Book Description

Space-time coding is an attractive technique to exploit the transmit diversity gain provided by a multiple-input multiple-output (MIMO) wireless system. Regarding a space-time code design, some important concerns are high rates, full diversity, large coding gain (diversity products) and low decoding complexity. However, a tradeoff exists among these goals and constructing a good code that optimizes some or all of these goals is a very practical and interesting problem that has attracted a lot of attention in the past 10 years. Furthermore, other design issues may also matter and should be taken into account when one considers certain special scenarios to which the space-time coding technique is applied. In this dissertation, we study both the code design at the transmitter side and the fast decoding algorithm at the receiver side for space-time coding. The first topic attempts to achieve both low decoding overhead and maximum (full) diversity for space-time block codes (STBC). By deploying a linear detector at the receiver, we can efficiently reduce the decoding complexity for an STBC and always obtain soft outputs that are desired when the STBC is concatenated with a channel code as in a real system. In this dissertation, we propose a design criterion for STBC to achieve full diversity with a zero-forcing (ZF) or minimum mean-square error (MMSE) receiver. Two families of STBC, orthogonal STBC (OSTBC) and Toeplitz codes, which are known to have full diversity with ZF or MMSE receiver, indeed meet this criterion, as one may expect. We also show that the symbol rates of STBC under this criterion are upper bounded by 1. Subsequently, we propose a novel family of STBC that satisfy the criterion and thus achieve full diversity with ZF or MMSE receiver. Our newly proposed STBC are constructed by overlapping the 2 x 2 Alamouti code and hence are named overlapped Alamouti codes. The new codes are close to orthogonal and have asymptotically optimal symbol rates. Simulation results show that overlapped Alamouti codes significantly outperform Toeplitz codes for any number of transmit antennas and also outperform OSTBC when the number of transmit antennas is above 4. The second topic concerns the design of space-time trellis codes (STTC) for their applications in cooperative communication systems, where transmission among different relay nodes that cooperate with each other is essentially asynchronous. A family of STTC that can achieve full cooperative diversity order regardless of the node transmission delays has been proposed and it was shown that the construction of this STTC family can be reduced to the design of binary matrices that can keep full row rank no matter how their rows are shifted. We call such matrices as shift-full-rank (SFR) matrices. We propose a systematic method to construct all the SFR matrices and, in particular, the shortest (square) SFR (SSFR) matrices that are attractive as the associated STTC require the fewest memories and hence the lowest decoding complexity. By relaxing the restriction imposed on SFR matrices, we also propose two matrix variations, q -SFR and LT-SFR matrices. In an extended cooperative system model with fractional symbol delays whose maximum value is specified, the STTC generated from q -SFR and LT-SFR matrices can still achieve asynchronous full diversity. As a result, more eligible generator matrices than SFR ones become available and some better STTC in terms of coding gain may be found. Finally, the third topic is to speed up the decoding algorithm for the vertical Bell Laboratories layered space-time (V-BLAST) scheme, a full rate STBC that however does not exploit any transmit diversity gain. A fast recursive algorithm for V-BLAST with the optimal ordered successive interference cancellation (SIC) detection has been proposed and two improved algorithms for it have also been independently introduced by different authors lately. We first incorporate the existing improvements into the original fast recursive algorithm to give an algorithm that is the fastest known one for the optimal SIC detection of V-BLAST. Then, we propose a further improvement from which two new algorithms result. Relative to the fastest known one from the existing improvements, one new algorithm has a speedup of 1:3 times in both the number of multiplications and the number of additions, and the other new algorithm requires less memory storage.