Detection Algorithm Based On Belief Propagation In A Massive MIMO System

MIMO technology in Wireless communication system has become more mature,and has become the key technology of LTE / LTE-Advanced. At present, the maximum number of antennas supported in LTE standards is 8, but the actual spectral efficiency is only 5 bps / Hz. Massive MIMO as the extension MIMO technology, is one of the key technology of 5G. It can increase the number of antennas to tens or even hundreds, it maximized the use of existing resources, and increased the channel capacity and spectral efficiency several times. However, in Massive MIMO system,it has a large number of antennas, complex received signal, huge dimension of channel matrix. These put forward higher requirements to the signal detection algorithm at the receiver, hoping to achieve good performance in lower complexity. It is against this background, this paper further research on belief propagation(BP) algorithm which is suitable for signal detection in Massive MIMO system, hope to achieve higher performance in different scenarios with less complexity degrees.BP as the superior detection algorithm in Massvie MIMO system, has attracted scholars’ highly attention. This paper focuses on how to further reduce the complexity of BP algorithms and how to solve the BP detection algorithm’s poor performance in the high-order modulation. This paper presented three methods to reduce the complexity of BP algorithm: the layered belief propagation, forcing convergence and node selection,and analyzed and compared between them; and also presented GTA-BP algorithms and GTA-BP-SIC algorithm, which are based on the BP algorithm, and simulated analysis of their performance.This paper introduces the research background and status of Massive MIMO system and its signal detection technology, and provides a brief overview of MIMO system model and common Massive MIMO detection algorithms. Then describes the BP algorithm based on Markov random field and the factor graph model and proposed three methods to reduce complexity of belief propagation algorithm. And introduced the principle of each method, and compared them with pure belief propagation algorithm in simulation. Then proposed GTA-BP detection algorithm which is suitable for Massive MIMO system with high order modulation. Discusses in detail how to use the Gaussian tree approximation in BP algorithm, and gives a detailed analysis of the process of the algorithm and simulation. It solves the issues of poor performance in BP algorithm in high order modulation system. Finally, inspired by MMSE-SIC algorithm proposed GTA-BP with successive interference cancellation which is called GTA-BP-SIC algorithms. Described the idea and each step of the algorithm: how to change BP algorithm’s update rules, how to use the optimal ordering. Finally simulated analysis the GTA-BP-SIC algorithm and it showed that further enhance the performance of the GTA-BP algorithm.This paper is focused on belief propagation algorithm and proposed high performance detection algorithms which are based on BP algorithm in a low complexity, It brings some fresh ideas and methods for the issue of the signal detection in Massive MIMO system.