Multiple Invariance ESPRIT

Abstract

ESPRIT is a recently developed technique for high-resolution signal parameter estimation. For the specific problem of direction-of-arrival (DOA) estimation, a decrease in computational complexity of orders of mag nitude over other high-resolution methods is achieved by exploiting an int'driance structure designed into the sensor array. Previously, ESPRIT took advantage of only one such invariance per dimension of the parameter vector. The ubiquitous uniform linear array is an example of a sensor array possessing many such invariances, and the question of which invariance to use naturally arises. In this paper, ESPRIT is extended to address the problem of exploiting all the invariances simultaneously. The nonlinear multiple invariance ESPRIT algorithm is derived and compared via simulation to various standard ESPRIT solutions possible for arrays with multiple invariances.