Tensor Decomposition Based Passive Localization for MIMO-OTFS System in High-Mobility Scenarios

In comparison to the orthogonal frequency division multiplexing (OFDM) signal, orthogonal time frequency space (OTFS) modulation is more suitable for high-mobility localization scenarios, such as autonomous vehicles. This paper proposes a tensor-based algorithmic framework for passive localization in MIMO-OTFS system, where the multipath components (MPCs) are exploited by sparse temporal-Doppler-spatial features of the received channel impulse response. Specifically, we propose a low-rank fourth-order channel tensor to characterize temporal-Doppler-spatial features, leveraging the distinctiveness of tensor decomposition to distinguish coherent multipath signals. This is followed by a sophisticated joint estimation technique. Subsequently, we establish the correlation between Doppler frequency shift (DFS) and target velocity, drawing upon the principles of relativistic Doppler effects in scenarios involving one more receivers existing in the sensing area. Moreover, we derive the Cramér-Rao lower bound (CRLB) to assess the performance of MPC parameter estimation and positioning. Our computational analysis reveals that the algorithm we propose exhibits superior accuracy in parameter estimation compared to cutting-edge methods currently available.