块处理Kalman算法及其在系统辨识中的应用

Kalman滤波器能以自适应的方式精确地辨识一个系统,但在通信等应用领域,Kalman系统辨识的意义往往在于求其逆系统并用作均衡等后级处理。由于求逆算子的敏感性,估计结果中小的误差也可能造成均衡器性能的显著损失。本文在研究Kalman估计的各种噪声抑制结构后提出一种块处理Kalman算法,该算法在静态或缓变系统中显示出比传统逐符号处理Kalman算法更好的估计误差的性能(误差功率与块处理长度成反比),而且其结构简单,易于实现。

Block Processing Kalman Filtering and its Application in System Identification

Kalman filter can adaptively identify a system precisely.In the fields such as telecommunication,however,Kalman system identification is valued in concern with the succeeded system inversion stage for equalization.Due to the sensitivity of inversion operator,minute errors in the estimation can incur considerable equalization performance degradation.After studying various error inhibition structures of the Kalman estimator,a block-processing Kalman algorithm is proposed,which shows better estimation error performance than classical symbol-processing Kalman algorithm(the estimation error power is inversely proportional to the block length),with simple structure and implementability.