Signal Detection Based On Sphere Decoding In Generalized Spatial Modulation

Multiple input multiple output(MIMO)technology could offer better capacity and reliability without transmission power and channel bandwidth increase,so that it becomes more attracted in wireless communication.Spatial multiplexed technology and spatial modulation technology are two important branches of MIMO.Spatial multiplexed technology could offer a high channel capacity,but the inter channel interference(ICI)and high decoding complexity cannot be avoided at the receiver.Compared to traditional MIMO systems,spatial modulation(SM)exploits the spatial dimension to improve spectral efficiency,so it requires only one RF chain at the transmitter and ICI can be complete avoided with low detection complexity.SM also has some disadvantages,for example the number of transmitting antennas has a specific requirement and the spectral efficiency cannot reach a high level.Spatially Multiplexed Generalized Spatial Modulation(SMu-GSM)takes advantage of spatial multiplexed and spatial modulation,so it can have a high spectral efficiency as well as reduce ICI.In this scheme,code words are assigned to transmitting antenna combinations and each active antenna transmits independent symbols,as in a Spatial Multiplexing system.The paper mainly studies the signal detection algorithm in SMu-GSM system,and focuses on the Sphere Decoding(SD)detection algorithm.Firstly,traditional SD algorithm is studied and analyzed.Then,several improved algorithms are proposed to reduce computational redundancy and accelerate the convergence of radius searching.The main work of this paper is summarized as follows:(1)Analyze the traditional SD detection algorithm Tx-SD and Rx-SD,then apply them to SMu-GSM system.(2)An algorithm called S-SD is proposed to accelerate the convergence of radius searching.(3)An algorithm called T-SD is proposed to reduce computational redundancy,in which the idea of tree searching is used to merge the repeated elements in the set of candidate vectors.(4)An algorithm called P-SD is proposed to further reduce computational redundancy,in which the idea of path searching is used to merge the repeated elements in the set of candidate vectors.Simulation results show that the improved algorithms can effectively reduce computational complexity with no bit error rate(BER)loss compared to traditional SD algorithm.