decoder has the advantage of extremely low computational complexity. However,
they suffer from the unsatisfactory block error rate (BLER) performance.
Secondly, the sphere decoding is elaborated. Similar to the ML decoding, sphere
decoding is another searching-based detector, and it can offer optimum BLER
performance. Nevertheless, its expected complexity is still very high, although it
has been dramatically reduced compared with the ML decoder. Thirdly, the
successive interference cancellation and the lattice reduction detection are
introduced. They both can be combined with the linear decoders, so as to improve
their BLER performance with certain complexity added.
2.2 Linear Decoders

As elaborated in Chapter 1, the complexity of ML decoder increases
exponentially with the number of jointly decoded symbols [38]. However, the
complexity of some sub-optimal decoders increases linearly with the number of
jointly decoded symbols, i.e. the number of symbols in one code block. These
17
decoders are called linear decoders. The well-known linear decoders include the
zero-forcing (ZF) decoder and the minimum-mean-square-error (MMSE) decoder.
2.2.1 Zero-Forcing
Zero forcing (ZF) detection [8] is the simplest linear detection algorithm. It
forces the impact of the channel matrix to be zero and is given by:
1
x ZF = Q ( H*r ) = Q ( H H H ) H H r
(2.1)
where H* is the pseudo-inverse matrix [45] of the channel matrix H , the

superscript H represents the complex conjugate transpose, Q ( i ) is quantization
to the constellation values. The decoder finds the value of x ZF which is the closest
to (H H H ) 1 H H r . However, the noise signal in each stream becomes correlated to
each other by the matrix H* , which results in decoding error. Thus ZF detection
is of suboptimum performance.
The diversity order of ZF detection is M R M T + 1 . The full diversity gain
given by ML decoding is M R . Thus, the performance of ZF is poor especially
when the number of antennas is large. The performance can be improved by the
minimum mean-square-error detection approach.
2.2.2 Minimum Mean-Square-Error

As suggested by its name, minimum mean-square-error (MMSE) [9] is an
approach to minimize the mean-square-error (MSE). The MMSE detection can be
written as:
x MMSE = Q ( H H H + n 2 I ) H H r
1
(2.2)
18
where n 2 denotes the noise power and I is the identity matrix. Compared with
the equation (2.1) of ZF, the difference lies in the term of n 2 I . ZF decoding
separates the co-channel signals and cancels all the inter-symbol-interference (ISI).
However, this inevitably leads to noise enhancement. On the other hand, MMSE
detection attempts to minimize the overall errors which are caused by the noise,
and make balance between the ISI mitigation and noise enhancement. Assuming
that the noise n 2 is zero, the MMSE becomes the same with ZF. Generally,
MMSE decoding tends to provide better error rate performance than ZF decoding.
Both ZF and MMSE are classified as linear detection algorithm. Although
they have very low complexity, their error rate performances are still far away
from being satisfactory.
2.3 Sphere Decoding

ML decoding applies an exhaustive search process, in which, all the lattice
points of the constellation are visited, and then choose the one with the minimum
distance to the received point as the solution. Although it is an optimum decoding
algorithm, its computational complexity increases dramatically with the increase
in the number of the antennas and the constellation size [5]. Sphere decoding [18]
(SD) is proposed aiming at reducing the decoding complexity, while retaining the
optimal BLER performance. It searches only a subset of the lattice points that are
located inside a hyper sphere centered at the received signal vector. The lattice
points which are located outside the hyper sphere will not be taken into account.
Thus, the number of lattice points searched by the algorithm depends on the initial
radius of the hyper sphere. Although the points inside the hyper sphere are not
19
searched exhaustively, the calculations are based on the branches in a tree which
are possible to lead to the final result.
There are two main problems to be solved in SD. One is how to determine the
initial radius. If the radius is too large, there will be a large number of points
inside the hyper sphere and the complexity will be too large. If the radius is too
small, there may be no point inside the sphere and the search process has to be
restarted. Usually, the ZF equalized result is taken to calculate the initial radius.
The other problem is how to tell which lattice points are located inside the sphere.
It is very difficult to identify whether a lattice point is located inside a hyper
sphere or not, but it is very easy to do so for a two-dimensional sphere by simply
checking whether the integer values of the lattice points lie in the interval of the
sphere. Inspired by this, for an N-dimensional sphere the points can be determined
from one dimension to the other successively. It means that for a - dimensional
point, if its ( 1) dimension values lie in the ( 1) dimensional sphere of a
certain radius, there will be a new interval for its th value to determine if it
lies in the - dimensional sphere.
SD is aimed at finding out the solution x SD which is the same with x ML that
has the minimum Euclidean distance from the received signal r . It searches the
lattice points x within a M T dimensional hyper sphere, which can be given by:
r Hx d 2
(2.3)
where d is the initial radius of the hyper sphere. The searching of the lattice
points is a kind of iterative algorithm. For simplicity, we have to separate the
original problem into several sub-problems. Thus, the channel matrix H is firstly
reduced into an upper triangular matrix by using the QR decomposition:
20
R
H =Q
0
(2.4)
r1,1
0
0
= [Q1 , Q 2 ] 0
where Q
M R M R
r1,2
r2,2
0
0
r1, MT
r2,M T
rM T , M T
0
MT MT
is an orthogonal matrix and R
matrix. Decomposing Q by Q = [Q1 Q 2 ] , where Q1
is an upper triangular
M R M T
and Q 2
M R ( M R M T )
the Euclidean distance involved in the ML decoding can be written as:
r Hx
R
= r [Q1 , Q 2 ] r
0
Q T
R
= 1T r x
0
Q 2
(2.5)
= Q1T r Rx + Q 2T r
Let r = Q1T r , (2.5) then becomes:

2
r Rx + Q 2T r
Thus, to minimize the Euclidean distance r Hx
r Rx
(2.6)
2
is equivalent to minimize
in the sphere decoding.

2
Let d = d 2 QT2 r , (2.3) can be rewritten as:

2
r Rx d 2
(2.7)
21
r1 r1,1
r 0
2

rM T 0
r1,M T x1
r2, M T x2
d 2
0 rM T , MT xM T
r1,1
r2,2
0
2
2
ri ri , j x j d
i =1
j =i
MT
(2.8)
where ri , j is the ( i, j ) th element of R . The inequality equation (2.8) is then

expanded to be:
MT
(rM T rM T , MT xMT ) 2 + (rM T 1 rMT 1, M T xM T rMT 1, M T 1 xM T 1 ) 2 + ... + (r1 r1,i xi ) 2 d 2

i =1
(2.9)
It can be seen from the inequality equation (2.9) that there is only one unknown
quality xMT in the first term (rM T rM T , MT xM T ) 2 . Similarly, there are two unknown
qualities xMT and xMT -1 in the second term (rM T 1 rM T 1, M T xM T rM T 1, M T 1 xM T 1 ) 2
and so on. The necessary condition of inequality equation (2.9) is:
(rM T rM T , M T xM T ) 2 d 2
(2.10)
So the value of xM can be solved at first. The boundary of xM is:

T
r d
r + d
MT
xM M T
T
rM T , M T
rM T , M T
(2.11)
It is easy to find out the possible values of xM by using (2.11). For example,
T
they are the odd numbers in the interval for QAM constellations. Usually, there
may be more than one possible value which are saved in the memory.
Secondly, one of the possible values of xM T is selected to solve xM T 1 by the
following inequality equation:
(rM T rM T , M T xM T ) 2 + (rM T 1 rMT 1, M T xMT rM T 1, MT 1 xM T 1 ) 2 d 2
(2.12)
22
The boundary of xM
is
2
d 2 ( r r
MT
M T , M T xM T ) + rM T 1 rM T 1, M T xM T
rM T 1, MT 1
xM 1
T
2
d 2 ( r r
MT
M T , M T xM T ) + rM T 1 rM T 1, M T xM T
xM T 1
rMT 1, M T 1
(2.13)
It can be seen from (2.13) that the signals for the previously detected
dimensions have been subtracted from the received signal.
If there is no possible value satisfying the inequality equation, the searching
process goes back to the previous step and then selects another possible value.
When all the element values of one lattice point are obtained, its Euclidean
distance away from the received point is calculated. The new distance is of course
smaller than the initial radius d , so we replace it by the new distance in (2.9) and
restart the searching process.
The searching process works in an iterative way to update the radius until no
more vectors satisfy the inequality equation (2.9). The last vector is then taken as
the decoding solution.
In conclusion, the searching process is done from the M T th dimension to
the 1st dimension. The intervals of x are calculated and the value of the each
element of x are then determined. When one lattice point is obtained, the radius
of the hyper sphere is reduced and the searching is restarted again. The searching
process goes on until only one branch is left, which is then taken as the final result.
23
Figure 2.1
The tress search structure of sphere decoding.
The searching process to determine all the lattice points in a

M T dimensional hyper sphere can also be illustrated by the tree search structure
shown in Fig. 2.1. Each level of the tree represents each dimension of the signal
vector x. The nodes in the figure are the values of the element of x. Take 16-QAM
constellation for example: a node emits four sub-nodes which are the values of -3,
-1, 1, 3 from the left to right. The black nodes denote the possible values that
have been visited. From the figure, we know that in the M T th dimension, the
possible values of xM T are -1 and 1. Then we select -1 to continue the searching
process down to the next dimension. It can be seen that there is one branch
emitted from the node xM T =-1 that is visited, which means all the element values
of one lattice point are obtained. Then we calculate the new radius and restart the
searching process. The search continues until no other branch can be found, the
lattice point corresponding to the last branch is selected as the final solution.
Although lots of new methods have been proposed to determine a proper
initial radius [14, 21], and many new searching strategies [22-24, 62-66] have
24
been developed to improve the searching efficiency, sphere decoding still suffers
from some drawbacks. The expected complexity of SD increases with the number
of antennas and the size of the constellation [15]. So SD is not suitable for large
size MIMO system. In addition, the complexity of SD is not fixed, but changing
with the number of the nodes visited in the searching strategies, which makes it
impractical for hardware implementation. As a result, the sub-optimum successive
interference cancellation and lattice reduction aided detection are developed to
improve the performance of linear detection with a comparable low complexity.
2.4 Successive Interference Cancellation

The successive interference cancellation (SIC) [88] is a sub-optimum nonlinear equalizer. Its basic idea is to estimate the current symbol and subtract the
impact of those interfering symbols which have been detected earlier. So the
distortion on the current detecting symbol caused by the previous symbols is then
eliminated. The SIC method can be combined with the linear detection algorithms
such as ZF and MMSE to improve their error rate performance. However this will
cause additional complexity.
Firstly, we apply QR decomposition of the channel matrix H :
H = QR
(2.14)
where Q is a unitary matrix and R is an upper triangular matrix. The MIMO

system model (1.5) can be transferred into:
r = QRx + n
QT r = Rx + QT n
(2.15)
Let r = QT r and n = QT n , the i th row of the r is:
25

MT
ri = Ri , j x j + ni
(2.16)
j =i
which is composed of the symbols from xi to xM T . Also, the M T th row of r

only involves the symbol xM T :
rM T = RM T , M T xM t + nM T
(2.17)
As a result, we could detect the symbol xM T first by:

xM T =
rM T
RM T , MT
(2.18)
Then the ( M T 1) th symbol xNt 1 will be detected next, where the interfering
symbol xMT can be canceled by subtracting its impact:
xM T 1 =
rM T 1 RMT , MT xMT
RMT 1, MT 1
(2.19)
Thus, the i th symbol is detected by:

xi =
ri Ri +1,i +1 xi +1
Ri ,i
(2.20)
The SIC works from the M T th symbol xMT upwards to the 1st symbol x1 till all
the symbols are detected in the end.
The drawback of the SIC detection is the error propagation caused when there
is wrong decision in one symbol. If the former detected symbol is wrong, the
latter ones have very large probability to suffer from errors. In order to solve this
problem, the methods of detection ordering are proposed [7, 10, 11]. The symbol
detection ordering is in accordance with the power of the symbols, which can be
determined by the SNR or the signal to interference plus noise ratio (SINR). That
means that the symbol which has large SNR or SINR will be detected first.
26
Similarly, the detection ordering will improve the error rate performance but
increase the complexity.
2.5 Lattice-Reduction Aided Detection

2.5.1 Lattice Reduction
The lattice structures are very popular in field of wireless communications
especially the MIMO system. A lattice is composed by a discrete set of N
dimensional vectors which is denoted by LN . Every lattice can be produced by a
linear combination of a set of independent integer vectors, which is called the
basis of the lattice given by {b1 , b 2 ,
b M } , M N . Any lattice has infinitely
many lattice bases, and the lattice basis composed of near orthogonal vectors
which are also with short lengths is the desirable one.
The procedure of determining a lattice basis with short and near orthogonal
vectors is termed as Lattice Basis Reduction [19]. To realize the reduction goal is
an N-P hard problem, whose running time varies exponentially with the
dimension of the lattice. A popular suboptimum reduction algorithm which is
called LLL algorithm [30] is proposed to reduce the complexity of Lattice Basis
Reduction.
In MIMO systems, for ill conditioned channel, the noise enhancement is large.
However, for orthogonal channel, there will be no noise enhancement. The lattice
reduction (LR) algorithm is employed to transform the original channel matrix
into a new channel matrix with much better channel condition [47, 76-81]. The
new channel matrix is composed of vectors with the shortest lengths or roughly
orthogonal to each other. The LR algorithm can be combined with the existing
27
linear detectors such as ZF and MMSE detectors. The linear detectors with a
better channel conditioned matrix will achieve better BLER performances.
2.5.2 Analysis of MIMO Detections in Lattice Space

This section is to analyze the different detections including ZF decoding, SIC
detection and ML decoding in the perspective of the lattice space. Herein, we
assume a 2 2 MIMO system using 4-PAM modulation. The original transmit
signals lattice is shown in Fig. 2.2.
Figure 2.2
The original transmit signal lattice.
With the distortion resulted from the channel matrix, the received signal
lattice without Gaussian noise is indicated in Fig 2.3. The vectors [h1 h 2 ] is the
basis of the channel matrix H [65].
28
Figure 2.3
The received signal lattice.
Fig. 2.4 and Fig 2.5 show the decision regions of the ZF decoding and SIC
detection, respectively.
Figure 2.4
ZF decoding decision region.
29
Figure 2.5
SIC detection decision region.
It can be seen that the ZF decoding transforms the received signal to the
transmitted signal space to make the decision, where the decision region becomes
parallelogram. The decision region of SIC detection is rectangular which is
different from that of the ZF decoding.
The ML decoding decision region as indicated in Fig. 2.6 is the best one,
which is formed by two roughly orthogonal bases. Any point in a decision region
is closer to the lattice point in its decision region than to other lattice points in
other regions. It thus inspires that if we make the decision of the signal in an
orthogonal or roughly orthogonal basis, then transform the decision value back to
the original signal space, the detection result will be much better than that in the
original basis.
30
ML decoding decision region.
Figure 2.6
2.5.3 Lattice Reduction Aided Linear Detections

The lattice reduction for MIMO system aims at transforming the original
channel matrix H into a new one H which contains more orthogonal and shorter
basis vectors by multiplying a unimodular matrix T . The unimodular matrix has
only integer elements in it and its determinant equals to 1 . For linear detections,
the decoding performance is much better for the case of orthogonal matrix.
Herein, H =HT and let z = T1x , the MIMO system model of (1.5) can be
written as:
r = Hx + n =HTT1x + n = Hz + n
(2.21)
where Hx and Hz represent the same point in the lattice space.

In this equivalent system model, the zero-forcing detection is applied to
obtain [29]:
z LR ZF = H*r
= H* Hz + n =z + H*n
=T1x ZF
(2.22)
31
The new pseudo-inverse matrix H* tends to generate less noise enhancement than
the original pseudo-inverse matrix H* . The solution z LR ZF is then quantitized
and multiplied by T, so as to recover the solution in the transmit signal lattice:
z LR ZF =TQ ( z LR ZF )
(2.23)
LR-ZF decision region.
Figure 2.7
Fig. 2.7 illustrates the decision region of LR-ZF, which is a parallelogram

with its sides parallel to the new channel basis vectors h1 and h 2 . The
parallelogram is much less stretched than the parallelogram of ZF shown in Fig.
2.5. This explains why the LR-ZF detector can offer better BLER performance
than ZF detection.
In MMSE detection the noise term is taken into consideration. Apply MMSE
detection in the system model (2.21) to obtain [65, 76]:
z LR MMSE = H H H + n 2 IT H T
=T1x MMSE
HH r
(2.24)
Similarly, the final solution becomes:

z LR MMSE =TQ ( z LR MMSE )
(2.25)
32
Another LR-MMSE detection for an extended channel matrix is introduced in

[29]. Define the extended channel matrix which is a ( M T + M R ) M T matrix:
H
H ext =
n I
Define a
(2.26)
( M T + M R ) 1 extended received vector
x ext :
x
x ext =
0
(2.27)
The extended MMSE detector is:
z ext MMSE = H ext H H ext
H ext H rext
=Text 1x ext
(2.28)
where H ext =H ext Text .

This extended MMSE detector is in fact the same as the LR-MMSE detector
given in (2.24) since they both follow the MMSE detection scheme of (2.2) in the
system model (2.21) except that the lattice reduction is conducted on H ext instead
of H . Both of them can outperform the MMSE detector because the noise
enhancement is less when perform on H or H ext .
The diversity order of the lattice reduction aided linear detectors can reach the
full diversity of M R , which is the same as that of ML detection, and much better
than that of the linear detectors. There are significant BLER performance
improvements of LR-ZF and LR-MMSE compared with ZF and MMSE.
However, the performances of LR-ZF and LR-MMSE are still worse than ML
decoding due to the noise enhancement. The complexity of LR aided linear
detectors is a little larger than that of their corresponding linear detectors due to
the additional polynomial-time for preprocessing the LLL algorithm.
33
2.6 Summary
Several existing detection algorithms for MIMO systems have been reviewed
thoroughly. As far as the comprehensive performance is concerned, they either
suffer from very high computational complexity, or rather bad error rate
performance. Thus, a lot more attentions should be paid on developing new
detection algorithms for MIMO systems, especially for the cases in which large
number of antennas and high level modulation are involved.
34
Chapter 3 Geometric Detection Algorithms
CHAPTER 3
GEOMETRIC DETECTION ALGORITHMS
3.1 Introduction
Since the minimum Euclidean distance principle of ML decoding could result
in the optimum error rate performance, the purpose of this chapter is to introduce
another perspective of reconsidering this principle. In the hyper space spanned by
the transmit lattice points, the Euclidean distance involved in the ML decoding is
found to be related to a series of concentric hyper ellipsoids. Searching the lattice
point with the minimum Euclidean distance away from the received signal point is
equivalent to searching for the lattice point that lies on the surface of the smallest
possible hyper ellipsoid. Decoding algorithms following this perspective are often
termed as geometrical detection. In this chapter, the geometrical analysis of signal
decoding for MIMO channels is presented. Then, the proposed ellipsoid searching
decoding algorithm (ESA) [69] are elaborated. It is an add-on to the standard
suboptimal detection schemes, such as ZF or MMSE. Simulation results
demonstrate that it offers the optimum error rate performance and higher diversity
gains than the standard suboptimal detection schemes.
3.2 Geometrical Analysis of Signal Decoding for MIMO Channels

The Euclidean distance involved in the ML decoding can be rewritten as:
f ( x ) = r Hx
= ( x x c ) 1 ( x x c )
T
(3.1)
where the matrix = ( HT H ) . The vector xc is the result of ZF equalization

1
35
which is determined by:

x c = ( HT H ) HT r
1
= x + ( HT H ) HT n
1
(3.2)
= x+n
where n = ( HT H ) HT n .
1
Substituting (3.1) into (1.8) yields:

x ML = arg min ( x x c ) 1 ( x x c )
T
(3.3)
It can be seen from (3.2) and (3.3) that in the absence of noise, i.e., the
transformed Gaussian noise term n = ( HT H ) HT n , both ZF equalization and
1
ML decoding result in the same correct solution. The reason why ML decoding
can offer much better performance than ZF equalization lies in the fact that the
transformed Gaussian noise has been minimized by the exhaustive search used in
ML decoding. However, the results of ZF equalization are directly distorted by the
transformed Gaussian noise n .
By using eigenvalue decomposition [72-73], the matrix M can be
decomposed into:
= ( HT H ) =VV T
1
where
arranged
MT M T
order,
and
= diag 1 , 2 ,..., MT
in
descending
(3.4)
, represents the
V = V1 , V2 ,
MT
, VNT
eigenvalues
MT M T
is
the
corresponding eigenvector matrix.

The condition of the channel can be indicated by the channel condition
number which is defined as:
( H ) =1 / M
(3.5)
It has been proven [68] that the channel condition number ( H ) has a profound
36
impact on the error rate performances of the linear detections. If ( H ) is low,

the error rate performances of the linear detections are very close to optimum.
However, if the ( H ) is very high, the error rate performances become
unacceptably bad due to the considerable noise contamination n . In the
geometric point of view, the contour surfaces of the probability density function
(PDF) of the noise n are the hyper ellipsoids with their directions of axis
decided by the eigenvector matrix V . The direction of the i th axis of the
hyper ellipsoids is given by the direction of the vector represented by the i th
column of V , and the length of the i th axis is proportional to the square root
of the corresponding eigenvalue.
Fig 3.1 illustrates the PDF of the received signal vector of ZF detections in
2 2 MIMO systems using BPSK modulation. The solid lines are the boundary
lines of the ZF decision regions and the dash lines are the boundary lines of the
ML decision regions. The channel condition number in Fig. 3.1 (a) is 1.5 which
can be considered as a good channel, while the channel condition number in Fig.
3.1 (b) is 7.5 which can be thought as a bad channel. In the case with good
channel condition, the boundary lines of ZF decision region is very likely close to
those of the boundary lines of ML decision region. So the performance of ZF
detection is good in this case. In the case with bad channel condition, the ML
decision regions are able to match to the PDF of the received signal vector, but the
ZF decision regions can not. The boundary lines of the ML decision region are
approximately orthogonal to the dominant principal axis which is corresponding
to the vector V1 . In general, the decision regions of ZF detector cannot have this
property since its boundary lines always go through the origin point.
37
(a)
(b)
Figure 3.1 Probability density function of the received signal vector of ZF
detections in 2 2 MIMO systems.
(a) Case for good channel condition. (b) Case for bad channel condition.
Geometrically, the Euclidean distance function
f ( x ) given in (3.1)
represents an elliptic paraboloid [70] in an M T + 1 dimensional space with its

axis perpendicular to an M T dimensional subspace spanned by the signal vectors
in . xc is the global minimum point of the elliptic paraboloid and is located
on the subspace spanned by the signal vectors in as shown in Fig. 3.2. From
(3.1) it can be known that the function f ( x ) reaches its minimum value at xc
38
if it is a continuous function, that is:

f (x) min = f (xc ) = 0
(3.6)
The horizontal-cross section of the elliptic paraboloid (3.1) is a M T dimensional

hyper ellipsoid given by:
f ( x) = a2
(3.7)
where a 2 represents the height of the cross section above the M T dimensional
space as shown in Fig. 3.2. The length and the direction of the i-th semiaxis of the
hyper ellipsoid are given as a i and Vi , respectively. With different value of
a 2 , a series of concentric hyper ellipsoids could be obtained and projected onto

the subspace spanned by the vectors as shown by the dash lines in Fig. 3.2. Thus,
searching the lattice point with the minimum Euclidean distance is equivalent to
searching for the lattice point that lies on the surface of the smallest possible
hyper ellipsoid.
Figure 3.2 Elliptic paraboloid with axis perpendicular to a subspace spanned by

lattice points.
39
3.3 Ellipsoid-Searching Decoding Algorithm

From section 3.2 we know that f (x) = a 2 represents a hyper ellipsoid
centered at the point xc , moreover, the length and the direction of its
i th semi-axis are given as a i and Vi , respectively. By choosing different
values of a , a group of similar hyper ellipsoids can be obtained. Thus, the

solution of ML decoding must be located on a hyper ellipsoid which has the
minimum surface area among its similar hyper ellipsoids.
Figure 3.3 Elliptic paraboloid in 3-dimensional space.
Fig. 3.3 shows a two dimensional lattice point space ( x1 x2 plane) with
three lattice points Point 1, Point 2, and Point 3 as shown in the figure. With
different a 2 , a group of similar hyper ellipsoids can be obtained, and their
projection onto the x1 x2 plane are ellipses which are all centered at the point
40
xc = H 1r . For each lattice point, there exists an ellipse that passes through it. The
corresponding ellipse of the ML solution is the one that has the minimum area. As
shown in Fig. 3.3, Point 1 is taken to be the ML solution while Point 2 and Point 3
are not, since Point 1 is located on the inner-most ellipse which has the minimum
area.
However, finding the smallest hyper ellipsoid containing the solution signal
vector is not an easy task. If we use the largest hyper ellipsoid which contains all
the signal vectors, then the complexity will be the same as ML decoding. Here we
propose an ellipsoid-searching decoding algorithm that uses a small hyper
ellipsoid containing the solution symbol vector to start the search and then
identify all the symbol vectors inside. The ESA consists of the following 4 steps:
3.3.1 Start with Zero-Forcing Points
It is well known that the ZF decoding is a kind of linear equalization

algorithm. Although it can not offer very good performance like ML decoding, its
solution however usually lies in the neighborhood of the transmit signal point.
Thus we can consider choosing the hyper ellipsoid that goes through the ZF
solution to start the searching process. Firstly, the ZF equalized x zf is solved.
Then its corresponding azf2 is computed. The starting hyper ellipsoid is obtained
as:
f ( x zf ) = azf2
(3.11)
3.3.2 Determine a Circumscribed Hyper Rectangle
After determining the hyper ellipsoid, the next key task is to identify whether
41
there are any lattice points located inside this hyper ellipsoid. The axes of the
M T -dimensional rectangular coordinate system for the lattice point space are
denoted as i - axes. Since the directions of the hyper ellipsoids semiaxes are not
in parallel with the axes of the coordinate system of the lattice point space, it is
rather complicated to directly use the surface equation (3.11) of the hyper
ellipsoid. Here we propose to use a circumscribed hyper rectangle as follows.
We set up a new M T -dimensional rectangular coordinate system with i axes ( i = 1, 2,3,..., M T ) being coincided with the i th semiaxis of the hyper
ellipsoid and the origin coincided with the global minimum point xc . We use the
superscript prime to denote the variables in the new coordinate system. The
coordinates of the 2M T apexes of the circumscribed hyper rectangle in this new
coordinate system are given by
k p = xp1 , xp 2 ,...xpMT
where p = 1, 2,3,...2 MT ,
(3.12)
xpj = azf j , and azf is related to the hyper
ellipsoid given by (3.11). It can be easily shown that, by using coordinate

transformation, the coordinates of the 2M T apexes in the original lattice point
space are:
k Tp = V ( k p ) + xc
T
(3.13)
where V is the eigenvector matrix in (3.4), and it serves as the transformation

matrix:
VT = V1 , V2 ,
v11
v12
, VM T = v13
v1M T
v21
v31
v22
v32
v23
v33
v2 M T
v3 MT
vM T 1
vM T 2
vM T 3
vM T M T
(3.14)
42
Thus the value of the i th component of k p can be obtained as

MT
x pi = ( vqi xpq ) + xci
(3.15)
q =1
where xci is the i th component of xc . Since xpq = azf q , the maximum and
minimum boundaries of the values of the each component in k p in the i - axes
can be expressed as:
MT
xi _ max = xci + vqi azf q
(3.16a)
q =1
MT
xi _ min = xci vqi azf q
(3.16b)
q =1
Since the circumscribed hyper rectangle encloses the hyper ellipsoid, any
lattice point s = s1
s2 ... sM T inside the hyper ellipsoid satisfies:
xi _ min < si < xi _ max
i = 1, 2,3,..., M T
(3.17)
It should be noted that this is not a sufficient condition for identifying the
lattice points lying inside the hyper ellipsoid.
From (3.17), we can obtain the possible value set i = { i1 , i 2 , i 3 ,
of the
i th element for the lattice points located inside the hyper ellipsoid. So the
search set becomes a larger hyper rectangle that encloses the circumscribed hyper
rectangle. For PAM and QAM, the elements of j are the odd numbers between
xi _ max and xi _ min , and it can be easily shown that the number of elements is:
MT
Numi = vqi azf q

q =1
(3.18)
3.3.3 Narrow the Search Set into Ellipsoid

As mentioned before, the search set becomes a larger hyper rectangle and the
43

MT
number of lattice points inside is
Num . If there is any

i
i =1,i l
Numi equals zero,
then it means that there is no lattice point located inside the hyper ellipsoid. The
searching process will terminate and the zero forcing point chosen before is
considered as the solution.
Otherwise, assuming the possible value set has the largest number of
elements among all the possible value sets, we form the combinations from the
other M T 1 possible value sets, and then substitute each of these combinations
into (3.11), to determine the lattice point elements of the possible value set
that are located inside the hyper ellipsoid. In doing so, the number of
combinations that need to be considered is smaller and hence lesser computation
complexity. Denoting the k th combination by:
Com k = 1,k , 2,k ,
1,k , +1,k
, M T , k
(3.19)
MT
k = 1, 2,...,
j =1, j
Num j
where j ,k represents an arbitrary element of the set j .

Geometrically, the Com k is a line pierced through the hyper ellipsoid. The
intersection of the line and the hyper ellipsoid consists of two points, known as
Emax,k and Emin, k along the th axis. Hence, the corresponding possible value
set ,k = { ,1,k , ,2, k ,...} for the th element of the lattice points are the
odd numbers between Emax,k and Emin, k . Thus, any lattice point that is located
inside the hyper ellipsoid can be expressed as:
x d ,k = 1, k , 2, k ,
1,k , ,d ,k , +1,k
, MT ,k
(3.20)
d = 1, 2,..., nk
where nk is the number of the elements of
,k for Com k .
44
3.3.4 Calculate the Euclidean Distance

All the Euclidean distances of the signal vectors inside the hyper ellipsoid can
be calculated recursively. Let the Euclidean distance the signal vector x d ,k is
denoted as d ,k . So the Euclidean distance d +1,k of the signal vector x d +1,k can
be written as:
d +1,k = d ,k + d ,k
(3.21)
where x d +1,k = x d ,k + 2 .
Substituting (3.21) into (3.1), we will get:
d ,k = 4 hi 4 ( r Hx d ,k ) hi
2
(3.22)
After all the Euclidean distances are calculated, the signal vector with the
minimum distance is then selected as the solution.
3.3.5 Examples
The following subsections will give two examples of the ESA in two
dimensional space and three dimensional space.
3.3.5.1 2-D lattice space

For a 2 2 8-PAM MIMO system, the lattice set is a 2-dimensional space as
shown in Fig. 3.4, where it is assumed that the ellipse and its circumscribed
rectangle have been determined using our proposed method as described
previously. The semiaxes of the ellipse are in parallel with vectors V1 and V2
with lengths azf 1 and azf 2 , respectively. The global minimum point xc
is marked by a triangle on the figure. The coordinates of the four apexes, A, B, C
and D, in the new coordinate system are given by A = ( azf 1 , azf 2 ) ,
45
B = azf 1 , + azf 2
),
C = azf 1 , azf 2
) , and
D = azf 1 , + azf 2
, respectively.
Substituting these vectors into (3.15) yields the corresponding coordinates in the
lattice point space. From (3.16) the x1 coordinates of points A and D are chosen
as
x1_ min
and x1_ max , respectively, and the x 2 coordinates of points B and C
are chosen as x2 _ min
and x2 _ max , respectively. Using (3.16), we can obtain a
possible set of values along each axis, i.e., two values {1, 3} along the x1 -axis
and one value {1} along the x 2 -axis. Since the number of values along the
x1 -axis is larger than that along the x 2 -axis, we substitute 2,1 = 1 into the hyper
ellipsoid equation (3.11). As shown in Fig. 3.5, the possible value along the
x1 -axis is 1,1,1 = 3 , so the point x1,1 = [3 1]
is obtained. Since it is the only
point located inside the ellipse, it would be the final solution.
Figure 3.4 2-D lattice space example.
3.3.5.2 3-D lattice space

Here, we continue to consider the case of 3-dimensional lattice space, namely
46
3 3 8-PAM. Fig. 3.5 shows a 3-dimensional ellipsoid with its circumscribed
rectangle which has been set up by the method introduced in section 3.3.2. xc is
the center of the ellipsoid, whose semiaxes are aligned along vectors V1 , V2 ,
V3 , with their lengths being azf 1 , azf 2 and azf 3 , respectively. By
substituting the coordinates of the eight points A to H to (3.15) and (3.16), x1_ min
and x1_ max , x2 _ min
and x2 _ max , x3_ min and x3_ max , which are all marked as
dots are obtained. The possible set of values along x1 -axis is {1, 3, 5}, and the
possible set of values along the x 2 -axis is {1, 3}. Along x3 -axis, the possible set
of value is {-1}. Since the number of possible values along the x1 -axis is the
largest
compared
to
those
Com1 = 2,1 , 3,1 = [1, 1]
determine Emax,k and Emin,k
and
along
the
other
axes,
Com 2 = 2,2 , 3,2 = [3, 1]
we
into
substitute
(3.11)
to
along the x1 -axis. As shown in Fig. 3.5, the
possible value set 1,1 along the x1 -axis is {1} for Com1 and 1,2 is {5} for
Com 2 , so the point x1,1 = [1 1 1]
and the point x1,2 = [5 3 1]
are
obtained. By calculating their corresponding Euclidean distance, it can be

concluded that the point x1,2 that has a smaller Euclidean distance is taken as the
final solution.
47
Figure 3.5 3-D lattice space example.
3.4 Simulation Results

The ESA algorithm for MIMO systems has been briefly introduced. It
contains three main steps: Firstly, determine the hyper ellipsoid. Secondly, find
out the probable value sets for each component of the lattice point that is located
inside the hyper ellipsoid. Finally, search for the ML solution. In the first step,
either ZF decoding or MMSE decoding can be selected for determining the hyper
ellipsoid. In the second step, we firstly determine a loose boundary for each
component of the lattice points that may be located in the hyper ellipsoid. Then,
by further shrinking the value set of the M T -th component, all the redundant
points can be discarded and the lattice points inside the hyper ellipsoid are exactly
detected.
Since the ESA algorithm strictly sticks to ML decoding, it can achieve the
same performance with ML decoding. The ML decoding searches the entire lattice
48
points while the ESA algorithm only searches a small subset. The ESA algorithm
is assessed by means of the simulation results of the error rate performance. In the
simulations, we used 4-QAM, 16-QAM, 64-QAM in Rayleigh flat fading
channels with i.i.d. complex zero-mean Guassian noise. Fig. 3.6 illustrates the
BLER performance of ESA compared with ML decoding and ZF decoding using
4-QAM. Fig. 3.7 shows the BLER performance of ESA compared with ML
decoding ZF decoding using 16-QAM. And Fig. 3.8 shows the BLER
performance of ESA compared with ML decoding and ZF decoding using
64-QAM. It can be seen that the performances of ESA can achieve that of ML
decoding and are much better than ZF decoding.
Table 3.1 compares the complexity of ML decoding and ESA. The numbers
of lattice points visited by ML decoding and ESA for transmitting 16-QAM and
64-QAM constellations in 2 2 to 4 4 MIMO systems are indicated. It can be
observed that compared with the ML decoding, the number of lattice points
visited by the ESA is substantially reduced from 95.7% to 99.8%. The more
number of antennas and the higher level of modulation the system applies, the
greater complexity reduction the ESA can achieve.
49
(a)
(b)
Figure 3.6 Comparison of BLER performance of ESA, ML decoding and ZF
using 4-QAM.
(a) 4 4 MIMO systems. (b) 6 6 MIMO systems.
50
(a)
(b)
using 16-QAM.
51

using 64-QAM in 4 4 MIMO systems.
Table 3.1 Number of lattice points visited using ML decoding/proposed GD

(reduction is indicated as percentage)
Modulation
22
33
44
16-QAM
256/11
95.7%
4096/20
99.5%
4096/69
98.3%
262144/647
99.7%
65536/265
99.6%
16777216/39361
99.8%
64-QAM
3.5 Summary
In this chapter, the geometrical analysis of signal decoding for MIMO
channels is presented. The ellipsoid searching decoding algorithm is introduced in
detail. It is an add-on to the standard suboptimal detection schemes. Simulation
52
results demonstrate that it can provide the optimum performance and higher
diversity gain compared to the standard suboptimal detection schemes, and has
considerable reduction in complexity compared with ML decoding.
53
Chapter 4 MIMO Detection Algorithms Based on Semidefinite Relaxation
CHAPTER 4
MIMO DETECTION ALGORITHMS BASED ON
SEMIDEFINITE RELAXATION
4.1 Introduction
In this chapter, the background of the convex optimization problems and the
concept of the semidefinite relaxation programming will be introduced. Then the
semidefinite relaxation methods for MIMO detection for both low-order
modulation system and high-order modulation systems will be presented in details.
The tightness of the semidefinite relaxation detectors will be compared and their
performances will be illustrated. Finally, the SDR-initiated sphere detectors will
be proposed together with the simulation results.
4.2 Convex Optimization Problems

Convex optimization [49] is an important branch of the mathematical
optimization, which has been studied for over decades. It is a significant
numerical tool for system analysis and design. For example, convex optimization
has been widely used in the field of communications and signal processing. The
corresponding problems can either be cast as or be converted into convex
optimization problems so that the original difficult non-convex problems can be
analyzed and solved by the simpler convex optimization techniques.
In order to formulate the convex optimization problems for communications
and signal processing applications, in this section we will provide the basic
54
concepts of convexity and convex optimization models, especially the most

commonly used semidefinite programming.
With no loss of generality, an optimization problem is usually defined in the
form as [49]:
minn f 0 ( x )
(4.1)
fi ( x )
(4.2)
xR
st.
bi , i = 1, , m
where f 0 ( x ) is the objective function, x is an n dimensional vector of variables;

fi ( x )
bi are the constraint functions and
represents = or or . Denoting
x as the optimal solution of the problem (4.1-4.2), then f 0 ( x ) is the smallest

objective value for any x R n with all the constraints fi ( x )
bi , i = 1, , m
being satisfied.
The optimization problems can be classified into three types according to the
objective function and constraint functions: the linear optimization problems, the
nonlinear optimization problems and the convex optimization problems, wherein
the objective function and constraint functions are linear, nonlinear and convex
respectively. This section will focus on the convex optimization problems.
So, what is convex, and what is a convex function? Firstly, a set S R n is
convex if it contains the straight line segment between any two distinct points in S:
x+ yS
(4.3)
where x, y S , , R , , 0 and + = 1 [49]. Fig. 4.1 illustrates the

examples of convex and non-convex set. Fig. 4.1(a) shows an octagon and its
inside region including the boundary constitute a convex set. However, Fig. 4.1 (b)
is a non-convex set because there are straight lines segments between the two
points in the set do not lie in the set [49].
55
(a)
Figure 4.1
(b)
(a) Convex set. (b) Non-convex set.
Secondly, the function f ( x ) which satisfies:

f ( x + y ) f ( x ) + f ( y )
(4.4)
is said to be a convex function, for x, y dom f , , R , + = 1 and
, 0 , where dom f represents the domain of the function f ( x ) and dom f

should be a convex set. The straight line segment joining
( x, f ( x ) )
and
( y, f ( y ) ) lies above the graph of f ( x ) . A function f ( x ) is termed as strictly

convex function if it satisfies:
f ( x + y ) < f ( x ) + f ( y )
(4.5)
for x, y dom f , , R , + = 1 , and , > 0 . A function f ( x ) is

considered as a concave function if - f ( x ) is convex.
56
(a) f ( x )
Figure 4.2
(b) - f ( x )
(a) Convex function. (b) Concave function.
Figure 4.2 shows the examples of convex function and concave function [49].
It can be seen that the line segment joining ( x, f ( x ) ) and ( y, f ( y ) ) lies above
the graph of the function f ( x ) , so f ( x ) is convex and of course - f ( x ) is
concave.
Thirdly, a generalized convex optimization problem can be formed as [49]:
minn f 0 ( x )
xR
st.
fi ( x ) 0, i = 1, , m1
(4.6)
hi ( x ) = 0, i = 1, , m2
xS
where the objective function f 0 ( x ) and the inequality constraints fi ( x ) 0 ,

i = 1, , m1 are convex functions; the equality constraints hi ( x ) =0 , i = 1, , m2
are affine functions; the constraint set S is a convex set. If any one of the above
conditions is not satisfied, the problem is a non-convex problem.
The optimization variable x S is feasible if it satisfies the constraints
functions fi ( x ) 0 and hi ( x ) =0 . If there exists a feasible variable x for the
problem (4.6), it is called a feasible problem. Otherwise, the problem is an
infeasible one. The set which contains all the feasible variables x is called the
57
feasible set. A feasible solution x is called globally optimal solution of the

problem (4.6) if it satisfies f 0 ( x ) f 0 ( x ) for all the feasible solution x [49].
Then, the optimal value f 0 ( x ) is attained and the problem is solvable. In case
that there exists some > 0 such that f 0 ( x + ) f 0 ( x ) for all feasible x satisfying
x + x , a feasible vector x + is called locally optimal solution [49]. For

convex optimization problem, if it has any locally optimal solution, it also has
globally optimal solution.
Convex optimization problems have been studied for decades, since they have
many advantages compared with the non-convex optimization problems. Firstly,
the convex optimization problem can be solved with high efficiency by using the
well developed methods such as interior-point method, whose worst-case
complexity grows only polynomially with the problem size. The second
advantage of the convex optimization is that if there exists a global optimal
solution, the problem can definitely be solved to obtain the optimal solution.
While for non-convex optimization problems, the global optimal solution is
extremely difficult to be determined. Last but not least, when there are several
objective functions in the optimization, they can be combined easily in terms of
linear and convex constraints.
For solving a system of equations, the linearity is a criterion for judging
whether the problem is easy or difficult. If the problem consists of a set of linear
equations, it is considered to be solvable analytically or using existing numerical
software [90]. For the problem of a system of nonlinear equations, it is usually
considered as a hard problem because there is no efficient and reliable solver for
nonlinear problems. While for optimization problems, the criterion of judging
58
whether the problem is easy or difficult is by the convexity of the problem. As

mentioned above, the convex optimization problems can be solved by the efficient
interior-point method. However, we always have many troubles handling the nonconvex optimization problems. Thus, converting or relaxing a difficult nonconvex problem into a convex problem is the key point for solving it efficiently.
4.3 Semidefinite Relaxation

4.3.1 Semidefinite Matrix
A set of n n symmetric matrices can be denoted by S n , where S S n . If all

the eigenvalues of a symmetric matrix are nonnegative, the matrix S is called
positive semidefinite matrix, viz. S
0 . In the case that S
0 , S 0 , S 0 , the
matrix S is called positive definite matrix, negative definite matrix and negative
semidefinite matrix, respectively.
The characteristics of the positive semidefinite matrix S are as follows:
1) S is positive semidefinite: S
0.
2) All the eigenvalues of S are nonnegative: i ( S ) 0 , i = 1, 2,
,n.
3) The principal submatrices of S are nonnegative: det ( Si ) 0 , i = 1, 2,
,n.
4) S, T 0 , for all T S n , where i denotes the inner product.

5) There exists a matrix A
mn
where rank ( A ) = rank ( S ) , so that
S =A T A .
4.3.2 Semidefinite Cone
A convex cone is a kind of convex set which is closed under positive scaling.
Let S +n denote a set of n n positive semidefinite matrices, then the set S +n of
59
particular dimension is called the convex semidefinite cone. The cone defines a
partial order for X, Y S n by X
Y.
4.3.3 Semidefinite Programming
Semidefinite programming (SDP) [89-90] is a subfield of the convex

optimization problem. It minimizes a linear objective function, and its constraints
are an affine combination of positive semidefinite matrices. It has attracted great
interest because of the reasons listed below.
Firstly, many engineering problems and combinatorial optimization problems
such as linear programming and quadratically constrained quadratic programming
(QCQP) can be formulated into semidefinite programming problems, so that the
semidefinite programming can provide a way to investigate the characteristics of a
large number of convex optimization problems.
Secondly, the semidefinite programming problems can be solved very readily
and efficiently by using the interior-point methods such as Sedumi.
Consider a linear function of variable x
, the semidefinite programming
problem is usually given by:
min cT x
(4.7)
n
st.
F ( x ) = F0 + xi Fi 0
(4.8)
AT x = a
(4.9)
i =1
where Fi S +n , i = 1, 2, , n , A
jn
. The inequality constraint F ( x ) 0 is a
linear matrix inequality, which indicates that F ( x ) is positive semidefinite. The

problem (4.7-4.9) is a convex optimization problem because cT x in (4.7) is a
60
linear function and the constraint functions (4.8-4.9) are convex functions. That
means that if F ( x ) 0 , F ( y ) 0 , then:
F ( x + y ) = F ( x ) + F ( y ) 0
(4.10)
where 0 , 1 and + =1 .
The semidefinite programming problems also have another form which is
give by:
st.
min cT Xc = Tr ( CX )
(4.11)
Tr ( A i X ) = ai , i = 1, 2, , n
(4.12)
(4.13)
where Tr ( i ) is the trace of the matrix, Ai is the i-th row of A and X is a

positive semidefinite matrix.
4.3.4 Interior-Point Methods

The interior-point methods (IPM) were firstly introduced to solve linear
programming problems in 1984 [54]. Recent developed IPM [55-59] for linear
programming problems has been developed to deal with the semidefinite
programming problems. The IPM has several advantages which attract great
attentions. They are very effective both in practices and theory. It has been proved
that the IPM is comparable with other simplex methods for the linear
programming problems with small size. Moreover, it is faster than other methods
for linear programming problems with more than 10000 constraints [60].
Similarly, the IPM is comparable with other methods for small-scale semidefinite
programming problems. Moreover, it is also more effective than other approaches
for medium to large problems. Theoretically, the worst-case complexity of IPM
61
for semidefinite programming problems increases no faster than a polynomial of

the problem size. What's more, the IPM for semidefinite programming problems
is usually solved by iterative methods [89], which can exploit the problem
structure.
There are a lot of toolboxes which are based on the IPM, such as CSDP,
Sedumi, SDPT3, DSDP, and SDPA. The Sedumi [35] runs on MATLAB and uses
the self-dual method for solving the general convex optimization problems. In this
thesis, when solving the SDP problems, we just invoke the Sedumi toolbox.
4.4 Semidefinite Relaxation Detection Algorithms for Low-Order

Modulation Systems
Since the semidefinite programming based algorithm is very effective in
dealing with practical engineering problems, the SDP techniques applied for
MIMO detection have received great attentions. As mentioned in Chapter 1, the
MIMO detection is to find out the transmitted vector from the received vector.
Most of the recent semidefinite relaxation (SDR) detectors have focused on the
SDP relaxation of the maximum likelihood detection, which is optimal in the
minimum error rate performance. The SDR detectors are solved and then an
approximation solution is obtained by the rounding procedure.
The ML decoding is a discrete least square problem:
x ml = arg min r Hx
(4.14)
where r
MR
, x
MT
, H
M R M T
In the following subsections, the SDR detectors for MIMO systems with
different level of modulations will be elaborated.
62
4.4.1 SDR Detection for 4-QAM

(BPSK) and quadrature amplitude modulation (4-QAM) signals in [36] and [37].
For BPSK and real valued 4-QAM, the signal vector space is = {1, +1} . The
problem (4.14) is not a homogeneous QCQP problem, so we can rewrite it to be
the following homogenized problem:
min r - Hx
st.
(4.15)
xi = 1 , i = 1, 2, , M T
HT H HT r
By introducing the matrix Q = T
, the problem can be
T
r
H
r
r
expressed as:
min x
Define
by: X= xT
HT H HT r x
1 T

T
r H r r 1
st.
xi 2 = 1, i = 1, 2, , M T
rank-one
1 xT
semidefinite
matrix,
(4.16)
which
is
given
1 , the objective function (4.16) can be transformed into:

min xT
H T H HT r x
1 T
=min Tr ( QX )
T
r H r r 1
(4.17)
Since xi equals either 1 or -1, and the matrix X is a positive semidefinite matrix,
the rank of X equals one and its diagonal entries are equal to 1. Assuming the
matrix X satisfies the above three characterising properties, so that the problem
becomes:
min Tr ( QX )
st.
diag ( X ) = e MT +1
(4.18)
rank ( X ) = 1
63
where the diag ( i ) means the diagonal elements, and e represents all ones vector
of M T dimension. The problem is a non-convex optimization problem because of
the rank-one constraint function, so we eliminate it to yield the basic semidefinite
relaxation [50] as:
min Tr ( QX )
st.
diag ( X ) = e MT +1
(4.19)
Now, the problem (4.19) becomes a semidefinite relaxation problem concerning

the matrix variable X , since its objective function is a linear function concerning
X , and one of its constraint functions is a linear function of X while the other is
semidefinite constraint. Then the optimal solution X of the relaxed problem

(4.19) can be solved by using the IPM. Due to the fact that the rank-one constraint
is dropped, the optimal solution X may not be a feasible one of the original
problem (4.18). Thus, it is necessary to convert the optimal solution X into a
feasible solution x of problem (4.18). If the matrix X has the rank of one, then
the feasible solution x=x* , where X =x*x*T . If the rank of matrix X is larger
than one, the feasible solution x is generally not an optimal solution and there are
some methods [51] such as dominant eigenvector approximation, randomization
or rank-one approximation to convert x* of problem (4.19) into a feasible solution
x of problem (4.18).
1)
Dominant eigenvector approximation
64
Let X = i qi qiT
be the eigenvalue decomposition of X , where
1 2 M
+1
are the eigenvalues and q = q1 , q2 , , qM T +1 are the
corresponding eigenvectors.
The feasible solution is :
x = Q qM T +1 q1 , q2 , , qM T
(4.20)
where Q ( i ) means the quantization to the signal vector set .
2)
Randomization
Firstly, the optimal solution X is Cholesky factorized to be X =FT F . Then,
generate several random M T +1 dimensional vectors u i , for i = 1, 2,
, N rand ,
which is uniformly distributed on an M T +1 dimensional unit sphere. Next,

compute
x i = Q ( FT ui ) , i = 1, 2,
l = arg max
i =1,2, , N rand
, N rand ,
and
choose
x = x l ,
where
x iT Qx i .
Consequently, the feasible solution is given by:

x = xM T +1 x1 , x2 ,..., xM T
3)
(4.21)
Rank-one approximation
Firstly, the optimal solution X is partitioned into
X
X = 1,1T
X 1,2
X1,2
X2,2
(4.22)
where X1,1 is a M T M T dimensional matrix, X1,2 is M T 1 dimensional vector

and X2,2 is a number.
Then, feasible solution can be given by:
65
x = Q ( X1,2 / X2,2 )
(4.23)
4.4.2 SDR Detection for 16-QAM

Define a rank-one semidefinite matrix, which is given by:
X = [ xT
1]T [xT
(4.24)
1]
It can be partitioned as:

X1,1
X= T
X1,2
X1,2
X 2,2
(4.25)
where X1,1 is a M T M T dimensional matrix, X1,2 is M T dimensional vector and

X 2,2 is a real number.
It can be observed that the matrix X possesses the following properties:

X =XT
(4.26)
X1,1 = X1,2 X1,2T
(4.27)
X1,2 =x {1
3}
MT
X 2,2 = 1
(4.28)
(4.29)
So the ML detection problem for 16-QAM can be rewritten as:
st.
T
T
H H H r
min Tr X T
T
r H r r
(4.30)
X =X T
(4.31)
X1,1 = X1,2 X1,2T
(4.32)
diag ( X1,1 ) {1,9}
(4.33)
X 2,2 = 1
(4.34)
However, this problem is still not a convex problem due to the non-convex
constraints diag ( X1,1 ) {1,9} and X1,1 = X1,2 X1,2T . Thus, relaxation of these
66
constraints will be engaged to transform the original problem into a semidefinite

problem, which can then be efficiently solved in polynomial time.
In 4-QAM MIMO systems, the semidefinite relaxation approach is to
eliminate the rank-one constraint. However, in the case of 16-QAM, the relaxed
problem still appears to be hard to solve because of the constraint function
diag ( X1,1 ) {1,9} . In the following, three methods: PI-SDR [40], BC-SDR [41]
and VA-SDR [42] for dealing with this constraint will be introduced.
4.4.2.1 Polynomial-Inspired SDR (PI-SDR)

The constraint xi {1, 3} is equivalent to the following equation:
( xi 1)( xi + 1)( xi 3)( xi + 3) = 0 , i = 1, 2,
, MT
(4.35)
This equation is equivalent to:
(x
1)( xi 2 9 ) = 0 , i = 1, 2,
, MT
(4.36)
By introducing ti = xi 2 , the equation becomes:
ti 2 10ti + 9 = 0 , i = 1, 2,
, MT
(4.37)
Define a rank-one matrix :

V = vvT
where vT = xT
tT
(4.38)
1 . It can be partitioned as:
V1,1
V = V1,2T
V1,3T
V1,2
V2,2
V
T
2,3
V1,3
V2,3
V3,3
(4.39)
where V1,1 , V1,2 and V2,2 are all M T M T dimensional matrixes, V1,3 and V2,3 are
M T dimensional vectors and V3,3 is a number.

It should be noted that the diagonal element of V1,1 are xi2 and V2,3 is ti for
i = 1, 2,
, M T . So the equation (4.37) is equivalent to diag ( V1,1 ) 10V2,3 + 9e = 0 ;

67
the equation ti = xi 2 is the same as diag ( V1,1 ) V2,3 = 0 . Then, (4.31) along with
(4.32) can be relaxed into V
0 and rank ( V ) = 1 .
Thus, the ML problem becomes:
st.
HT H 0 HT y

min Tr V 0
0
0
yT H 0 yT y
(4.40)
diag ( V1,1 ) V2,3 = 0
(4.41)
diag ( V1,1 ) 10V2,3 + 9e = 0
(4.42)
V3,3 = 1
(4.43)
(4.44)
rank ( V ) = 1
(4.45)
Dropping the non convex rank-one constraint can yield the SDR problem:
st.
HT H 0 HT y

min Tr V 0
0
0
yT H 0 yT y
(4.46)
diag ( V1,1 ) V2,3 = 0
(4.47)
diag ( V1,1 ) 10V2,3 + 9e = 0
(4.48)
V3,3 = 1
(4.49)
(4.50)
Then the optimal solution V can be solved by the IPM and the approximation
is the final
procedure. The first 2M T elements of the last row in the solution V
solution x .
The PI-SDR precisely transforms the alphabet constraint (4.33) into linear
constraints by a higher dimensional matrix, so this is an increased-dimension
relaxation.
68
4.4.2.2 Bound-Constrained SDR (BC-SDR)

Firstly, the constraint (4.33) implies 1 xi2 9 , where xi denotes the i-th
component of x . Please be noted that the diagonal elements of X1,1 are xi2 , for
i = 1, 2, , M T , thus, it satisfies that e diag{Y1,1} 9e . Also (4.31) along with

(4.32) are relaxed to be Y
0 and rank ( Y ) = 1 . Herein, a new symbol Y is
introduced to distinguish from the aforementioned X in (4.24) since they are

actually different matrixes after relaxation. Consequently, the BC-SDR problem is
obtained as:
st.
HT H HT r
min Tr Y T
T
r H r r
(4.51)
(4.52)
e diag{Y1,1} 9e
(4.53)
Y2,2 = 1
(4.54)
rank ( Y ) = 1
(4.55)
The problem is a non-convex optimization problem due to the rank-one

constraint function which is dropped to become a convex problem:
st.
T
T
H H H r
min Tr Y T
T
r H r r
(4.56)
(4.57)
e diag{Y1,1} 9e
Y2,2 = 1
(4.58)
(4.59)
This BC-SDR problem can then be solved by any of the SDP solvers, such as
Sedumi, based on interior-point methods. Although (4.57) could be deduced from
(4.31) and (4.32), and (4.58) can also be deduced from (4.33), the BC-SDR
problem is however not exactly equivalent to the ML detection problem. Thus, the
69
solution obtained from solving BC-SDR has more errors than ML solution. Once
the solution is found, the approximation procedure approach is applied to quantize
till the constraint (4.33) is satisfied.
the resulting Y
1,2
This SDR method transfers alphabet constraint (4.33) into the linear
constraints in the same dimensional space, so it is a fixed-dimension relaxation.
4.4.2.3 Virtually-Antipodal SDR (VA-SDR)

In this method, a new matrix U is introduced to express the constraint
x {1
3}
MT
. The signal x could be expressed as:

x = UpT
where U = [ I 2I ] , p = [p1 p 2 ] , I
(4.60)
MT M T
and pi {1}
MT
, i = 1, 2,3, 4 .
Substituting (4.60) into (4.24), and defining a matrix Z , which is given by:
Z = [p 1]T [p 1]
(4.61)
the objective function (4.30) can be equivalently transformed into (4.62), and the
constraints (4.33) and (4.34) are relaxed into (4.64). Similarly, (4.31) along with
(4.32) can be relaxed into (4.63). Thus, we obtain the VA-SDR problem given by:
st.
T
T
T
T
U H HU U H r
min Tr Z
T
r Tr
r HU
(4.62)
(4.63)
diag ( Z ) = 1e
(4.64)
rank ( Z ) = 1
(4.65)
Then, drop the rank-one constraint to yield the SDR problem:
st.
T
T
T
T
U H HU U H r
min Tr Z
T
r Tr
r HU
(4.66)
(4.67)
70
diag ( Z ) = 1e
(4.68)
can be solved by the IPM. After the

Next, the optimal solution Z
approximation procedure, the approximated solution of the problem is z , the final
solution is conducted by using:
x = Uz
(4.69)
The VA-SDR method transfers the alphabet constraint (4.33) into the linear
constraints in the same dimensional space. It is also a fixed-dimension relaxation.
4.4.3 Simulation Results

Simulations were conducted to evaluate the performance of these three SDR
detectors. An uncoded MIMO system with independent Rayleigh fading channel
was taken into account and the Sedumi toolbox within Matlab software was used
to implement the SDR detection algorithms.
Fig. 4.3 compares these three approximation procedures which convert the
SDR solution x to a feasible solution x* . The approximation includes dominant
eigenvector approximation, randomization ( N rand =10 ), randomization ( N rand =30 )
and rank-one approximation. The BLER performances of the BC-SDR combined
with the three approximation procedures show that the randomization performs
better than the dominant eigenvector approximation and the rank-one
approximation. Whats more, the randomization ( N rand =30 ) outperforms the
randomization ( N rand =10 ). Thus, in the following simulation, we apply the
randomization ( N rand =30 ) as the SDR approximation procedure.
71
Figure 4.3
Comparison of the three approximation procedures.
We conducted the simulation for two typical MIMO transmission scenarios,

namely, 4 4 and 8 8 16-QAM systems. Fig. 4.4 shows the BLER
performances of the three SDR detectors, ML decoding and zero-forcing decoding,
respectively. It can be observed that the PI-SDR, the BC-SDR detector and the
VA-SDR detector provide exactly the same BLER performance. Although the
SDR detectors are suboptimum, it provides a significant performance
improvement over the ZF decoding.
72
Figure 4.4
Comparison of BLER performance of SDR detectors, ML
decoding, ZF using 4 4 16-QAM MIMO systems.
(b)
Figure 4.5
Comparison of BLER performance of SDR detectors, ML
decoding, ZF using 8 8 16-QAM MIMO systems.
73
4.5 Semidefinite Relaxation Detection Algorithms for High-Order

Modulation Systems
4.5.1 Challenges and Motivation
To enable high data rate in the future wireless communication systems, the
ability to utilize the high-order modulation is in great need. The sphere decoding
(SD) can provide optimum BLER performance at the low-order signal
constellations or small system dimensions. However, SD cannot efficiently handle
high-order symbol constellations or high problem dimensions, especially at low
SNR. In contrast, the decoding algorithms based on semidefinite relaxation (SDR)
approach have become more and more attractive simply because of the fact that
SDR problems can be solved very efficiently even for high-order symbol
constellations or high problem dimensions.
(BPSK) and quadrature amplitude modulation (4-QAM) signals. Then the
extensions to different SDR techniques for 16-QAM signals had been proposed,
all exhibit acceptable BLER performance and relatively low complexity. In [43],
it has been proved that there exists equivalence among PI-SDR, BC-SDR and
VA-SDR for 16-QAM. However, due to its high complexity, the PI-SDR is not
suitable for extension to 256-QAM system. In the following subsections, the
extension of BC-SDR and VA-SDR to 256-QAM has been investigated, a new
SDR detector is proposed for 256-QAM system. Then, a comprehensive
comparison between the proposed method and the previous SDR detectors is
made.
74
4.5.2 Extension of existing SDR Detectors to 256-QAM Systems
4.5.2.1 Extended BC-SDR

It is easy to find that the ML detection problem using 256-QAM constellation
can be rewritten as:
st.
HT H HT r
min Tr Y T
T
r H r r
(4.70)
Y = YT
(4.71)
Y1,1 = Y1,2 Y1,2T
(4.72)
Y1,2 =xi {1, 3, 5, 7, 9, 11, 13, 15}
MT
Y2,2 = 1
(4.73)
(4.74)
It can be observed that the high complexity of the ML detection is due to the
presence of the two non-convex constraints (4.72) and (4.73). Thus, relaxation of
these constraints will be engaged to transform the original problem into a SDR
problem, which can then be efficiently solved in polynomial time. Firstly, the
constraint (4.73) implies 1 xi2 225 , where xi denotes the i-th component of x .
Please be noted that the diagonal elements of Y1,1 are xi2 , for i = 1, 2, , M T , thus,
it satisfies that e diag{Y1,1} 225e . Secondly, the constraints (4.71) and (4.72) can
be relaxed into Y 0 . Consequently, the BC-SDR problem is obtained as:
st.
T
T
H H H r
min Tr Y T
T
r H r r
(4.75)
(4.76)
e diag{Y1,1} 225e
(4.77)
Y2,2 = 1
(4.78)
75
The BC-SDR problem can then be solved by any of the SDP solvers. Once
the solution is found, the randomization approach is applied to quantize the
till the constraint (4.73) is satisfied.
resulting Y
1,2
4.5.2.2 Extended VA-SDR

It is worth noting that when the constraint (4.73) is expected to be satisfied,
the signal xi {15, 13, 11, 9, 7, 5, 3, 1} could be expressed as:
x = UpT
where U = [ I 2I 4I 8I ] , p = [p1 p 2 p3 p 4 ] , I
(4.79)
M T M T
and pi {1}
MT
i = 1, 2,3, 4 .
Define a matrix Z , which is given by:

Z = [p 1]T [p 1]
(4.80)
By substituting (4.79) and (4.80) into (4.30), the objective function becomes
T
T
T
T
U H HU U H r
min Tr Z
. Finally, we obtain the VA-SDR problem
T
r Tr
r HU
given by:
st.
U T H T HU U T H Tr
min Tr Z
T
r Tr
r HU
(4.81)
(4.82)
diag ( Z ) = 1e
(4.83)
The approximate solution of the problem z is solved by the IPM together

with the approximation procedure. Then the optimum solution x is reconstructed
by:
76
x = Uz
(4.84)
4.5.3 Proposed SDR Detector for 256-QAM Systems
As introduced in section 4.5.2, one of the key issue for deducing the
corresponding SDR detector lies in how to deal with the non-convex constraint of
the alphabet set xi {15, 13, 11, 9, 7, 5, 3, 1} . In the extended BC-SDR,
the alphabet set is directly relaxed into 1 xi2 225 . This is reasonable for the
low-order constellations, such as 16-QAM. However, for 256-QAM system, it
will inevitably cause considerable errors. In the extended VA-SDR, the alphabet
set is reformed by expanding the dimension of the problem into four times of its
original size. This will result in the significant increase of computational
complexity. In order to keep the relaxation as tight as possible, while the
complexity as low as possible, a novel method to deal with the alphabet set is
proposed.
Actually, the signal xi {15, 13, 11, 9, 7, 5, 3, 1} in 256-QAM
constellations could also be expressed as:
x = VqT
where V = [ I 4I ] ,
q = [q1 q 2 ] , I
M T M T
(4.85)
and q1 , q 2 {1, 3}M .

T
Table 4.1 The values of x
77
Table 4.1 gives the values of xi for the possible combinations of q1i and q2i ,
where xi , q1i and q2i denote the j-th element of x , q1 and q 2 , respectively.
By substituting (4.85) into (4.24), and defining a matrix W , which is given by:
W = [q 1]T [q 1]
(4.86)
the objective function (4.30) can be equivalently transformed into (4.87).

Moreover,
it
can
be
known
from
(4.85)
that
the
constraint
xi {15, 13, 11, 9, 7, 5, 3, 1} is equivalent to q1 , q 2 {1, 3}M , which are

T
essentially the well known indices used to characterize the 16-QAM constellation.
Herein, the set operation method [44] is engaged to formulate the alphabet
constraint (4.73) into (4.88) and (4.89). Similarly, (4.71) along with (4.72) can be
relaxed into (4.90). Also (4.74) can be reformulated as (4.91). Thus, we obtain the
proposed SDR problem given by:
(4.87)
1e diag {W1,1} 9e
(4.88)
st.
V T H T HV V T H Tr
T
r Tr
r HV
min Tr W
diag {W1,1} 4diag {W1,2 } + 3e
W2,2 = 1
0,
(4.89)
(4.90)
(4.91)
is then
The problem can be solved by using IPM. The optimal solution W
. The final solution x

approximated by the approximation procedure to obtain w
is reconstructed by:
x = Vw
(4.92)
4.5.4 Comparison of Tightness
78
As mentioned above, these three SDR problems are all relaxed from the
original ML problem, and their objective functions are equivalent, which are
actually calculating the Euclidean distance given by (4.14). Thus, the tightness of
the constraints of each SDR algorithm implies how close it is to the ML decoding.
In what follows, we will compare their tightness.
4.5.4.1 Equivalence of BC-SDR and VA-SDR

Firstly, we will demonstrate that the constraints of the BC-SDR problem are
equivalent to those of the VA-SDR problem [43]. The proof is organized in the
following two steps:
Step 1 For each matrix Z
( 4 M T +1)( 4 M T +1)
(4.83), there must be a matrix Y
that satisfies the constraints (4.82)-
( M T +1)( M T +1)
which satisfies the constraints
(4.76)-(4.78).
Proof: For any matrix Z which satisfies the constraints (4.82)-(4.83), it has
the following form:
Z
Z = 1,1
Z 2,1
4 M T 4 M T
1 4 M T
Z1,2
1
4 M T 1
(4.93)
Since Z 0 , there should be a reversible matrix:

= 1 , 2 ,
M T , MT +1 , MT + 2 ,
2 MT , 2 M T +1 , 2 M T + 2 ,
, 3 M T , 3 M T +1 , 3 M T + 2 ,
4 M T , 4 M T +1
(4.94)
that satisfies:
Z = T
where m
( 4 M T +1)1
(4.95)
, m = 1, 2, , 4 M T + 1 .
79

( M T +1)( M T +1)
Now, we construct a semidefinite matrix Y

Y = GZG T = ( G T )
( G )
I
where G =
=
0 1 0
U 0
MT MT
1 M T
2I
0
which is defined as:
MT MT
4I
1 M T
(4.96)
MT MT
8I
1 M T
MT MT
M T 1
1 M T
From (4.93)-(4.95) and (4.83), we get:

n = ( Tn n )
1/2
= 1 , n = 1, 2,
, 4M T
(4.97)
T4 M T +1 4 M T +1 = 1
(4.98)
Substituting (4.94) into (4.96) gives:

Y1,1 Y1,2
Y=
Y2,1 Y2,2
( 1 + 22 + 43 + 84 )T ( 1 + 22 + 43 + 84 )
=
T
T4 MT +1 ( 1 + 22 + 43 + 84 )
( 1 + 22 + 43 + 84 )
T4MT +1 4MT +1
4 MT +1
(4.99)
( 4 M T +1) M T
where 1 = 1 , 2 ,
Nt
2 = MT +1 , MT + 2 ,
2 MT
( 4 M T +1) M T
3 = 2 Nt +1 , 2 Nt + 2 ,
3 Nt
( 4 M T +1) M T
4 = 3 Nt +1 , 3 Nt + 2 ,
4 Nt
( 4 M T +1) M T
,
,
,
.
From (4.98) and (4.99) it can be known that Y2,2 = 1 , and (4.78) is satisfied.
Moreover, we have:
Y1,1 = ( 1 + 2 2 + 43 + 8 4 )
( 1 + 22 + 43 + 8 4 )
(4.100)
and the element located in the j-th row and the j-th column of Y1,1 is:
y jj = j + 2 M T + j + 4 2 MT + j + 8 3 M T + j
(4.101)
80
From the perspective of geometry and (4.97), it is easy to know that:

1 = 8 3 M T + j j 2 MT + j 4 2 MT + j j + 2 M T + j + 4 2 MT + j + 8 3 M T + j
j + 2 M T + j + 4 2 M T + j + 8 3 MT + j = 15
(4.102)
Thus we have:
1 y jj 225
(4.103)
1e diag{Y1,1} 225e
(4.104)
The proof is complete.
Step 2 For each matrix Y

there must be a matrix Z
( M T +1)( M T +1)
that satisfies the constraints (4.76)-(4.78),
( 4 M T +1)( 4 M T +1)
which satisfies the constraints (4.82)-
(4.83).
Proof: For any matrix Y that satisfies the constraints (4.76)-(4.78), it has the
following form:
Y
Y = 1,1
Y2,1
MT M T
1 M T
Y1,2
1
M T 1
(4.105)
Since Y 0 , there must be a reversible matrix = 1 , 2 , 3 ,
Nt , MT +1
that satisfies:
Y = T
where l
( M T +1)1
(4.106)
, l = 1, 2, , M T + 1 .
From (4.105)-(4.106) and (4.77), it can be obtained that:
M T +1 = TM T +1 M T +1
1 j = ( Tj j )
1/ 2
1/ 2
=1
15 , j = 1, 2,
(4.107)
, MT
(4.108)
81
From the perspective of geometry, it is easy to know that there should be vectors
j1 , j 2 , j 3 , j 4
( M T +1)1
which satisfy:
j = j1 + 2 j 2 + 4 j 3 + 8 j 4
(4.109)
j1 = j 2 = j 3 = j 4 = 1
(4.110)
Now we construct another matrix in the basis of (4.106) and (4.109). which is
given by:
= 11 , 12 , 13 , 14 , 21 , 22 , 23 , 24 ,
Nt 1 , Nt 2 , Nt 3 , Nt 4 , Nt +1
( M T +1)( 4 M T +1)
(4.111)
( 4 M T +1)( 4 M T +1)
Next, we construct a semidefinite matrix Z

Z = ( )
T

(4.112)
Substituting (4.111) into (4.112), the element located in the ( j i ) -th row and the
( j i ) -th column of Z is given as:

z( ji )( ji ) = ( Tji ji ) = ji
=1
(4.113)
where j = 1, 2, , M T , i = 1, 2,3, 4 . Moreover, the ( 4M T + 1) -th row and the
( 4M T + 1) -th column of Z is:
z( 4 M T +1)( 4 M T +1) = TM T +1 M T +1 = 1
(4.114)
From (4.113) and (4.114), it can be obtained that:

diag ( Z ) = 1e
(4.115)

From both step 1 and step 2, the equivalence of BC-SDR and VA-SDR can be
concluded.
82
4.5.4.2 Proposed SDR tighter than BC-SDR and VA-SDR

Secondly, we will demonstrate that the constraints of the proposed SDR
problem are tighter than those of the BC-SDR problem, and also tighter than those
of the VA-SDR due to the aforementioned equivalence. The proof consists of the
following two steps:
( 2 M T +1)( 2 M T +1)
Step 1 For each matrix W

there must be a matrix Y
( Nt +1)( Nt +1)
which satisfies the constraints (4.76)-(4.78).
Proof: For any matrix W that satisfies the constraints (4.88)-(4.91), it has the
following form:
W
W = 1,1
W2,1
2 M T 2 M T
W1,2
1 2 M T
2 M T 1
(4.116)
Since W 0 , there should be a reversible matrix:

= 1 , 2 ,
MT , MT +1 , MT + 2 ,
2 MT , 2 MT +1
(4.117)
that satisfies:
W = T
where s
( 2 M T +1)1
(4.118)
, s = 1, 2, , 2 M T + 1 .
From (4.116)-(4.118), (4.88) and (4.91), gives:

1 k = ( Tk k )
1/2
3 , k = 1, 2,
, 2M T
T2 M T +1 2 M T +1 = 1
(4.119)
(4.120)
Now we construct a semidefinite matrix Y

Y = EWET = ( ET )
( M T +1)( M T +1)
( E )
T

(4.121)
83
I
where E =
=
0 1 0
V 0
MT MT
MT MT
4I
1 M T
1 M T
M T 1
0
1
Substituting (4.118) into (4.120) gets:

T
Y1,2 ( 1 + 4 2 ) ( 1 + 4 2 )
=
T
Y2,2 T
2 M T +1 ( 1 + 4 2 )
Y1,1
Y=
Y2,1
( 1 + 4 2 )
2 MT +1
2 MT +1 2 M T +1
T
(4.122)
where
1 = 1 , 2 ,
M T
( 2 M T +1) M T
, 2 = M +1 , M
T
+2
2 MT
( 2 M T +1)M T
From (4.120) and (4.122) it can be known that Y2,2 = 1 , and (4.78) is satisfied.
Moreover, we have:
Y1,1 = ( 1 + 4 2 )
( 1 + 4 2 )
(4.123)
and the element located in the j-th row and the j-th column of Y1,1 is:
2
y jj = j + 4 M T + j , j = 1, 2,
, MT
(4.124)
From the perspective of geometry and (4.119), it is easy to know that:

1 4 M T + j j j + 4 M T + j j + 4 M T + j 15
(4.125)
Thus we have:
1 y jj 225
(4.126)
1e diag{Y1,1} 225e
(4.127)

Step 2 For each matrix Y
there must be a matrix W
( M T +1)( M T +1)
( 2 M T +1)( 2 M T +1)
which satisfies the constraints (4.88),
(4.90) and (4.91). However, this W may not satisfy the constraint (4.89).
84
Through some deduction process similar to that given in the step 2 of section
( 2 M T +1)( 2 M T +1)
4.5.4.1, it is easy to construct a matrix W
constraints (4.88), (4.90) and (4.91) in the basis of Y
which satisfies the
( M T +1)( M T +1)
. Next, we
consider a special example, let:

2 MT
2 MT
W = [1.5 1.5 1.5 1] [1.5 1.5 1.5 1]

T
(4.128)
It is easy to know that W 0 , diag {W1,1} = 2.25e and diag {W1,2 } = 1.5e , thus we
have:
diag {W1,1} 4diag {W1,2 } + 3e
= 2.25e 6e + 3e = 0.75e
(4.129)
Obviously, (4.129) being negative can not satisfy the constraint (4.89).
From both step 1 and step 2, it can be concluded that the constraints of the
proposed SDR are tighter than those of the BC-SDR, and also tighter than those of
the VA-SDR due to the aforementioned equivalence.
4.5.5 Comparison of Complexity

Firstly, the BC-SDR given in (4.75)-(4.78) consists of a ( M T + 1) ( M T + 1)
matrix variable Y and O ( M T + 1) linear constraints. Since the constraints (4.77)
and (4.78) are separable, the complexity of the BC-SDR detector is O ( M T + 1)3.5
[41].
Secondly,
the
VA-SDR
( 4 M T + 1) ( 4 M T + 1) matrix variable
given
in
(4.81)-(4.83)
involves
Z and O ( 4 M T + 1) linear constraints. The
constraint (4.83) are also separable, thus the complexity of the VA-SDR detector
is O ( 4 M T + 1)3.5 . Finally, the proposed SDR shown in (4.87)-(4.91) consists of a
( 2 M T + 1) ( 2 M T + 1) matrix variable
W and two O ( 2 M T + 1) linear constraints.
85
According to [44] and [52], it can be known that the complexity of the proposed
SDR detector is O ( 2 M T + 1)3.5 .
4.5.6 Simulation Results

Computer simulations were conducted to evaluate the performance of these
three SDR detectors. An uncoded MIMO system with independent Rayleigh
fading channel was taken into account and the Sedumi toolbox within Matlab
software was used to implement the SDR detection algorithms. Fig. 4.6 and Fig.
4.7 show the BLER performances of the three SDR detectors for 4 4 and 8 8
256-QAM systems, respectively. It can be observed that the BC-SDR detector and
the VA-SDR detector provide exactly the same BLER performance, while, the
proposed SDR detector can provide the best BLER performance among these
three detectors. This agrees well with the analysis presented in section 4.5.4. What
is more, it can be found that the improvement concerning the BLER performance
provided by the proposed SDR detector in the case of 8 8 MIMO system is
larger than that in the case of 4 4 MIMO system. The relaxation of the alphabet
set engaged in the BC-SDR and the VA-SDR will cause increasing errors with the
increase of the number of the antennas. That is because the inaccurate detection of
any dimension of the transmitted signal will result in the failure of the detection of
the whole signal vector, and the higher the dimension of the problem is, the bigger
the possibility of failure will be. The main merit of the proposed SDR over the
BC-SDR and the VA-SDR is that it can offer more accurate relaxation of the
alphabet set. Moreover, the computational times for solving one signal vector of
these SDR detectors are illustrated in Fig. 4.8 and Fig. 4.9. These figures
86
demonstrate that the BC-SDR has the lowest computational complexity and the
VA-SDR has the highest computational complexity while the proposed SDR is in
between the two detection methods. It is also found that the complexity of SDR
detectors is independent of SNR. This is a distinct advantage over the SD whose
complexity varies as a function of SNR, making SD difficult to be implemented in
practice.
Figure 4.6
The BLER performance of the SDR detectors for 4 4 256-QAM

MIMO systems.
87
Figure 4.7
The BLER performance of the SDR detectors for 8 8 256-QAM

MIMO systems.
Figure 4.8
The computational time of the SDR detectors for 4 4 256-QAM

MIMO systems.
88
Figure 4.9
The computational time of the SDR detectors for 8 8 256-QAM

MIMO systems.
4.6 SDR-initiated Sphere detector

Sphere decoding (SD) is able to provide optimum BLER performance with
less complexity than ML decoding by searching only a subset of the entire lattice
space within a hyper sphere. Nevertheless, it is still impractical when the
constellation size is large and the SNR is low. This is mainly caused by the large
initial radius of the hyper sphere which results in a very time-consuming
searching process. One of the most important issues of SD is the selection of the
initial radius d of the hyper sphere. If the radius is too large, the sphere contains
very large number of lattice points, and hence results in very high search
complexity. If the radius is too small, there may not be any lattice points in the
89
hyper sphere and the searching has to be restarted with a new initial radius [14,
53]. A traditional way to attain the initial radius is the ZF equalization. The radius
is the distance between the received point and the ZF equalized point. For the
MIMO systems with small constellation size, such as 8-QAM and 16-QAM, the
ZF-initialed SD has acceptable complexity. However, due to the poor BLER
performance of ZF or MMSE for cases with large constellation size, such as 256QAM, the complicated searching process becomes unacceptable.
The decoding algorithms based on SDR have become more and more
attractive simply because of the fact that SDR problems can be conveniently
solved in polynomial time. Although the SDR detectors can offer significantly
low computational complexity, as well as better BLER performance than ZF and
MMSE, their BLER performance is still worse than SD detector because of the
relaxation process.
The purpose of this section is to propose a new detection algorithm for 256QAM signals which combines the SDR with the SD. In this method, the SDR
algorithms are engaged to obtain a primary result. The feasible solution of the
SDR problem is proposed to be the initial point of SD. The radius becomes the
distance between the received point and the feasible point. Then, a hyper sphere is
constructed which is centered at the received signal and has its radius equals to the
Euclidean distance between the primary result and the received signal. Finally, the
SD searching strategy is employed to determine the final result which satisfies the
principle of maximum likelihood. This method can offer optimum BLER
performance as well as lower computational complexity than the traditional SD
detectors. Since the SDR detections have much better BLER performance than ZF
90
for 256-QAM MIMO systems, the radius given by SDR tends to be smaller. Thus,
the number of lattice points to be visited inside the sphere is smaller, which means
it can reduce the complexity of SD.
Matlab simulation has been used to assess the performances of the proposed
SDR-SD used in the 4 4 MIMO systems transmitting 256-QAM in a fading
channel. The BC-SDR, VA-SDR and the newly proposed SDR are separately
combined with SD. Their BLER performances are compared with the stand-alone
SD detector and SDR detectors and the results are shown in Figure 4.10. It can be
seen that the three SDR-SD detectors can offer the same BLER performance as
the SD detector, and have much better BLER performance than the stand-alone
SDR detectors. Figure 4.11 shows the comparison of the complexity of the
different detection algorithms. The complexity of a detection algorithm is
measured by the average computational time required. It can be seen that all three
combined SDR-SD detectors are faster than the stand-alone SD detector.
Furthermore, the BC-SDR-SD has the lowest complexity.
91
Figure 4.10
Comparison of the BLER performance of the SDR-SD detectors,

SDR detectors and SD.
Figure 4.11
Comparison of the computational time of the SDR-SD detectors

and SD.
92
4.7 Summary
The semidefinite relaxation approach usually consists of following four steps:
1) Convert the objective and constraint functions into convex or affine functions.
2) Relax the non convex constraints.
3) Solve the SDR problem by using interior-point methods.
4) Convert the optimal solution to a feasible solution of the original problem.
In this chapter, we firstly reviewed the existing SDR detectors for 16-QAM
constellations, which include PI-SDR, BC-SDR and VA-SDR. Then, the BC-SDR
and VA-SDR were extended into high order modulation MIMO system, such as
256-QAM constellations. After that, we proposed a new SDR detector which can
transform the 256-QAM constellation into an equivalent 16-QAM constellation.
Next, we analyzed the tightness and the complexity of these three SDR detectors
for 256-QAM. Both theoretical analysis and simulation results demonstrate that
the proposed SDR can provide the best BLER performance among these three
detectors, while the BC-SDR detector and the VA-SDR detector provide exactly
the same BLER performance. Moreover, the BC-SDR offers the lowest
computational complexity and the VA-SDR is with the highest computational
complexity, while the complexity of the proposed SDR is higher than that of the
BC-SDR and lower than that of the VA-SDR. Finally, the principle of the
combined SDR-SD detection algorithms was proposed. The simulation results
show that the SDR-SD detectors have much lower complexity compared with the
stand-alone SD detector while maintaining the optimum BLER performance.
93
Chapter 5 Lattice-Reduction-Aided Semidefinite Relaxation Detection Algorithms
CHAPTER 5
LATTICE-REDUCTION-AIDED SEMIDEFINITE
RELAXATION DETECTION ALGORITHMS
5.1 Introduction
The semidefinite relaxation (SDR) method for MIMO detection is an
attractive alternative to the ML decoding because it is very computationally
efficient. In the case of the BPSK and 4-QAM constellation, the BLER
performance of SDR detection is fairly close to the optimum BLER performance.
It has been proven that the receive diversity order of SDR detector is M R / 2 in
real valued cases, which is equal to that of the optimal ML decoding. However,
for higher level modulation, the SDR detection could not achieve the full diversity
which can be observed from the simulation results presented in Chapter 4.
For BPSK and 4-QAM, the real elements of the constellation belong to the set
{-1, 1}. When deducing the corresponding SDR problem, the constraints that can
represent the characteristic of the elements of the constellation are given by:
diag ( X ) = 1e
(5.1)
rank ( X ) = 1
Although there is no explicit integer constraint, these three conditions together are
exactly the same as x {1,1} . In order to obtain the SDR problem, the nonconvex constraint rank ( X ) = 1 is removed so that it can be efficiently solved by
the SDP solvers. But due to the elimination of this rank-one constraint, the
may not be restricted to be rank-one, so we need to convert it

resulting solution X
94
into a feasible solution x by using rank-one approximations. Since the elements

of the constellation are either 1 or -1, the constraint of the alphabet set x {1,1}
shall strictly hold by diag ( X ) = 1e . Thus the BLER performance of the SDR
detection mainly depends on the method of the rank-one approximation which
.
extracts the feasible solution x from solution X
But in the cases of higher level modulation such as 16-QAM (24-QAM), 32-
QAM (25-QAM), 64-QAM (26-QAM) or 256-QAM (28-QAM), the characteristic

of the constellation can not be simply expressed equivalently to the several
constraints similar to those listed in (5.1). As elaborated in Chapter 4, in BC-SDR,
the characteristic of the constellation is described by:
X
1e diag ( X1,1 ) ( 2q /2 1) e
2
(5.2)
X 2,2 = 1
rank ( X ) = 1
It should be noted that (5.2) can cover all the elements of the constellation of 2qQAM.
{ ( 2
q 2
However,
1) ,
, 1,1,
it
is
not
exactly
equivalent
to
the
alphabet
set
, ( 2q 2 1) . It means that in BC-SDR, the constraint of the
alphabet set can not be strictly satisfied besides the rank-one constraint being
omitted. For PI-SDR and VA-SDR, similar problems will inevitably be
encountered when deducing their corresponding SDR problems. Thus, in higher
level modulation MIMO systems, the BLER performance of SDR detection is
much worse than that of the ML detection. In what follows, we will apply the
lattice reduction (LR) algorithm to SDR detection, so that the novel LR-aided
SDR detectors could offer much improved BLER performance.
95
5.2 Lattice Reduction

Lattice reduction algorithm is used to transform the original channel matrix
into a new channel matrix with better channel condition [77]. It can be combined
with the linear detectors to improve the error rate performance.
The singular value decomposition of the channel matrix H can be expressed
as the factorization of the form:
H = UDV H
(5.3)
where U is an M R M R real unitary matrix, V is an M T M T real unitary matrix,

D is an M T M R diagonal matrix, which has nonnegative real numbers in its
diagonal entries. The diagonal entries are called the singular values of the channel
matrix and denoted as 1 , 2 ,
min ( M
,M R )
The channel condition number is defined as the maximum singular value

divided by the minimum singular value of the channel matrix [77-78]:
(H) =
max ( i )
i
min ( i )
(5.4)
When the condition number is close to 1, the channel is a well-conditioned

channel. Otherwise, the channel is bad-conditioned. Lattice reduction (LR)
techniques can be used to transform a bad-conditioned channel matrix into a wellconditioned channel matrix.
5.2.1 Lattice Reduction Techniques

The columns of the channel matrix h l are assumed to be the lattice basis,
where 1 < l < M T . Assuming the possible transmit vectors are given by ,
96
MT
, the possible received signals which are not disturbed by the Gaussian
noise are represented by the lattice as:
MT
L ( H ) = L h1 , , h MT = hl
For
any
lattice
= h1 h 2
L,
there
are
many
(5.5)
l =1
possible
lattice
bases.
Herein,
h MT is one of the bases for the lattice L. For any matrix
T such that T is unimodular, = T is also a basis for the lattice L:
( )
L ( H ) = L = HT
(5.6)
The matrix T is unimodular means that T and T1 has only integer entries and
the determinant of T equals 1 or -1. Apparently, we have Hx = ( HT ) ( T-1x ) = z ,
where z = T1x . Therefore, a point which is denoted by x in the basis H can also
be denoted by z in the basis .
By using a unimodular matrix T, LR transforms the matrix H into a new
matrix . The newly derived basis vectors are more orthogonal to each other and
have shorter lengths, so that a better decision domain can be obtained. It is worth
noting that although the decision domain is changed with the basis vector, the
lattice still remains itself unchanged.
Although the LR technology may not always lead to the optimum solution, it
indeed could improve the BLER performance of the detectors. The more
correlated the columns of the original channel matrix H are, the more significant
improvement of the BLER performance could be obtained. Consequently, to find
the optimum lattice basis is the so-called lattice reduction problem.
5.2.2 LLL algorithm
97
There are three main lattice reduction algorithms: Minkovosky algorithm, KZ

algorithm and LLL algorithm [30]. None of these can find exactly the shortest
basis of the lattice, but LLL algorithm is able to find the approximate shortest
basis of any lattice in polynomial time.
Fig 5.1 indicates an example of LLL reduction algorithm: the basis h1 , h 2
shown in orange solid lines is obtained from the basis ( h1 , h 2 ) indicated in blue
dash lines by using LLL reduction algorithm. It can be seen that the basis vectors
h1 and h 2 are more orthogonal to each other and their lengths are shorter.
Moreover, the lattice points remain unchanged during the LR process.
Figure 5.1
Example of LLL algorithm.
We apply QR decomposition to the reduced basis H :
H = QR
(5.7)
where Q = q1 q 2 q MT is an M R M T matrix whose columns are

orthogonal and with unit length. R
MT MT
is an upper triangular matrix, and
ri , j denotes its element located in the i th row and j th column. Define a
98
parameter where 1/ 4 < < 1 , the elements in the matrix R should fulfill the
following conditions:
rl ,k < 1/ 2 rl ,l
for 1 l < k M T
rk21,k 1 rk2,k 1 + rk21,k
for k = 2, M T .
(5.8)
(5.9)
The parameter is a factor to be selected so as to achieve the tradeoff between

the quality and complexity of the algorithm. We assume = 3 / 4 in this thesis as
proposed in [30].
The LLL reduction algorithm consists of two main steps including the size
reduction step and the basis vector swapping step [30]. The former step aims at
making the basis vector more orthogonal to each other and shorter in length by
evaluating (5.8). In the latter steps, if the condition (5.9) is not satisfied, the two
consecutive basis vectors will be swapped. Then go back to the first step to make
the lengths of the vectors shorter. Table 5.1 shows the detailed description of the
LLL reduction algorithm. The inputs include Q and R which are derived from the
QR decomposition of the basis H. The outputs are Q , R and the unimodular
matrix T.
Table 5.1 LLL Lattice Reduction Algorithm [27, 30, 67]

INPUT: Q, R. OUTPUT: Q , R , T.
Initialization step
1: Q := Q , R := R , T := I MT
2. k=2
3. while k M T
Size reduction step
99
4.
for l = k 1,
,1
5.
= round ( R ( l , k ) / R ( l , l ) )
6.
if 0
7.
R (1: l , k ) := R (1: l , k ) R (1: l , l )
8.
T (:, k ) := T (:, k ) T (:, l )
end
9.
10.
end
Basis vector swapping step

11.
if R ( k 1, k 1) > R ( k , k ) + R ( k 1, k )
2
12.
swap R (:, k ) and R (:, k 1)
13.
swap T (:, k ) and T (:, k 1)
14.
Calculate the given rotation matrix G such that element

R ( k , k 1) become zero:
G=
R ( k , k 1)
R ( k 1: k , k 1)
where
R ( k 1, k 1)
R ( k 1: k , k 1)
15.
R ( k 1: k , k 1: M T ) := GR ( k 1: k , k 1: M T )
16.
Q (:, k 1: k ) = Q (:, k 1: k ) G T
17.
k := max {k 1, 2}
18.
else
19.
k := k + 1
20.
end
and
21. end
100
5.3 Lattice-Reduction-Aided SDR Detection

5.3.1 Fundamental of LR-SDR
As mentioned before, the lattice reduction combined with linear detectors
provides much better BLER performance compared with that of the linear
detectors. Therefore, we propose the lattice-reduction-aided semidefinite
relaxation detection algorithm. By considering the reduced lattice basis H instead
of the original basis H, the performance of the SDR detection can be greatly
improved.
Fig 5.2 shows the block diagram of the conventional SDR detectors for
MIMO system. The SDR detector compensates for the original channel H and the
received signal r are detected directly by the SDR detector and the output is the
solution x .
Fig 5.3 indicates the block diagram of the proposed lattice-reduction-aided
SDR (LR-SDR) detector for MIMO system. The system performs a basis change
via an unimodular matrix T. Then the SDR detector is applied to compensate for
the new channel matrix H , and detect the signal z as the output. After that, the
final solution x is obtained by x = Tz .
Figure 5.2
The conventional SDR detector.
101
Figure 5.3
The proposed Lattice-Reduction-Aided SDR detector.
5.3.2 The LR-SDR detector 1 for 16-QAM

The MIMO system shown in Fig 5.3 is written as:
r = Hx + n = ( HT ) ( T1x ) + n
= Hz + n
(5.10)
It can be considered that the channel matrix is H and the transmitted signal is z.
Thus, the ML detection problem given in (1.8) becomes:
z ML = arg min r Hz
(5.11)
x=T1z
where = {3, 1,1,3} for 16-QAM.

Problem (5.11) can be rewritten as:
min rT r 2rT Hz - zT HT Hz
st.
z = T-1 x
x {3, 1,1,3}
(5.12)
(5.13)
In order to derive the SDR problem, the optimization problem (5.12-5.13) should
be formulated in a higher dimension. Thus, we define a rank-one semidefinite
matrix Z, which is given by:
102

T
Z = z T 1 z T 1
T1x ( T1x )T
=
T
1
( T x )
T1x
T1xxT ( T1 )T
=
T 1 T
x ( T )
T 1x
(5.14)
Then, the objective function (5.12) is equivalent to:

T
T H H
HT r T
min z T 1 T
1
z
T
r H r r
(5.15)
HT H HT r
= min Tr Z T
r H rT r
We apply block decomposition to the rank-one matrix Z in (5.14), it becomes:

Z1,1
Z=
Z 2,1
Z1,2
Z 2,2
(5.16)
where Z1,1 = T1xxT ( T1 ) , Z1,2 = T1x , Z 2,1 = xT ( T1 ) , Z 2,2 = 1 .

T
Please be noted that the diagonal elements of

i = 1, 2,
TZ1,1TT are xi2 , for
M T . It satisfies the alphabet set diag ( TZ1,1TT ) {1,9} . Relax it into
1e diag ( TZ1,1TT ) 9e , which is a convex half space constraint.

Thus, the ML problem becomes:
st.
HT H HT r
min Tr Z T
r H rT r
(5.17)
(5.18)
1e diag ( TZ1,1TT ) 9e
(5.19)
Z 2,2 = 1
(5.20)
rank ( Z ) = 1
(5.21)
103
It is well known that the extremely high complexity of the ML detection is caused
by the presence of the non-convex rank-one constraint (5.21). Thus, relaxation of
this constraint will be engaged to transform the non-convex problem into a
semidefinite relaxation problem:
st.
HT H HT r
min Tr Z T
r H rT r
(5.22)
(5.23)
Z 2,2 = 1
(5.24)
(5.25)
The SDR detector can be solved by any of the SDP solvers, such as Sedumi
[35] which is based on the interior-point methods [52].
The problem (5.22-5.25) is not exactly equivalent to the problem (5.12-5.13)
since the rank-one constraint (5.21) has been relaxed, and the constraint (5.24) is a
sufficient but non-necessary condition of the constellation of 16-QAM given in
(5.13). Therefore, the solution obtained from solving the problem (5.22-5.25) may
have more errors than ML decoding. Once the solution of problem (5.22-5.25) is
to
solved by the SDP solver, the randomization approach is applied to Z
2,1
recover z .
After the solution z is obtained, the estimate of x is given by x = TQ {z } .
Since the lattice reduction is based on the integer lattice, Q {i} is the quantization
operation to the integer lattice. However, the quantization does not fit the
boundary region of the constellation of x, and the final solution x may be outside
the boundary of the constellation and will result in errors. Thus, the solution
z needs shifting and scaling to obtain a suitable solution x .
104
( )
If L ( H ) = L H is satisfied and the transmitted signal is located in the
( )
infinite integer space , the lattice L ( H ) and the reduced lattice L H represent
the same lattice. However, the transmitted signal of QAM modulation is limited in
a constellation space and does not fulfill the latter assumption. Thus should
be considered as an amendment version of a subset of the infinite integer
space [79-81]:
= a + 1MT
2
M
M
where =
,
+ 1,
2
2
(5.26)
1 and M is the modulation level, a is

2
the QAM energy normalization parameter. Therefore, the transmitted signal can
be expressed as:
1
x = a + 1MT
2
(5.27)
Define an integer vector z = T1 T1 , so the signal z in the reduced

lattice can be written as:
z = T1x
1
= aT1 + 1MT
2
= a z + T11M T
2
(5.28)
In order to obtain the final result x , z is shifted and quantized in integer region:
1
1
1
z sq = a Q z T11MT + T11MT
2
2
a
(5.29)
Thus, the final result is:

1
1
1
x = Tz sq = aT Q z T11M T + T11MT
2
2
a
(5.30)
105
5.3.3 The LR-SDR detector1 for M-ary QAM

In this section, we extend the LR-SDR detector to higher level QAM. Let
M = 2q , the ML problem is an optimization problem that can be written as:
min rT r 2rT Hz - zT HT Hz
(5.31)
z = T-1 x
st.
x 2
+ 1, 2
+ 3,
,2
(5.32)
Define a rank-one semidefinite matrix Z = z T 1 z T 1 , the optimization

problem becomes:
st.
HT H HT r
min Tr Z T
r H rT r
(5.33)
(5.34)
0
2
q
1e diag ( TZ1,1TT ) 2 2 1 e
(5.35)
Z 2,2 = 1
(5.36)
rank ( Z ) = 1
(5.37)
Similarly, in order to obtain the convex optimization problem, the rank-one

constraint should be dropped:
st.
HT H HT r
min Tr Z T
r H rT r
(5.38)
(5.39)
0
2
q
1e diag ( TZ1,1TT ) 2 2 1 e
(5.40)
Z 2,2 = 1
(5.41)
The problem can be solved by common solver Sedumi, then the
randomization approach is applied to Z

2,1 to recover z . Consequently, the final
solution x is obtained by (5.30).
106
5.3.4 The LR-SDR detector 2 for 16-QAM

In section 5.3.3, the constraint (5.24) is a sufficient but non-necessary
condition of the alphabet constraint (5.13), and this will result in detection errors.
In this section, we propose another LR-SDR detector for MIMO system to
improve the error rate performance. It has more tightened alphabet constraints
than the aforementioned LR-SDR detector, thus its BLER performance is closer to
the optimum performance.
Define five real sets: 1 := ( , 3) , 2 := ( 3, 1) , 3 := ( 1,1) , 4 := (1,3)
and 5 := ( 3, + ) . For 16-QAM, signal vector is given by
x {3, 1,1,3}
= 1 2 3 4 5
= 1 2 3 4 5
(5.42)
= 1 5 3 2 4
where 1 5 3 = [ 3, 1] [1,3] , 2 = [ , 3] [ 1, + ] ,
4 = [ ,1] [3, + ] .
Fig. 5.4 shows the interval of these three sets: 1 5 3 , 2 and 4 .
Apparently, it can be observed that the intersection set of these three sets is
exactly the alphabet set of 16-QAM: {3, 1,1,3} .
These sets can also be expressed by two quadratic inequalities equations:
xi 2 + 4 xi + 3 0 and xi 2 4 xi + 3 0 , together with the constraint 1 xi 2 9 . Due

to the solution of xi 2 + 4 xi + 3 0 is 2 = [ , 3] [ 1, + ] , the solution of
xi 2 4 xi + 3 0 is 4 = [ ,1] [3, + ] and the 1 xi 2 9 is equivalent to

4 = [ ,1] [3, + ] .
107
Figure 5.4
The three real sets.
Therefore, the ML problem given in (5.11) can be represented as:

min r - Hx
st.
(5.43)
1 xi 2 9
(5.44)
xi 2 + 4 xi + 3 0
(5.45)
xi 2 4 xi + 3 0
(5.46)
where the constraints (5.44)-(5.46) is equivalent to (5.13).

When lattice reduction is involved, the ML problem becomes:
min r - Hz
st.
z = T-1 x
(5.47)
(5.48)
1 ( Tz )i 9
(5.49)
( Tz )i
+ 4 ( Tz )i + 3 0
(5.50)
( Tz )i
4 ( Tz )i + 3 0
(5.51)
108
To derive a SDR, the problem is equivalent to:

HT H HT r
min Tr Z T
r H rT r
st.
(5.52)
(5.53)
(5.54)
diag ( TZ1,1TT ) + 4TZ1,2 + 3
(5.55)
diag ( TZ1,1TT ) 4TZ1,2 + 3
(5.56)
Z 2,2 = 1
(5.57)
rank ( Z ) = 1
(5.58)
But it is not convex due to the rank-one constraint, so we eliminate the constraint:
HT H HT r
min Tr Z T
r H rT r

st.
(5.59)
(5.60)
(5.61)
diag ( TZ1,1TT ) + 4TZ1,2 + 3
(5.62)
diag ( TZ1,1TT ) 4TZ1,2 + 3
(5.63)
Z 2,2 = 1
(5.64)
At this point, the SDP solvers could be used to solve the problem and obtain
, then the randomization procedure is applied to Z
the solution Z
2,1 to recover z .
Consequently, the final solution x is obtained by equation (5.30).
5.3.5 The LR-SDR detector 2 for M-ary QAM
For higher level QAM MIMO system, the LR-SDR problem is formulated as:
HT H HT r
min Tr Z T
r H rT r
(5.65)
109
st. Z
(5.66)
q +1
diag ( TZ1,1TT ) 2 2 + 4i + 4 TZ1,2
, i = 0,1,
q
q
2
2
+ 2 + 2i + 1 2 + 2i + 3 0
(5.67)
, 2
(5.68)
Z 2,2 = 1
(5.69)
The problem can be solved by common SDP solver Sedumi, then the
randomization approach is applied to Z

2,1 to recover z . Consequently, the final
solution x is again obtained by (5.30).
5.4 Simulation Results

Computer simulations are conducted to evaluate the performance of the
proposed LR-SDR detectors. An uncoded MIMO system with independent
Rayleigh fading channel was taken into account and the Sedumi toolbox within
Matlab software was used to implement the BC-SDR, LR-SDR1, LR-SDR2, LRZF detector and sphere decoding (SD). Fig. 5.5 compares the BLER performances
of these five detectors for 4 4 and 6 6 16-QAM systems, respectively. It can
be seen that the LR-SDR detectors can provide near optimum BLER
performances and they are much better than the BC-SDR detector, which is
credited to the lattice reduction. The LR-SDR detectors can also reach the full
diversity order which is the same as that of the SD. The performance of the LRSDR detector 2 is closer to that of SD than LR-SDR detector 1 since the
110
constraints of LR-SDR detector 2 is more tightened than those of LR-SDR

detector 1.
Fig. 5.6 and Fig 5.7 compare the BLER performances of BC-SDR, LR-SDR1,
LR-SDR, LR-ZF detector and SD for 64-QAM and 256-QAM systems,
respectively. From these figures, similar conclusions can be reached. That is, the
LR-SDR detector 2 also outperforms the other detectors and then is followed by
LR-SDR detector 1.
Both the LR algorithm and the SDR detector can be solved in polynomialtime, so the LR-SDR detectors also have very low polynomial-time complexity.
The average computational times of the BC-SDR, LR-SDR1 and LR-SDR2
detectors are indicated in Table 5.2 to Table 5.7 to evaluate their complexity. It
can be seen that the LR-SDR detectors consume a bit more time than the BC-SDR
detectors because of the lattice reduction procedure. Moreover, the LR-SDR1
detector is faster than LR-SDR2, since the LR-SDR2 involves more constraints
than LR-SDR1 to deal with.
111
(a)
(b)
Figure 5.5
The BLER performance of the LR-SDR detectors, SDR detector,

SD and LR-ZF detector using 16-QAM.
112
(a)
(b)
Figure 5.6

113
(a)
Figure 5.7
(b)
114
Table 5.2 The computational time of 4 4 16-QAM MIMO System.

Detector
Computational time
BC- SDR
0.32s
LR-SDR1
0.37s
LR-SDR2
0.40s

Detector
Computational time
BC- SDR
0.36s
LR-SDR1
0.44s
LR-SDR2
0.48s

Detector
Computational time
BC- SDR
0.32s
LR-SDR1
0.37s
LR-SDR2
0.49s

Detector
Computational time
BC- SDR
0.36s
LR-SDR1
0.44s
LR-SDR2
0.60s
115

Detector
Computational time
BC- SDR
0.32s
LR-SDR1
0.37s
LR-SDR2
0.67s

Detector
Computational time
BC- SDR
0.36s
LR-SDR1
0.44s
LR-SDR2
0.83s
5.4 Discussion
The LR-SDR detection algorithms presented in this chapter are inspired by
the LR technology combined with the linear detectors. It has been introduced in
Chapter 2 that the decision region of ZF decoder can be dramatically improved by
combining LR technology and the LR-ZF decoder can offer much better BLER
performance than ZF decoder. The proposed LR-SDR detection algorithms are
also able to provide significantly improvement in BLER performance compared
with the sole SDR detector, and with only a little complexity added. However,
since the SDR MIMO detection method is an empirical method, it is quite difficult
to theoretically analyze how its BLER performance is affected and by what
factors. Nevertheless, it is for sure that the performance of SDR decoders is
influenced by the condition of the channel. By using LR to improve the channel
116
condition, the SDR decoder could offer better BLER performance. This has been
fully verified by the simulation results presented in section 5.3. For the
completeness of the theory, it would be better if this improvement can be
supported by theoretical analysis. In our future work, we will try to investigate
this problem.
117
Chapter 6 Conclusions and Recommendations
CHAPTER 6
CONCLUSIONS AND RECOMMENDATIONS
6.1 Conclusions
The background, system model, working principle and detection design
details of MIMO systems have been comprehensively analyzed and simulated in
this thesis. The main achievements of the thesis can be summarized as follows:
1.
The geometric analysis of the signal detection for MIMO systems is

presented. Based on this, an optimum ellipsoid-searching decoding
algorithm which is add-on to standard suboptimal detection schemes
is proposed. It can provide the optimum error rate performance which
is the same as the ML decoding, while its complexity is much lower
than that of ML decoding.
2.
The possibility of using convex optimization in MIMO detection is

investigated. The semidefinite relaxation (SDR) detection algorithms
for low-order modulation are introduced. Two existing SDR detection
algorithms for low level constellation have been extended into the
cases of high level constellations. A new SDR detector for 256-QAM
constellation MIMO system is proposed. Then, the tightness and the
complexity of these three SDR detectors are comparatively analyzed.
Both theoretical analysis and simulation results demonstrate that the
proposed SDR can provide the best BLER performance with
polynomial-time complexity.
118
3.
The semidefinite relaxation initiated sphere detector is proposed. The

SDR-SD detectors can offer the same BLER performance as the SD,
and have much better BLER performance than the stand-alone SDR
detectors. The average computational times of the combined SDR-SD
detectors are much faster than the SD.
4.
The
lattice-reduction-aided
semidefinite
relaxation
detection
algorithms are proposed. This kind of detectors can reach the full
diversity order and provide near optimum BLER performances. The
LR-SDR detectors have very low polynomial-time complexity. Its
computational time keeps unchanged with different SNR conditions
and is only a little longer than that of the SDR detector.
6.2 Recommendations
Some suggestions for future research are:
1.
It is believed that the performance of SDR detectors can be influenced

by the channel condition. This has been verified by the simulation
results. The theoretical analysis on the relationship between the
performance of SDR detectors and the channel condition number can
be investigated.
2.
The software solver of the SDR detectors used in this thesis is based
on the interior-point method. New improved interior-point methods to
reduce the iteration complexity can be investigated.
3.
The geometric decoding presented in Chapter 3 can deduce tighter

constraints for the possible value of each dimension of the signal
119
vectors that may be solved as the final solution. These constraints

could be combined with the SDR detector presented in Chapter 4, so
as to further improve its BLER performance.
120
List of Figures
LIST OF FIGURES
Figure 1.1 Structure of MIMO system....................................................................3
Figure 1.2 MIMO space-time block code system. ..................................................4
Figure 1.3 MIMO spatial multiplexing system.......................................................6
Figure 1.4 Sphere decoding. .................................................................................11
Figure 2.1 The tree search structure of sphere decoding ....................................24
Figure 2.2 The original transmit signal lattice. .....................................................28
Figure 2.3 The received signal lattice. ................................................................29
Figure 2.4 ZF decoding decision region. ..............................................................29
Figure 2.5 SIC detection decision region..............................................................30
Figure 2.6 ML decoding decision region. .............................................................31
Figure 2.7 LR-ZF decision region.........................................................................32
Figure 3.1 Probability density function of the received signal vector of ZF
detections in 2 2 MIMO systems.
(a) Case for good channel condition. (b) Case for bad channel condition. ..........38
Figure 3.2 Elliptic paraboloid with axis perpendicular to a subspace spanned by
lattice points...........................................................................................................39
Figure 3.3 Elliptic paraboloid in 3-dimensional space. ........................................40
Figure 3.4 2-D lattice space example....................................................................46
Figure 3.5 3-D lattice space example....................................................................48
using 4-QAM.
(a) 4 4 MIMO systems. (b) 8 8 MIMO systems. .........................................50
121
List of Figures
using 16-QAM.
(a) 4 4 MIMO systems. (b) 8 8 MIMO systems. .........................................51
using 64-QAM in 4 4 MIMO systems..............................................................52

Figure 4.1 (a) Convex set. (b) Non-convex set. ....................................................56
Figure 4.2 (a) Convex function. (b) Concave function. ........................................57
Figure 4.3 Comparison of the three approximation procedures............................72
Figure 4.4 Comparison of BLER performance of SDR detectors, ML decoding,
ZF using 4 4 16-QAM MIMO systems. ...........................................................73

Figure 4.5 Comparison of BLER performance of SDR detectors, ML decoding,
ZF using 8 8 16-QAM MIMO systems. ...........................................................73

Figure 4.6 The BLER performance of the SDR detectors for 4 4 256-QAM
MIMO systems. .....................................................................................................87

Figure 4.7 The BLER performance of the SDR detectors for 8 8 256-QAM
MIMO systems. .....................................................................................................88

Figure 4.8 The computational time of the SDR detectors for 4 4 256-QAM
MIMO systems. .....................................................................................................88

Figure 4.9 The computational time of the SDR detectors for 8 8 256-QAM
MIMO systems. .....................................................................................................89

Figure 4.10 Comparison of the BLER performance of the SDR-SD detectors,
SDR detectors and SD. ..........................................................................................92

Figure 4.11 Comparison of the computational time of the SDR-SD detectors and
SD. .........................................................................................................................92
Figure 5.1 Example of LLL algorithm..................................................................98
122
List of Figures
Figure 5.2 The conventional SDR detector.........................................................101

Figure 5.3 The proposed Lattice-Reduction-Aided SDR detector......................102
Figure 5.4 The three real sets. .............................................................................108
Figure 5.5 The BLER performance of the LR-SDR detectors, SDR detector, SD
and LR-ZF detector using 16-QAM.

(a) 4 4 MIMO systems. (b) 6 6 MIMO systems........................................ 112


123
List of Tables
LIST OF TABLES
Table 3.1 Number of lattice points visited using ML decoding/proposed GD
(reduction is indicated as percentage)....................................................................52
Table 4.1 The values of x. .....................................................................................77
Table 5.1 LLL Lattice Reduction Algorithm.........................................................99
Table 5.2 The computational time of 4 4 16-QAM MIMO systems. ............115
Table 5.3 The computational time of 6 6 16-QAM MIMO systems.............. 115
Table 5.4 The computational time of 4 4 64-QAM MIMO systems.............. 115
Table 5.5 The computational time of 6 6 64-QAM MIMO systems. ............ 115
Table 5.6 The computational time of 4 4 256-QAM MIMO systems. .......... 116
Table 5.7 The computational time of 6 6 256-QAM MIMO systems. .......... 116
124
Abbreviations
Abbreviations
2G
second-generation
3G
third-generation
4G
fourth-generation
MIMO
multiple-input multiple-output
SM
spatial multiplexing
DBLAST
diagonally-bell laboratories layered space-time
VBLAST
vertical-bell laboratories layered space-time
LTE
long-term-evolution
WLAN
wireless local area network
UWB
ultra-wide-band
CR
cognitive radio
STC
space-time coding
STTC
space-time trellis codes
STBC
space-time block codes
SNR
signal to noise ratio
ML
maximum likelihood
NP-hard
non-deterministic polynomial time hard
BER
bit-error rate
EUB
exact union bound
LDC
linear-dispersion code
SD
sphere decoding
FSD
fixed-complexity sphere decoding
AWGN
additive white Gaussian noise
PEP
pair-wise-error probability
OSTBC
orthogonal space-time block code
ML
maximum likelihood
ZF
zero-forcing
MMSE
minimum mean-square error

125
Abbreviations
SD
sphere decoding
SIC
successive interference cancellation
LRD
lattice reduction aided decoding
SDR
semidefinite relaxation
F-P
Fincke-Pohst
S-E
Schnorr-Euchner
LLL
LenstraLenstraLovsz
SDP
semidefinite programming
BPSK
binary phase-shift keying
QAM
quadrature amplitude modulation
PI-SDR
polynomial-inspired SDR
BC-SDR
bound-constrained SDR
VA-SDR
virtually-antipodal SDR
BLER
block-error rate
MSE
mean-square-error
ISI
inter-symbol-interference
SINR
signal to interference plus noise ratio
LR
lattice reduction
ESA
ellipsoid searching decoding algorithm
PDF
probability density function
QCQP
quadratically constrained quadratic programming
IPM
interior-point method
LR-SDR
lattice-reduction-aided SDR
126
References
REFERENCES
[1]. A. Paulraj and T. Kailath, Increasing capacity in wireless broadcast systems

using distributed transmission/directional reception (DTDR), US Patent
No.5345599, Sept. 1994.
[2]. I. E. Telatar, Capacity of multi-antenna gaussian channels, Tech. Rep.,
AT&T Bell Labs., 1995.
[3]. G. J. Foschini, Layered space-time architecture for wireless communication
in a fading environment when using multi-element antennas, Bell Labs
Tech. J., vol. 1, no. 2, pp. 4159, 1996.
[4]. G. D. Golden, G. J. Foschini, R. A. Valenzuela, and P. W. Wolniansky,
Detection algorithm and initial laboratory results using V-BLAST
space-time communication architecture, Electron. Lett., vol. 35, pp. 1416,
Jan. 1999.
[5]. A. Paulraj, R. Nabar and D. Gore Introduction to space time wireless
communication, Cambridge University Press, 2003.
[6]. S. M. Alamouti, A simple transmitter diversity scheme for wireless
communications, IEEE J. on Selected Areas in Communications, vol. 16,
pp. 1451-1458, Oct. 1998.
[7]. P. W. Wolniansky, G. J. Foschini, G. D. Golden and R. A. Valenzuela,
V-BLAST: An architecture for realizing very high data rates over the
rich-scattering wireless channel, in IEEE Proc. of ISSSE-98, Pisa, Italy,
Sept. 1998.
[8]. R. W. Lucky, J. Salz and E. J. Weldon, Principles of Data Communications,
127
References
New York: McGraw-Hill, 1968.

[9]. B. R. Saltzberg, Intersymbol interference error bounds with application to
ideal band-limited signaling, IEEE Trans. on Information Theory, vol.14,
pp. 563-568, 1968.
[10]. G. J. Foschini, G.. D. Golden, R. A. Valenzuela and P. W. Wolniansky,
Simplified processing for high spectral efficiency wireless communication
employing multi-element arrarys, IEEE J. on Selected Areas in
Communications, vol. 17, no. 11, pp. 1841-1852, Nov. 1999.
[11]. S. W. Kim and K. P. Kim, Log-likelihood-ratio-based detection ordering in
V-BLAST, IEEE Trans. on Communications, vol. 54, no. 2, pp. 302-307,
Feb. 2006.
[12]. J. R. Barry, E. A. Lee and D. G.. Messerschmitt, Digital Communications,
Springer, 3rd ed, 2004.
[13]. D. Tse and P. Viswanath, Fundamentals of Wireless Communication,
Cambridge University Press, May 2005.
[14]. E. Viterbo and J. Boutros, A universal lattice code decoder for fading
channels, IEEE Trans. on Information Theory, vol. 45, no. 5, pp.
1639-1642, July 1999.
[15]. H. Vikalo and B. Hassibi, On the expected complexity of sphere
decoding, in the Thirty-Fifth Asilomar Conference on Signals, Systems and
Computers, vol. 2, pp. 1051-1055, 2001.
[16]. O. M. Damen, H. E. Gamal, and G. Caire, On the complexity of ML
detection and the search for the closest lattice point, IEEE Trans. on
Information Theory, vol. 59, no. 10, pp. 24002414, Oct. 2003.
128
References
[17]. E. Zimmermann, W.Rave, G. Fettweis, On the complexity of sphere

decoding, in Proc. of the International Conference on Wireless Personal
and Multimedia Communications (WPMC04), Abano, 2004.
[18]. U. Fincke and M. Pohst, Improved methods for calculating vectors of short
length in a lattice, including a complexity analysis, Math. Comput., vol. 44,
pp.463-471, 1985.
[19]. C. P. Schnorr, M. Euchner, Lattice basis reduction: improved practical
algorithms and solving subset sum problems, Math. Program., vol. 66, no.
2, pp.181-191, 1984.
[20]. E. Agrell, E. Thomas, A. Vardy, and K. Zeger, Closest point search in
lattices, IEEE Trans. on Information Theory, vol. 48(8), pp. 2201-2214,
Aug. 2002.
[21]. O. Damen, A. Chkeif and J. Belfiore, Lattice code decoder for space time
codes, IEEE Communications Lett., vol.4, pp.161-163, 2000.
[22]. Q. Liu and L. Yang, A novel method for initial radius selection of sphere
decoding, in IEEE Proc. of VTC, vol. 2, pp. 1280-1283, 2004.
[23]. A. Ghaderipoor and C. Tellambura, Senior Member, A statistical pruning
strategy for Schnorr-Euchner sphere decoding, IEEE Communication Lett.,
vol. 12, no. 2, Feb. 2008.
[24]. K. Su and I. J. Wassell, A new ordering for efficient sphere decoding,
IEEE International Conference on Communications 2005, pp. 1906-1910,
2005.
[25]. J. Jaldn and B. Ottersten, An exponential lower bound on the expected
complexity of sphere decoding, in Proc. ICASSP, Montreal, QC, Canada,
129
References
2004.
[26]. C. P. Schnorr and M. Euchner, Lattice basis reduction: Improved practical
algorithms and solving subset sum problems, Math. Program., vol. 66, no.
13, pp. 181199, Aug. 1994.
[27]. C.
Windpassinger,
L.
Lampe,
and
R.
F.H.
Fischer,
From
lattice-reduction-aided detection towards maximum-likelihood detection in

MIMO systems," in Proc. International Conf. Wireless Optical Commun.,
Banff, Canada, July 2003.
[28]. H. Yao and G.W. Wornell, Lattice-reduction-aided detectors for MIMO
communication systems," in Proc. IEEE Global Telecommun. Conf., vol. 1,
pp. 424-428, Taipei, Taiwan, Nov. 2002.
[29]. D. Wbben, R. Bhnke, V. Khn, and K.-D. Kammeyer, Near
maximum-likelihood detection of MIMO systems using MMSE-based
lattice reduction," in Proc. IEEE International Conf. Commun., vol. 2, pp.
798-802, Paris, France, June 2004.
[30]. A. K. Lenstra, H. W. Lenstra, and L. Lovsz, Factoring polynomials with
rational coefficients," Mathematicsche Annalen, vol. 261, pp. 515-534,
1982.
[31]. J. Jaldn, D. Seethaler, G. Matz, Worst- and average-case complexity of
LLL lattice reduction in MIMO wireless systems, in Proc. of ICASSP, pp.
2685- 2688, May, 2008.
[32]. Y. H. Gan and W. H. Mow, Novel joint sorting and reduction technique for
delay-constrained LLL-aided MIMO detection, IEEE Signal Processing
Lett., vol.15, pp. 194-197, 2008.
130
References
[33]. Y. H. Gan, C. Ling and W. H. Mow, Complex lattice reduction algorithm

for low-complexity full-diversity MIMO detection, IEEE Trans. on Signal
Processing, vol. 57, no. 7, Jul. 2008.
[34]. A. Ben-Tal, A. Nemirovski, Lectures on modern convex optimization.
MPS-SIAM Series on Optimization, 2001.
[35]. J. F. Sturm,Using SEDUMI 1.02, A Matlab toolbox for optimizations over
symmetric cones, Optim. Meth. Soft., vol. 1112, 1999.
[36]. W.-K. Ma, T. N. Davidson, K. M. Wong, Z.-Q. Luo, and P.-C. Ching,
Quasimaximumlikelihood multiuser detection using semidefinite
relaxation with application to synchronous CDMA, IEEE Trans. on Signal
Processing, vol. 50, no. 4, pp. 912-922, 2002.
[37]. P. H Tan, L. K. Rasmussen, The application of semidefinite programming
or detection in CDMA, IEEE J. on Selected Areas in Communications, vol.
19, no.8, pp. 1442-1449, 2001.
[38]. B. Steingrimsson, Z.-Q. Luo, and K. M. Wong, Soft quasi-maximum
likelihood detection for multiple-antenna channels, IEEE Trans. on Signal
Processing, vol. 51, no. 11, pp. 2710-2719, Nov. 2003.
[39]. J. Jaldn and B. Ottersten, The diversity order of the semidefinite
relaxation detector, IEEE Trans. on Information Theory, vol. 54, no. 4, pp.
1406-1422, 2008.
[40]. A. Wiesel, Y. C. Eldar, and S. Shamai (Shitz), Semidefinite relaxation for
detection of 16-QAM signaling in MIMO channels, IEEE Signal
Processing Lett., vol. 12, no. 9, pp. 653-656, 2005.
[41]. N. D. Sidiropoulos and Z.-Q. Luo, A semidefinite relaxation approach to
131
References
MIMO detection for high-order QAM constellations, IEEE Signal

Processing Lett., vol. 13, no. 9, pp. 525-528, 2006..
[42]. Z. Mao, X. Wang, and X. Wang, Semidefinite programming relaxation
approach for multiuser detection of QAM signals, IEEE Trans. on Wireless
Communications, vol. 6, no. 12, pp. 4275-4279, 2007.
[43]. W.-K. Ma, C.-C. Su, J. Jaldn, T.-H. Chang, and C.-Y. Chi, The
equivalence of semidefinite relaxation MIMO detectors for higher-order
QAM, IEEE J. on Selected Areas in Communications, vol. 3, no. 6,
pp.1038-1052, 2009.
[44]. Y. Yang, C. Zhao, P. Zhou, and W. Xu, MIMO detection of 16-QAM
signaling based on semidefinite relaxation, IEEE Signal Processing Lett.,
vol. 14, no. 11, pp. 797-800, 2007.
[45]. G.. Strang, Linear Algebra and Its Applications, Fort Worth, TX: Saunders
College Publishing, Brace Jovanovich College Publishers, 3rd edition, 1988.
[46]. E. Agrell, T. Eriksson, A. Vardy, and K. Zeger, Closest point search in
lattices, IEEE Trans. on Information Theory, vol. 48, pp. 2201-2214, Aug.
2002.
[47]. D. Tse and P. Viswanath, Fundamentals of Wireless Communication, New
York: Cambridge University Press, 2005.
[48]. V. Kuhn, Wireless Communications over MIMO Channels, Chichester,
England: John Wiley & Sons Ltd, 2006.
[49]. S. Boyd and L. Vandenberghe, Convex Optimization, Cambridge University
Press, 2004.
[50]. C. Helmberg and F. Rendl, Solving quadratic (0, 1) problems by
132
References
semidefinite programs and cutting planes, Math. Program., vol. 82, no. 3,
pp. 219315, 1998.
[51]. W.-K. Ma, P.-C. Chung and Z. Ding, Semidefinite relaxation based
multiuser detection for M-Ary PSK multiuser systems, IEEE Trans. on
Signal Processing, vol. 52, No. 10, pp. 2862- 2872, Oct. 2004.
[52]. C. Helmberg, F. Rendl, R. J. Vanderbei, H. Wolkowicz, An interior-Point
method for semidefinite programming, SIAM J. Optim., vol. 6, no. 2,
342-361, 1996.
[53]. B. M. Hochwald and S. ten Brink, Achieving near-capacity on a
multiple-antenna channel, IEEE Trans. on Communications, vol. 51, no. 3,
pp. 389399, Mar. 2003.
[54]. N. Karmarkar, A new polynomial-time algorithm for linear programming,
Combinatorica, 4 (1984), pp. 373-395.
[55]. M. Kojima, N. Megiddo, T. Noma, and A. Yoshise, A united approach to
interior point algorithms for linear complementarity problems, Lecture
Notes in Computer Science, Springer-Verlag, 1991.
[56]. Y. Nesterov and A. Nemirovsky, A general approach to polynomial-time
algorithms design for convex programming, Tech. report, Centr. Econ. &
Math. Inst., USSR Acad. Sci., Moscow, USSR, 1988.
[57]. Optimization over positive semidefinite matrices: Mathematical background
and user's SEMIDEFINITE PROGRAMMING 49 manual, USSR Acad. Sci.
Centr. Econ. & Math. Inst., 32 Krasikova St., Moscow 117418 USSR, 1990.
[58]. Self-concordant functions and polynomial time methods in convex
programming, Tech. Rep., Centr. Econ. & Math. Inst., USSR Acad. Sci.,
133
References
Moscow, USSR, Apr. 1990.

[59]. Conic formulation of a convex programming problem and duality, Tech.
Rep., Centr. Econ. & Math. Inst., USSR Academy of Sciences, Moscow
USSR, 1991.
[60]. I. J. Lustig, R. E. Marsten, and D. F. Shanno, Interior point methods for
linear programming: Computational state of the art, ORSA Journal on
Computing, 1994.
[61]. H. Wolkowicz, Some applications of optimization in matrix theory, Linear
Algebra and its Applications, 40:101-118, 1981.
[62]. R. Gowaikar and B. Hassibi, Efficient statistical pruning for maximum
likelihood decoding, IEEE International Conference on Acoustics, Speech,
and Signal Processing, vol. 5, pp. 49-52, Apr., 2003.
[63]. T. Cui, T. Ho and C. Tellambura, Statistical prunning for near maximum
likelihood detection of MIMO systems, IEEE International Conference on
Communications 2007, vol. pp. 5462-5467, 2007.
[64]. A. Burg, M. Borgmann, M. Wenk, M. Zellweger, W. Fichtner and H.
Bolcskei, VLSI implementation of MIMO detection using the sphere
decoding algorithm, IEEE J. on Solid-State Circuits, vol. 40, pp.
1566-1577, Jul., 2005.
[65]. D. Wbben, R. Bhnke, V. Khn, and K.-D. Kammeyer, MMSE-based
Lattice-Reduction for near-ML detection of MIMO Systems," in ITG
Workshop on Smart Antennas, March 2004.
[66]. M. Damen, H. E. Gamal and G. Caire, On maximum-likelihood detection
and the search for the closest lattice point, IEEE Trans. on Information
134
References
Theory, vol. 49, pp. 2389-2402, Oct. 2003.

[67]. B. Hassbi and B. M. Hochwald, High-rate codes that the linear in space
and time, IEEE Trans. on Information Theory, vol. 48, pp.1804-1824, July
2002.
[68]. H. Artes, D. Seethaler and F. Hlawatsch, Efficient detection algorithms for
mimo channels: A geometrical approach to approximate ML detection,
IEEE Trans. on Signal Processing, vol. 51, no. 11, pp. 28082820, Nov.
2003.
[69]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, A Simple and optimum geometric
decoding algorithm for MIMO systems, 4th International Symposium on
Wireless Pervasive Computing 2009, Melbourne, Australia, 2009.
[70]. R. A. Horn and C. R. Johnson, Matrix Analysis, Cambridge University Press,
1985.
[71]. M. Samuel and M. P. Fitz, Geometric decoding of PAM and QAM lattices,
in Proc. IEEE Global Telecommunications Conf., pp. 42474252, Nov.
2007.
[72]. D. Seethaler, H. Artes, H and F. Hlawatsch, Efficient near-ML detection for
MIMO channels: The Sphere-Projection Algorithm, GLOBECOM,
pp.20982093, 2003.
[73]. G. H. Golub and C. F. Van Loan, Matrix Computations, 3rd ed. Baltimore:
Johns Hopkins University Press, 1996.
[74]. Z. Guo and P. Nilsson, Algorithm and implementation of the K-best sphere
decoding for MIMO detection, IEEE J. on Selected Areas in
Communications, vol. 24, pp. 491-503, Mar., 2006.
135
References
[75]. J. B. Anderson and S. Mohan, Sequential coding algorithms: a survey and

cost analysis, IEEE Trans. on Communications, vol. 32, pp. 169-176, Feb.
1984.
[76]. D. Wbben, R. Bhnke, V. Khn, and K.-D. Kammeyer, MMSE extension
of V-BLAST based on sorted QR decomposition, in IEEE Proc. VTC,
Orlando, Florida, USA, Oct. 2003.
[77]. W. Zhang and X. Ma, Low-complexity soft-output decoding with
lattice-reduction-aided detectors, IEEE Trans. on Communications, vol. 58,
no.9, pp. 2621-2629, Sept. 2010.
[78]. M. Taherzadeh, A. Mobasher, and A. K. Khandani, Communication over
MIMO broadcast channels Using Lattice-Basis reduction, IEEE Trans. on
Information Theory, vol.53, no. 12, pp. 4567-4582, Dec. 2007.
[79]. C.
Windpassinger
and
R.
Fischer,
Low-complexity
near-maximum-likelihood detection and precoding for MIMO systems using

lattice reduction, in Proc. of Information Theory Workshop, 2003.
[80]. C.
Windpassinger,
R.
F.
H.
Fischer,
and
J.
B.
Huber,
Lattice-reduction-aided broadcast precoding, in 5th International ITG

Conference on Source and Channel Coding, pp. 403-408, Erlangen,
Germany, Jan. 2004.
[81]. C. Windpassinger, R. Fischer, and J. B. Huber, Lattice-reduction-aided
broadcast precoding, IEEE Trans. on Communications, pp. 2057-2060,
Dec. 2004.
[82]. V. Tarokh, N. Seshadri, and A. R. Calderbank, Space-time codes for high
data rate wireless communication: Performance criterion and code
136
References
construction, IEEE Trans. on Information Theory, vol. 44, pp. 744-765,

Mar. 1998.
[83]. V. Tarokh, H. Jafarkhani, and A. R. Calderbank, Space-time block codes
from orthogonal designs, IEEE Trans. on Information Theory, vol. 45, pp.
1456-1467, July 1999.
[84]. I. E. Telatar, Capacity of multi-antenna gaussian channels, in Europ.
Trans. Telecommun., pp. 585-595, Nov. 1999.
[85]. T. L. Marzetta and B. M. Hochwald, Capacity of a mobile
Multiple-Antenna communication link in rayleigh flat fading, IEEE Trans.
on Information Therory, vol. 45, no. 1, pp. 139-157, Jan.1999.
[86]. A. Edelman, Eigenvalues and condition numbers of random matrices,
Ph.D. dissertation, MIT, Cambridge, MA, May 1989.
[87]. G. H. Golub and C. F. Van Loan, Matrix Computations, 3rd ed. Baltimore:
Johns Hopkins Univ. Press, 1996.
[88]. S. M. Kay, Fundamentals of Statistical Signal Processing: Detection Theory.
Upper Saddle River, NJ: Prentice-Hall, 1998.
[89]. S. M. Kay, Fundamentals of Statistical Signal Processing: Estimation
Theory. Englewood Cliffs, NJ: Prentice-Hall, 1993.
[90]. Z.-Q. Luo and W. Yu, An introduction to convex optimization for
communications and signal processing, IEEE J. on Selected Areas in
Communications, vol. 24, no. 8, Aug. 2006.
137
Publications
PUBLICATIONS
Book Chapter:
[1]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, Geometrical detection algorithm for
MIMO systems, MIMO Systems, Theory and Applications, InTech, April
2011.
Articles:
[1]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, A Simple and Optimum Geometric
Decoding Algorithm for MIMO Systems, The 4th International
Symposium on Wireless Pervasive Computing (ISWPC), Melbourne,
Australia, pp. 1-5, Feb. 2009.
[2]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, A fast geometric decoding
algorithm for MIMO systems, The 11th International Conference on
Advanced Communication Technology (ICACT), Korea, vol. 3, pp.
2108-2111, Feb. 2009.
[3]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, Semi-definite relaxation decoder
for 256-QAM MIMO system, IET Electronics Letters, vol. 46, no. 11, pp.
796-797, 2010.
[4]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, Combined semi-definite relaxation
and sphere decoding method for multiple antennas systems, 2011
International ITG Workshop on Smart Antennas (WSA), Aachen, Germany,
pp. 1-4, Feb. 2011.
[5]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, Comparison of semidefinite
relaxation detection for 256-QAM MIMO system, submitted to Signal
138
Publications
Processing.
[6]. Z. Y. Shao, S. W. Cheung and T. I. Yuk, Lattice-Reduction-Aided MIMO
Detections Based on Semidefinite Relaxation, submitted to IEEE Signal
Processing Letter.
139

Full Text

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Full Text

Uploaded by

Copyright:

Available Formats

Title

Design and analysis of detection algorithms for MIMO wireless

Yuk, TTI; Cheung, SW

The author retains all proprietary rights, (such as patent rights)

Design and Analysis of Detection Algorithms for

A thesis submitted in partial fulfillment

To my beloved parents and husband

Abstract of thesis entitled

Total words: 343

MIMO Wireless Communication System...........................2

Detection Problem for MIMO Wireless Communication

Motivation and Contribution of the Thesis .......................14

Thesis Outline ...................................................................15

STATE-OF-THE-ART MIMO DETECTION

Linear Decoders ................................................................17

Sphere Decoding ...............................................................19

Successive Interference Cancellation ...............................25

Lattice-Reduction Aided Detection...................................27

Geometrical Analysis of Signal Decoding for MIMO

Ellipsoid-searching decoding algorithm ...........................40

Simulation Results ............................................................48

MIMO DETECTION ALGORITHMS

Convex Optimization Problems ........................................54

Low-Order Modulation Systems.................................................62

High-Order Modulation Systems ................................................74

SDR-initiated Sphere detector ..........................................89

Lattice Reduction ..............................................................96

Lattice-Reduction-Aided SDR Detection .......................101

Simulation Results .......................................................... 110

Discussion ....................................................................... 116

Conclusions ..................................................................... 118

Recommendations ........................................................... 119

Wireless communications is the most vital field in digital communications.

second-generation (2G) mobile communications services which provide up to 115

1.1 MIMO Wireless Communication System

Figure 1.1 Structure of MIMO system.

1.1.2 Space-Time Coding for MIMO Systems

better performance. In 1998, Alamouti proposed a kind of STBC called Alamouti

Figure 1.2 MIMO space-time block code system.

1.1.3 Spatial Multiplexing for MIMO Systems

In MIMO systems, different antennas transmitting the signal in parallel can

1.1.4 MIMO System Model

The Spatial multiplexing MIMO system is illustrated in Fig.1.3, in which,

Figure 1.3 MIMO spatial multiplexing system.

Thus, the MIMO system is modeled as [5]:

where r is the N R -dimensional received signal vector, and x is the

zero-mean Gaussian noise vector with elements having a fixed variance n 2 .

The dimension of r and x are M R = 2 N R and M T = 2 NT , respectively. H

1.1.5 Channel Capacity

where det ( i ) denotes the determinant of a matrix, is an identity matrix,

1.2 Detection Problem for MIMO Wireless Communication

1.2.2 Maximum Likelihood Decoding

1.3 Literature Review

performance of ZF is poor since the noise is enlarged by the generalized inverse

ordered successive interference cancellation algorithm based on the maximum

bound-constrained SDR (BC-SDR) [41] and virtually-antipodal SDR (VA-SDR)

1.4 Motivation and Contribution of the Thesis

performance. The thesis aims at investigating detection algorithms to reduce the

Geometrically analyzing the signal detections of MIMO system, which is

another perspective of reconsidering the principle of the ML decoding. Based on

Giving the extension of the existing SDR detection algorithms to

high-order modulation MIMO system, and proposing a novel SDR detection

Combining the SDR detection algorithms with the sphere decoding,