Fourth-order cumulant matrices for DOA estimation

A novel class of fourth-order cumulant matrices for direction-of-arrival (DOA) estimation is presented. The suitable adjustment of the components of the cumulant matrices makes it possible to preserve the structures which can be directly applied to the well developed algorithms such as MUSIC and ESPRIT and make these algorithms more applicable to DOA problems. By using the proposed cumulant matrices along with MUSIC or ESPRIT algorithms, a uniform linear array with L sensors can be used to resolve 2L-2 DOAs; twice the maximum number of DOAs that can be estimated by the covariance-based MUSIC or ESPRIT. Besides, the cumulant matrix pencil, which is structured like the covariance one used in ESPRIT, can be formed to estimate DOAs for general arrays which may not consist of matched sensor doublets. Both the order and the components of the proposed matrices can be adjusted to improve the performance. Computer simulations are carried out to illustrate the potential of the proposed cumulant matrices