Signal Detedtion Algorithm Based On LLR In MASSIVE MIMO Systems

Massive MIMO system scales up the number of antennas to dozens or even hundreds,which leads to changes in system characteristics,significantly improving system spectral efficiency,channel capacity,and transmission reliability.Nevertheless,at the same time,if we still employ traditional signal detection scheme at the receiver,we will get poor detection performance as well as tremendous computational complexity which is not conductive to hardware implementation.So in this paper,we study low-complexity signal detection scheme for Massive MIMO system.This thesis first compare point to point MIMO and multi user MIMO system model,On this basis,expanding the antenna configuration,leads to Massive MIMO system model,and illustrates the actual advantages of Massive MIMO technology,as well as the actual implementation of the bottleneck.In the aspect of signal detection,traditional methods of linear detection and nonlinear detection are sorted out,and the adaptive communication scene and its shortcomings are pointed out.Research shows that when the number of BS antennas is much larger than the number of the uplink user antennas,simple linear detection is appropriate in term of both complexity and performance.So we focus on the system of which the antenna configuration is almost balanced in both transmitter and receiver.In order to further increase the detection accuracy,we choose MMSE as the starting point and utilize soft information of signal—Log Likelihood Ratio(LLR)to judge final signal.Secondly,based on the previous one-dimension dominant eigenvector search algorithm,an improved multi-dimensional eigenvector search algorithm is proposed.The former assumes that target signal exists in only one direction of dominant eigenvector,and traverse several dominant eigenvectors we choose and decide the optimal direction.On the basis,multidimensional eigenvector search algorithm combines multiple eigenvectors,as a result,not only the detection performance is superior,but also the number of transmitted signal candidates is decreased.According to the statistical result of computation complexity,the former’s complexity is always more than the latter’s.So the modified algorithm achieves better performance and lower hardware implementation complexity.Finally,antenna selection algorithms based on detection error modeling were proposed,which assume that signals of some antennas are absolutely equal to some modulation symbols.The final target signal is the one which minimizes the detection error.If we just choose one antenna,the scheme is called single antenna selection scheme,otherwise,it’s multiple antenna selection scheme.The former scheme has no eigenvalue decomposition and less matrix operation,so the computational complexity is relatively low.The latter’s number of signal candidates is increased exponentially.Simulation result shows that complexity of single antenna selection scheme is superior to adapt to the change of system antenna configuration,and the scheme can achieve low complexity while maintaining detection performance and realize low complexity Massive MIMO detection algorithm.