基于NLMS的CDMA盲自适应多用户检测算法研究

多用户检测是抑制DS-CDMA系统多址干扰最有效的技术之一。由于所需的先验知识仪有期望用户的地址码,盲多用户检测技术的研究尤受重视。最小输出能量(MOE)准则被广泛用于盲线性多用户检测。目前已提出的该类检测器多采用LMS或RLS算法。本文则研究基于NLMS算法的盲自适应检测技术,并进一步提出盲自适应变步长NLMS检测器和参数可变的盲自适应变步长NLMS检测器。它们具备很好的收敛速度和跟踪能力,以及较高的输出信干比,同时计算复杂度仅为O(3N)或O(4N),非常适合硬件实现。

NLMS Based Blind Adaptive Multiuser Detection Algorithms for CDMA Systems

Multiuser detection (MUD) is one of the most effective techniques to suppress multi-access interferences ( MAI) in CDMA systems. In research of this scope, much attention is paid to blind MUD in which prior knowledge of only the signature waveform and timing of the interested user is needed. A widely-used cost function to derive adaptive algorithm of MUD is the minimum output energy(MOE) . Most of proposed schemes based on MOE adopted LMS or RLS algorithms. While in this paper, Normalized LMS( NLMS) is introduced to derive MUD algorithm. Furthermore, blind adaptive MUD based on variable step-size NLMS algorithm and blind adaptive MUD based on variable step-size NLMS with adaptive parameter are proposed. They show fast convergence speed and strong tracking ability, as well as high output signal-to-interference ratio. Besides, their computation complexity are only O(3N) or O(4N) , easy to be implemented by hardware.