GLRT-based detection-estimation of uncorrelated Gaussian sources in circular antenna arrays

The problem of estimating the number of m independent Gaussian sources and their parameters impinging upon an M-sensor uniform circular antenna array is considered for scenarios that cannot be properly addressed by existing techniques. In particular, pre-asymptotic performance is considered for "conventional" scenarios (m < M), where traditional techniques such as MUSIC experience "performance breakdown". "Superior" scenarios with more sources than sensors (m ≥ M) are also addressed. The method proposed adopts the same generalized likelihood-ratio test (GLRT) methodology that was developed for sparse linear antenna arrays, and approximate Toeplitz embedding for the associated covariance matrix.