An effective direction estimation algorithm in multipath environment based on fourth-order cyclic cumulants

In order to improve the direction estimation performance of noise-suppression and signal-selectivity in multimath environment of mobile communications, we develop a novel multipath direction finding algorithm by exploiting higher order cyclostationarity of communication signals. Firstly, the fourth-order cyclic cumulant matrices of the received signals are constructed and the generalized steering vector is estimated based on EVESPA algorithm, then Forward-Backward Linear Prediction (FBLP) method with subarrays is used to resolve the DOAs of each group of coherent signals of interest (SOIs), which can improve resolution of closely spaced coherent signals in small data length. Part of the array can be arbitrary, and unlike EVESPA algorithm, the total number of signals (SOIs and interferences) impinged on the array can be more than 2K(K-1)/3 with K-sensor uniform linear subarray and a maximum of 2K(K-1)/3 coherent SOIs can be resolved. The simulations show the advantages of the proposed method compared with EVESPA algorithm.