基于变分自由能的免授权上行SCMA系统活动用户检测与信道估计  

Active user detection and channel estimation of uplink grant-free SCMA system based on variational free energy

作　　者：袁泉[1] 王振永[1] 李德志 郭庆[1] 王振邦 YUAN Quan;WANG ZhenYong;LI DeZhi;GUO Qing;WANG ZhenBang(School of Electronic and Information Engineering,Harbin Institute of Technology,Harbin 150001,China)

机构地区：[1]哈尔滨工业大学电子与信息工程学院,哈尔滨150001

出　　处：《中国科学：技术科学》2020年第7期964-970,共7页Scientia Sinica(Technologica)

摘　　要：基于贝叶斯理论的联合消息传递算法(JMPA)是5G免授权SCMA系统的多用户检测方法.该方法是通过在因子图上进行消息传递直至收敛,从而达到概率逼近的目的.但是由于因子图消息传递存在闭合环路(circle),因此基于JMPA的活动用户检测以及信道估计存在不收敛的可能性,这对于5G通信系统的可靠性影响来说是致命的.本文基于变分法对免授权上行SCMA系统的活动用户检测与信道估计过程进行了分析,并推导得出了求解用户活跃概率以及信道增益后验概率的不动点方程组(fix points equations).对方程组进行理论分析,得到了不动点方程组收敛的充分条件.基于该分析结果对原有的迭代过程进行改进,仿真结果表明,改进后的算法较原有的JMPA算法在性能没有下降的同时具有更好的鲁棒性.Joint message passing algorithm(JMPA) that is based on Bayesian theory is a multi-user detection method for the 5G unlicensed SCMA system. This method allows to pass messages in the factor graph until convergence to achieve probability approximation.However, owing to the closed loop of factor graph messaging, there is a possibility of non-convergence for active user detection and channel estimation based on JMPA, which is fatal to the reliability impact of 5G communication systems. This study analyzes the process of active user detection and channel estimation in the uplink grant-free SCMA system that is based on the variational method and derives the fixed point equations that solve the user active probability and channel gain posterior probability. Through the theoretical analysis of the system of equations, sufficient conditions for the convergence of the system of fixed-point equations are obtained. On the basis of the analysis results, the original iterative process is improved. The simulation results show that the improved algorithm is more robust than the original JMPA algorithm and achieves performance without degradation.

关 键 词：联合消息传递算法 免授权SCMA 因子图 变分法 不动点方程组 

分 类 号：TN929.5[电子电信—通信与信息系统]