Programmable Blind Adaptive Multivariate Filtering

点击翻译按钮获取中文摘要

In wireless communications, including cellular communication systems, spread spectrum overlay systems, and signals intelligence applications, the degradation caused by rapidly time-varying multipath and unknown cochannel interference can be reduced by adaptive spatial filtering using antenna arrays. This research is pursuing a flexible framework for adapting a spatial filter without using a training signal, array calibration data, or knowledge of spatial characteristics of the desired or interfering signals. It is derived from two conceptually different perspectives: canonical correlation analysis from multivariate statistics, and the conditional maximum likelihood problem for a data-derived training signal obtained from a user-programmable transformation of the received data. The transformation can be programmed by the user to sort and separate the received signals on the basis of the differing degrees to which they exhibit one or more user-selected statistical properties. Many such useful transformations are being examined. The performance of the new method is being investigated analytically and by computer simulations to quantify its capabilities of signal separation, multipath mitigation, and interference rejection. Preliminary results suggest that the new method can converge very quickly to yield signal estimates that are comparable to those obtained by the MSE method that uses perfectly known training signals. More generally, the objective of this research is to develop and evaluate a new and very general approach to programmable blind adaptive multivariate filtering that has broad application to sensor array processing in addition to wireless communications.