Enhanced Target Sensing for mmWave MIMO-OFDM ISAC Systems Using Tensor Decomposition and Virtual Array

Millimeter-wave (mmWave) integrated sensing and communications (ISAC) equipped with multiple-input multiple-output (MIMO) antenna arrays can simultaneously deliver high data rates and sensing performance. However, the angular resolution and estimation accuracy of existing mmWave MIMO-orthogonal frequency division multiplexing (OFDM) ISAC schemes remain constrained by the limited physical aperture. In this work, we propose an enhanced target sensing framework that fuses virtual array synthesis and tensor processing for mmWave MIMO-OFDM ISAC systems. Firstly, we model the target echo of the OFDM-ISAC signal as a third-order tensor, which includes the angle, delay, and Doppler parameters of the targets. Then, we apply CANDECOMP/PARAFAC decomposition to the third-order tensor to obtain factor matrices containing the target parameters. By exploiting the fact that the target shares the same physical angle relative to both transmit and receive arrays in a monostatic sensing system, the Kronecker product is employed to combine the transmit and receive factor matrices, which enables the construction of a virtual array for extending the array aperture. Numerical results demonstrate that the estimation accuracy and spatial resolution of the proposed method are superior to state-of-the-art baselines, due to our specially-tailored algorithm design and virtual aperture construction.