Research on Adaptive Non-Least Squares Linear DS/CDMA Receivers

为了寻找具有良好性价比的直接序列码分多址（DS/CDMA）接收机设计，我们考虑了自适应线性分接权重滤波器。所考虑的结构是最常规的，但所选择的设计优化准则不是。基于随机近似的概念，我们提出了递归算法的发展，直接最小化诱导的错误概率（平均误码率）。这种方法的新颖之处在于将性能评估措施直接纳入设计过程。本项目的性质包括对诱导线性滤波器误码率表面的可行性研究、开发和理论分析。我们计划首先关注无记忆，时不变的DS/CDMA信道，然后我们的目标是在多径衰落信道中成功运行的相关接收器的开发和性能评估。就所提议的工作的影响而言，我们认识到这条研究路线远远超出了DS/CDMA通信，并可能导致线性和非线性滤波器的非最小二乘（可能是MBER）优化研究。

In search of Direct Sequence Code Division Multiple Access (DS/CDMA) receiver designs with favorable cost versus performance characteristics, we consider adaptive linear tap-weight filters. The architecture under consideration is the most conventional, but the chosen design optimization criterion is not. Based on stochastic approximation concepts, we propose the development of recursive algorithms that minimize directly the induced probability of error (average Bit-Error-Rate). The novelty of this approach is the incorporation of the performance evaluation measure directly into the design process. The nature of this project involves feasibility studies, development, and theoretical analysis of the induced linear filter Bit-Error-Rate surfaces. We plan to focus initially on memoryless, time-invariant DS/CDMA channels and then we aim at the development and performance evaluation of pertinent receivers that operate successfully in multipath fading channels. In terms of the impact of the proposed work, we recognize that this line of research goes well beyond DS/CDMA communications and may lead to studies for non-least squares (possibly MBER) optimization of linear and non-linear filters.