Research On Low-Complexity Detection Techniques For Large MIMO Systems

The development of the society demands increasingly high requirements on wireless communication technology. Multi-Input Multi-Output(MIMO) technology is increasingly applied to the mainstream standard of wireless communication technology because of its advantage to achieve spatial multiplexing gains and diversity orders and its ability of improving the channel capacity and communication channel link reliability. Large MIMO system gradually becomes a research hotspot due to its configurations of tens to hundreds of antennas in both base stations and communication terminals which bring improvments of data rates and diversity orders. However detection technology has become the bottleneck of the implementation of Large MIMO systems because of the increase of the detection complexity and the degradation of the detection performance due to the use of conventional MIMO detection technologies in Large MIMO systems with tens to hundreds of antennas. The research reported in this paper is concerned with seeking low-complexity high-performance detection techniques which are suitable for detection in Large MIMO systems.First the mathematical model of MIMO system is given from the perspective of the whole system, and then analysis of channel capacity of MIMO system in different scenarios is given to illustrate the advantages of MIMO technology. Some conventional detection algorithms for MIMO system are analyzed in detail, including ML, MF, ZF, MMSE, ZF-SIC, MMSE-SIC and ZF-OSIC, MMSE-OSIC. A brief introduction of LR-Aided linear detection algorithm and SD is given as well. Then low-complexity detection algorithms based on tabu search method for the signal detection in Large MIMO Systems are analyzed, including the RTS(Reactive Tabu Search) algorithm, LTS(Layered RTS) algorithm, and ZF-OLTS(The Ordered Layered Tabu Search Algorithm with ZF initial solution vector). Finally, MMSE-OLTS algorithm with two improvements on ZF-OLTS is proposed, one of the improvements is using MMSE solution vector instead of ZF solution vector as the initial solution vector of the detection algorithm, and the other is seeking a lower-complexity sorting method without the reduction of the detection performance.Through the study of the simulationson MMSE-OLTS, ZF-OLTS, MMSE-OSIC, ZF-OSIC, MMSE and ZF, it can be seen that the improved MMSE-OLTS algorithm can greatly reduce complexity with higher detection performance than that of ZF-OLTS and other algorithms under the same simulation condition. In addition, the simulation results show that the detection performance of MMSE-OLTS detection algorithm will increase with the increasing antennas, showing the characteristics of MMSE-OLTS suitable for Large MIMO systems. What is more, the MMSE-OLTS algorithm has more advantages than conventional detection methods in the scenario of higher order modulation.