A study on the accurate estimation of the number of weak coherent signals

Estimation of the number of incident signals (NIS) is an important problem for array signal processing, such as direction-of-arrival (DOA) estimation. A method for estimating the number of signals without eigendecomposition (MENSE) is superior to algorithms based on the spatial smoothing preprocessing (SSP) and the Akaike information criterion (AIC). The advantages of the MENSE can be achieved by employing the Hankel correlation matrices which can suppress the correlation between the incident coherent signals and the influence of noise. In this paper, we propose a new metric based on the multiplicity criterion to improve the accuracy of the MENSE. Computer simulation results show that the proposed method is superior to the MENSE and the SSP-AIC methods in an environment arriving closely spaced coherent signals.