ABSTRACT
This chapter extends Unitary Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT) to the R-dimensional case to solve the harmonic retrieval problem. Automatic pairing of the R-dimensional frequency estimates is achieved through a simultaneous Schur decomposition of R real-valued, non-symmetric matrices that reveals their "average eigenstructure." The chapter presents the multidimensional data model and also discusses a related problem, the estimation of shifts of known functions. It explores two alternatives: a covariance and a square-root approach. The chapter discusses the application of R-D Unitary ESPRIT in wireless channel sounding campaigns to estimate the multipath propagation structure of the mobile radio channel. It investigates design and calibration issues for the employed antenna array architectures. The chapter also covers the practical resolution capability of the high-resolution parameter estimation scheme in the multiple dimensions.
