Channel Estimation for OTFS Systems Based on Basis Expansion Model With UPF

Orthogonal Time Frequency Space (OTFS) is an emerging modulation scheme operating in the delay-Doppler (DD) domain, offering superior performance over conventional waveforms like Orthogonal Frequency Division Multiplexing (OFDM) in high-mobility scenarios. However, accurate channel estimation in OTFS remains a critical challenge due to fast time-varying and non-stationary channel characteristics, which significantly degrade system performance. To address this issue, we reformulate the DD domain channel state information (CSI) estimation into a basis coefficient tracking problem using the basis expansion model (BEM) and propose a novel BEM-unscented particle filter (BEM-UPF) algorithm. Specifically, BEM is integrated with UPF to track fast time-varying channels effectively. The unscented Kalman filter (UKF) captures the dynamics of basis coefficients and time-domain correlation coefficients, while the particle filter (PF) approximates their posterior distribution through adaptive resampling, ensuring both accuracy and robustness in high-mobility scenarios. Simulation results demonstrate that BEM-UPF achieves superior bit error rate (BER) performance and robustness against channel estimation errors in high-mobility scenarios.