Box-Constrained Dichotomous Coordinate Descent Based Detection Technology For Multiuser And Multiple-antenna Systems

With the rapid growth of mobile communication services,in addition to the need to meet higher system capacity,more reliable transmission and greater coverage,future mobile communication systems must pursue low cost and green environmental protection.Therefore,it is urgent to design and efficient,low-complexity algorithms to improve the transmission quality of the system.Multiuser detection(MUD)technology can detect and judge the useful information of each user from a large number of user signals,effectively suppressing multiple access interference and improving code division multiple access(CDMA).Multi-antenna technology,that is Multiple-Input Multiple-Output(MIMO),improves system capacity and performance by using spatial multiplexing and diversity techniques.Multiuser and multi-antenna detection techniques have been widely used to improve spectral efficiency.The maximum likelihood algorithm can provide optimal detection performance,but its complexity grows exponentially with the increasing system.Although many researchers have proposed many suboptimal detection algorithms,there are still significant performance gaps between these algorithms and optimal detection algorithm.Traditional decorrelator and Minimum Mean Square Error(MMSE)detector can not provide good detection performance,and need to inverse the matrix operation.The optimal detection algorithm can be equivalent to solving a constrained equation.However,the computational complexity of solving this problem is usually very high.In this work,some low complexity detection techniques with high performance are proposed.These detection techniques are mainly based on Dichotomous Coordinate Descent(DCD)algorithm.The DCD algorithm uses shifting instead of multiplication and division,so it is attractive for real-time implementation.We propose a detection algorithm based on box-constrained DCD(DCD-A)for unsteady diagonal loading iteration(DCD-DL).Compared with MMSE,DF and PDA,the data results show that the proposed algorithm can not only improve the detection performance of multiuser,but also provide a result close to a single user in the case of high load.In addition,the DCD-DL technology is also tested in the 20×20 and 60×60 MIMO system.The data shows that under the condition of low SNR,when the number of iteration is 5,the group error rate of DCD-DL is significantly lower than the DF,DCD-A and MMSE,close to the performance of PDA,and the average complexity is lower than MMSE and PDA.Under the condition of high SNR,when the number of iteration is 3,the performance of DCD-DL algorithm is better than PDA.DCD-DL provides a better detection performance,but it can only detect real signals.In order to solve the complex signal detection problem,the detection technology is improved.We proposed the complex detection of zero-input two-order box-constrained DCD(two-order DCD-B)and random-input two-order boxconstrained DCD(two-order DCD-BR).The simulation results show that the detection effect of the two-order DCD-B is better than the box-constrained DCD under 4QAM and 16 QAM modulation.In the 60×60 MIMO system and high SNR conditions,the two-order DCD-BR detection performance of the random data as the initial value is better than the two-order DCD-B detection with the zero input as the initial value.