基于等价多入多出结构的多用户检测方法

Multi user detection method based on MIMO structure

对于一个异步高斯直接序列码分多址系统的多用户检测问题，现有的解相关线性检测方法存在对高维矩阵求逆，计算量大的问题。该文根据系统基带模型的等价多入多出（ Ｍ Ｉ Ｍ Ｏ）结构，基于 Ｍ Ｉ Ｍ Ｏ 和周期时变滤波器的等价性，设计了周期时变的最小均方误差均衡器，并在此基础上设计了用户增益估计的批处理和自适应方法，克服了常规多用户检测方法对用户增益敏感的缺点。仿真结果表明，该文方法的误码率性能与解相关检测方法的基本相同，而计算量减少了，并且减少了解码延迟。

The existing de correlating linear detector for an asynchronous direct sequence code division multipe access Gaussian channel requires the inverse of a high dimension matrix, which is a huge computation burden for a practical system. This paper established the MIMO structure of the system, designed a MMSE (minimun mean sequare error) based periodically time variant equalizer according to the well known equivalence between a vector stationary system and a periodical stationary system. The equalizer can achieve multi user detection with the same performance as the decorrelating detector, but with lower computation cost and also smaller decoding delay. An amplitude estimator (in both batch and adaptive way) was proposed, which makes the approach be near far resistant. The extensive simulations show the efficiency of the approach.