Robust Optimal Filtering Method for Cyclostationary Signals

Abstract

This paper addresses the optimal filtering method of cyclostationary signals in the presence of interfering signals and non-Gaussian symmetric alpha-stable distribution impulsive noise. By employing spectral correlation analysis method, the problem of optimally filtering the cyclostationary signals was discussed in the reference. Based on the max-output signal-to-noise ratio criterion, the analytic expression for the cyclic correlation matched filter is derived, which is just the same as the conventional matched filter except for different signal model. But the performance of conventional cyclic matched filter suffers from severe degradation under impulsive noise. Although the fractional lower-order statistics based filtering methods are robust to impulsive noise, the interfering signals which are present at the same time and also occupy the same spectral band as the source signal can be particularly problem. By fusing cyclostationarity and covariation, we introduce a new type of cyclic matched filter. The new method exploits fractional lower-order cyclostationarity property of signals with fractional lower-order cyclic statistics. Simulation results indicate that our new method is highly tolerant to interference, Gaussian and impulsive noise compared with conventional cyclic matched.