Tensor Decomposition-Based Beamspace Esprit Algorithm For Multidimensional Harmonic Retrieval

Beamspace processing is an efficient and commonly used approach in harmonic retrieval (HR). In the beamspace, measurements are obtained by linearly transforming the sensing data, thereby achieving a compromise between estimation accuracy and system complexity. Meanwhile, the widespread use of multi-sensor technology in HR has highlighted the necessity to move from a matrix (two-way) to tensor (multi-way) analysis. In this paper, we propose a beamspace tensor-ESPRIT for multidimensional HR. In our algorithm, parameter estimation and association are achieved simultaneously.