基于子空间跟踪的盲多用户检测算法

A Better Blind Multiuser Detection Method Based on Subspace Tracking

基于修正的最小均方误差(MMSE)准则,提出了一种适用于多径衰落信道中的同步DS-CDMA盲自适应多用户检测算法。该方法使用了一种改进的信号子空间跟踪算法,将内部模型法则与PAST算法相结合,并在每一次自适应迭代中对信号子空间正交化,解决了传统算法因引入特征值估计误差而导致检测性能下降的问题,提高了算法的跟踪精度和鲁棒性,且计算复杂度较低。仿真结果验证了本文算法的有效性。

Aim. Existing algorithms are, in our opinion, somewhat unsatisfactory in one or more of the following three respects： （1） slow convergence, （2） insufficient robustness, （3） computational complexity. We now present what we believe to be a better algorithm. In the full paper, we explain our algorithm in some detail; in this abstract, we just add some pertinent remarks to listing the two topics of explanation. The first topic is： the model of direct-sequence code-division multiple-access （DS-CDMA）. Its two subtopics are： the routine subspace model （subtopic 1. 1） and the subspace model of modified minimum mean-square error （MMSE） multiuser detector （MUD） （subtopic 1. 2）. In subtopic 1.2, we point out that a subspace tracking algorithm that uses strong orthogonality can greatly reduce the errors induced by subspace estimation. The second topic is： the MMSE MUD based on our improved subspace tracking algorithm. In this topic, we prove that our improved subspace tracking algorithm orthogonalizes the parameters of signal subspace during each iteration and has global convergence. Finally, we simulate our subspace tracking algorithm; the simulation results, shown in Figs. 1 through 4 in the full paper, demonstrate preliminarily that our algorithm can raise the output signal-to-interference-plus-noise ratio （SINR） by about 7. 7% and reduce the bit error rate （BER） by about 16. 7%, compared with the projection approximation subspace tracking （PASTd） algorithm. The results also verify preliminarily that our algorithm is better in the overall sense： it is not better in all three respects--convergence, robustness, and computational complexity--but is better in that it dose not exceed the allowable limit in any one respect.