Studies Of Space-time Coding Technique With Low Complexity For MIMO Systems

Space-time coding techniques can exploit the space diversity gains by utilizing the multi mutual independent transmission paths of MIMO systems to greatly increase the ability against channel fadings. In this thesis, we study the space-time coding techniques with low complexity, which is easy to implement, and propose four new coding design schemes based on multi-dimensional algebraic lattices.The MIMO channel can be equivalent to multi parallel virtual SISO channels if space-time codes satisfy so-called generalized orthogonal constraint (GOC) conditions, so that the transmitted signals can be decoupled. In the current space-time codes designed for improving reliability, Orthogonal Space-Time Block Codes (OSTBC) and Restricted Full-rank Single-symbol Decodable Designs (RFSDD) have already satisfied the GOC condition. However, the OSTBC has just very low code rate and the RFSDD has the defect of unbalanced power distribution though its code rate is a little higher than the OSTBC. In this thesis, we propose a new space-time block code with generalized orthogonal constraint (named as GOC-STBC) to achieve full diversity, where the code rate is enhanced much compared with the OSTBC by relaxing the full-rank condition of weight matrices; the full diversity is obtained by using multi-dimensional real lattice constellations; in addition, we systematically analyze the coding gain, code rate and decoding complexity issues of the proposed GOC-STBC. In the meanwhile, the GOC-STBC resolves the unbalanced power distribution problem of the RFSDD. As the coding gain is maximized by using the ready-made algebraic lattice theory, the GOC-STBC has smaller design expense than another typical STBC with lower decoding complexity- Quasi-Orthogonal STBC (QSTBC). Moreover, the performance of GOC-STBC is the same or even better than the QSTBC when the decoding complexity is comparable.Based on the structured GOC-STBC, we investigate its full diversity strategy in closed-loop systems with limited feedback bits and propose an adaptive GOC-STBC design. Because the property of equivalent virtual channel matrices of the proposed code is improved by the precoding codebooks, which are adaptively choosen by 2~4 bits feedback information at the transmitter, the proposed code can obtain full diversity under the simple zero-forcing (ZF) decoding. Furthermore, the performance of ZF decoding is equivalent to that of maximum likelihood (ML) decoding, which is single-symbol decodable (SSD). In addition, the proposed adaptive GOC-STBC is constellation free for both of the aforementioned decoding methods and at the same time the code rate achieve one for any number of transmit antennas. In contrast to the adaptive-QSTBC (A-QSTBC), which is a typical code with rate one and also has full diversity under ZF decoding, the proposed code with just 2 feedback bits has the same performance as the A-QSTBC with perfect feedback; when the proposed code uses more that 2 feedback bits, its performance is better than that of the A-QSTBC with perfect feedback.The coding gain of rate-one QSTBCs is upper-bounded by the minimum Euclidean distance of signal constellations. When the constellation energy is fixed, it is very difficult to increase the minimum Euclidean distance of the current complex constellations (such as 4QAM and 16QAM). To further improve the coding gain of QSTBCs, in this thesis we propose a Cyclotomic Lattice-based QSTBC (CL-QSTBC) where the previous rate one 4×4- and 8×8- QSTBCs are transformed so that they are suitable to use multi-dimensional complex lattice constellations. Then, we prove that the coding gain of the CL-QSTBC is upper-bounded by the minimum Euclidean distance of lattice points and then the upper-bound is reachable. Thus, the coding gain is completely determined by the the minimum Euclidean distance of multi-dimensional lattice constellations. So, we maximize the above minimum Euclidean distance by lattice packing theory, which is equivalent to maximize the coding gain. As a result, the coding gain of the proposed CL-QSTBC will not be constrained by the common complex constellations. According to the comparison of simulations, the CL-QSTBC has higher coding gain than the upper-bound of previous QSTBCs and obtains better performance than the current best QSTBCs but with comparable decoding complexity.The mainest factor which dominates the performance of space-time codes is the diversity gain instead of coding gain. In this thesis, we propose a Joint Orthogonal Space-Time Code (Joint-OSTBC) scheme, where the anti-fading property of multi-dimensional algebraic lattices is combined with the equivalent channel model of OSTBCs, so that the extra time diversity can be obtained by jointly coding multi OSTBC codewords along time domain. In the meanwhile, the full space diversity can be still kept in the above coding scheme. In addition, we derive the Chernoff bound of symbol pairwise error probability and prove that the total diversity gain of the proposed Joint-OSTBC is equal to the product of the number of transmit antennas, receive antennas and jointed codewords. In other words, the Joint-OSTBC with M codewords jointed has the M times diversity gains than the previous OSTBCs. From simulations, we find that the Joint-OSTBC with just double codewords jointed has already significant performance gain over both the QSTBC and even the Golden code, which is the best known code currently, but the decoding complexity is lower much.