Study On Signal Detection Algorithms For Large-dimension MIMO And Spatial Modulation

Multiple-input multiple-output(MIMO) technique can achieve a higher spectral efficiency and/or higher reliability. Recently, during the evolution of the 5G technique, the large-dimension MIMO equipped with dozens and even hundreds of antennas at the transceiver, also called large scale MIMO or large MIMO, has attracted many attentions from academic and practical industry. Moreover, it is well known that the frequency selective fading MIMO has also an equivalent large MIMO presentation. Spatial modulation(SM) is a novel multiplexing scheme proposed recently which can achieve higher energy efficiency at the transmitter than the conventional MIMO scheme. For both large MIMO and SM, an important challenge is to design efficient signal detection algorithm.For the spatial-multiplexing large MIMO, the reactive tabu search(RTS) algorithm and the improved layered tabu search(LTS) algorithm are first studied in this thesis. RTS and LTS can approach the optimal detection performance for low-order and high-order modulations, respectively. However, they require a large number of search iteration, which results in high computational complexity. Next, this thesis studies the lattice reduction(LR) aided linear detection algorithm. Through LR, the equivalent channel matrix becomes more orthogonal, which can improve the linear detection performance. Compared with RTS and LTS, LR achieves a better performance-complexity tradeoff..For the frequency selective fading MIMO channel, this thesis first studies the application of RTS algorithm in the cyclic prefix(CP) single carrier modulation. Then, the detection algorithm for the zero padding single carrier modulation is researched. Specifically, utilizing the Toeplitz property of the equivalent channel matrix, three algorithms, bidirectional decision algorithm, M-algorithm, and factor graph-based Gaussian approximation method, and their combinations are proposed.For SM, an improved detection algorithm is first proposed. The improved algorithm redefines the reliability metric in the conventional ordered block minimum mean square error(OB-MMSE) method, and achieves better performance especially for the quadrature amplitude modulation. Moreover, for an extension scheme of SM, called generalized space-time shift keying, two low-complexity detection algorithms, orderedsuccessive sphere decoding and pattern-aware sequential Monte Calro method, are proposed.