Research On Key Technologies Of Generalized Spatial Modulation Systems

Due to the low complexity and low inter-channel interference,Spatial Modulation(SM)and Generalised Spatial Modulation(GSM)will be the promising candidates for 5G large scale MIMO systems.The goal of this research is to improve the performance of GSM system while maintaining low computational complexity.The main tasks of this research are to develop alternative detection algorithms with low-complexity and simplified active antenna combinations selection algorithms for GSM to improve the performance of the whole system.The Maximum Likelihood(ML)decoder offers optimal BER performance,while it requires extremely high computational comlexity,which makes it difficult to be applied to practical applications.On the contrary,linear decoders can provide complexity reduction,while they can't offer satisfactory BER performance.This paper proposed a joint active antenna combinations and modulated symbol estimation algorithm for GSM system by combining ML and linear decoders.Simulation results illustrated that proposed algorithm can achieve near-optimal performance with low complexity.Furthermore,this paper applied the latest Compressed Sensing(CS)technology to GSM detection,and studied the Orthogonal Matching Pursuit(OMP)and Basis Pursuit(BP)algorithms.An improved CS-based algorithm for GSM detection is proposed.This algorithm combines CS with linear decoders and reduces the dimension of the signal which will be detected.Simulation results illustrated that proposed algorithm can reduce the complexity compared with that of BP-based detection algorithm,while only a little performance loss appears.Additionly,this paper studied the optimal antenna combinations selection algorithm for GSM by maximizing the minimum Euclidean distance between different received vectors,while the computational complexity of this algorithm is extremely high.Thus,in order to reduce the complexity,another two algorithms are proposed,which are based on the criterions of Antenna Correlation and Maximal Volume respectly.Simulation results illustrated that proposed algorithms can offer satisfactory performance with lower complexity.