基于预处理共轭梯度法的多用户检测方法

Research on Multi-User Detection Based on Preconditioned CG Method

针对准同步CDMA（码分多址）系统中解相关检测运算复杂度高以及常规CG（共轭梯度）法收敛速度慢的问题,提出了SSOR-PCG（对称超松弛预处理共轭梯度）模型求解方法。通过SSOR预处理降低了扩频码矩阵的条件数,从而提高收敛速度,减少了迭代次数。实验结果表明,本文提出的方法相对于解相关检测,运算复杂度大大降低,而误码率与解相关检测器相当;相对于共轭梯度法,收敛速度更快,有效减少了迭代次数。

Multi-user decorrelation detection computation has a high level of complexity and the CG（Conjugate Gradient） algorithm slows convergent speed in an asynchronous CDMA（Code-Division Multiple-Access） system.To solve these problems,a multi-user detector based on SSOR（Symmetric Successive Over-Relaxation） preconditioned CG algorithm is proposed in this paper.The method uses SSOR preconditioner to reduce the number of conditions of the coefficient matrix and speeds up the convergence speed.Computer simulation shows that method achieves the performance of decorrelation receiver while requiring much lower complexity.Moreover,the rounds of iterations are reduced and convergence is much faster than the conventional CG method.