Reliable DOA Estimation in Spatially Correlated Noise With Nonuniform Arrays
Fauzia Ahmad, Piya Pal, Tongdi Zhou · 2021 55th Asilomar Conference on Signals, Systems, and Computers · 2021
In this paper, we consider source direction-of-arrival (DOA) estimation with nonuniform arrays in the presence of unknown spatially correlated noise. Subspace-based techniques and sparse recovery methods typically proceed under the assumption of white noise. However, in many applications, this assumption is rarely valid and the noise is, in fact, correlated along the array. We investigate the offerings of the "correlation-aware" LASSO (Co-LASSO) method for DOA estimation under spatially correlated noise. Co-LASSO does not require strong assumptions on the distribution or correlation of the noise, and its theoretical performance guarantees can be established for any (bounded) noise pattern. We apply Co-LASSO to a finite number of signal snapshots received with a co-prime array under spatially correlated noise, modeled as a first-order autoregressive process, and compare its performance with coarray-based MUSIC algorithm for low and high noise spatial correlations. The results illustrate that the Co-LASSO provides superior DOA estimates as compared to coarray-based MUSIC, provided the number of sources does not approach the maximum level of recoverable sparsity.