Interference Mitigation in Blind Source Separation by Hidden State Filtering

Abstract

Radio frequency (RF) sources are observed by a uniform linear array (ULA) in the presence of interference. The activity of the sources of interest is sparse, intermittent and assumed to follow a hidden Markov model (HMM). The interfering jammer is active during the entire period of observation. Blind Source Separation (BSS) is performed using direction of arrival (DOA) as criterion of separating the sources as well as the jammer. It is shown that an interfering jammer has a deleterious effect on the performance of the BSS. Leveraging the HMM activity model of the sources, a method is proposed to mitigate the effect of an interfering jammer. The proposed method is essentially a state filtering technique, and it is referred to as Hidden State Filtering (HSF). Two different HSF methods are introduced and compared. The HSF concept is extended to include estimating the HMM model parameters from the observed data. Numerical results demonstrate that the proposed approach is capable of mitigating the effects of interference and enhance source separation.