Tensor Decomposition-Based Sensing for Hybrid-Field Movable Antenna ISAC Systems

Abstract

This paper investigates sensing parameter estimation for movable antenna (MA) integrated sensing and communication (ISAC) systems under hybrid-field propagation. Specifically, to characterize the coexistence of near-field spherical wavefronts and far-field plane wavefronts, a hybrid-field sensing signal model is first established. The received observations are then organized into a third-order tensor, whose factors correspond to the receiver-side hybrid-field response, delay-dependent frequency response, and transmitter-side hybrid-field response. Based on this tensor representation, an alternating least squares (ALS)-based method is proposed to recover the underlying factor matrices. The associated sensing parameters, including delay, range, angle-of-arrival, and angle-of-departure, are subsequently extracted from the structured factors. Finally, simulation results demonstrate that the proposed tensor decomposition-based method can achieve superior delay and angle estimation accuracy compared with baseline algorithms.