Cumulant-based subspace tracking

We have developed a cumulant-based preprocessing method that can be combine d with any fast rank and subspace tracking algorithm to achieve low computa tional complexity. We applied this method to Virtual-ESPRIT to track moving targets; its generalizations to other applications of array signal process ing are straightforward Our algorithm is suitable for a slowly time-varying environment. Although it has slower convergence than a correlation-based t racker, due to the nature of cumulants, it preserves all advantages of cumu lant-based array signal processing methods. Simulation results show this me thod works well for both rank and target tracking in the presence of Gaussi an noise. (C) 1999 Elsevier Science B.V. All rights reserved.