Enhanced Target Sensing for OFDM ISAC With L-Shaped Arrays Using Coupled Tensor Decomposition

Abstract

Integrated sensing and communication (ISAC) is widely regarded as a key enabler for future wireless networks. However, ISAC schemes employing orthogonal frequency division multiplexing (OFDM) signals are often limited to range and velocity estimation, whereas existing target sensing schemes for joint range-velocity-angle estimation in the two-dimensional angular domain typically suffer from high array hardware cost and degraded estimation accuracy. In this paper, we propose a target parameter estimation method for OFDM ISAC systems with L-shaped arrays under a coupled tensor decomposition framework. Specifically, we establish a generic OFDM ISAC framework employing an L-shaped receiver that consists of two mutually orthogonal uniform linear arrays (ULAs). By jointly leveraging the echoes observed on the horizontal and vertical axes, we model the received echo signals as two third-order tensor that follows a coupled canonical polyadic decomposition (CPD). Building on this coupled structure, we formulate multi-target parameter estimation as a coupled CPD problem and derive the corresponding identifiability conditions. The analysis reveals that the proposed coupled CPD formulation enjoys more relaxed uniqueness conditions than ordinary CPD approaches that decompose the two axes separately, thereby enabling more reliable parameter recovery. Furthermore, we develop a two stage coupled CPD-based estimation algorithm, where the azimuth and elevation angles, Doppler shifts, time delays, and complex reflection coefficients are sequentially extracted from the estimated coupled factor matrices. Numerical results demonstrate that the proposed method achieves a higher estimation reliability and improved accuracy compared with representative baselines.