Denoised Modified Incoherent Signal Subspace Method for DOA of Coherent Signals

Incoherent signal subspace method (ISM) is ineffective to deal with direction of arrival (DOA) of coherent wideband signals. Although modified methods were proposed to solve this drawback, such as ISM based on modified MUSIC (MMUSIC-ISM), their performance wm seriously deteriorate when the signal-to-noise ratio (SNR) is low. This paper proposes a method to construct the covariance matrix, namely denoised modified ISM (DMISM), which attempts to effectively reduce the influence of stationary white Gaussian noise. Numerical experiments affirm that DMISM outperforms MMUSIC-ISM.

I. Introduction

Incoherent Signal Subspace Method (ISM) [1] is a conventional method for estimating direction of arrival (DOA) estimation of wideband signals. According to ISM, a wideband signal can be considered as the superposition of numerous narrowband signals at different frequencies [2]. Therefore, algorithms dealt with DOA in narrowband coherent signals such as Modified Multiple Signal Classification (MMUSIC) [3] can be applied to each separated narrowband component, then DOA can be obtained by averaging the estimation result of each component. However, the performance of DOA estimation algorithms is largely affected by the signal-to-noise ratio (SNR).