大规模MIMO系统中软输出信号检测方法

发布时间：2018-02-27 08:49

本文关键词： 大规模MIMO 多用户检测迫零软输出 Lanczos算法　出处：《华中科技大学学报(自然科学版)》2017年04期 　论文类型：期刊论文

【摘要】：为解决在上行多用户大规模多输入多输出(MIMO)系统中,迫零(ZF)检测算法可取得近似最优性能,但ZF检测算法涉及复杂度高达O(K3)的矩阵求逆运算(其中K为用户数)的问题,基于Lanczos算法提出一种软输出信号检测方法,避免了高阶矩阵求逆运算,使复杂度由O(K3)降为O(K2).该方法为了计算软输出信息——对数似然比(LLR),通过对基于Lanczos算法的迭代计算解向量的过程进行分析,给出了一种低复杂度的LLR近似计算方法.仿真结果表明:提出的软输出信号检测方法的误比特率(BER)性能与计算复杂度均优于基于Neumann级数近似的信号检测算法,同时,最多仅需5次迭代就可取得逼近于ZF检测算法的性能.
[Abstract]:In order to solve the problem that in uplink multi-user large-scale multi-input multi-output MIMO-system, the zero-forcing ZF-based detection algorithm can achieve approximate optimal performance, but the ZF detection algorithm involves matrix inversion operations (where K is the number of users) with a complexity of up to OK3. Based on Lanczos algorithm, a soft output signal detection method is proposed, which avoids high order matrix inversion. In order to calculate the soft output information-logarithmic likelihood ratio (LLR), the process of calculating the solution vector by iteration based on Lanczos algorithm is analyzed. In this paper, a low complexity LLR approximation calculation method is presented. The simulation results show that the performance and computational complexity of the proposed soft output signal detection method are better than that of the signal detection algorithm based on Neumann series approximation. The performance of ZF detection algorithm can be achieved by only five iterations.
【作者单位】：重庆邮电大学通信与信息工程学院;
【基金】：国家高技术研究发展计划资助项目(2014AA01A705) 国家科技重大专项资助项目(2015ZX03001033-002) 重庆市基础与前沿研究计划资助项目(cstc2015jcyjA40040)
【分类号】：TN919.3

【参考文献】