Non-Parametric Adaptive Thresholding for Channel Estimation of OTFS-Based 6G Communication Links

Sixth Generation (6G) wireless networks are expected to focus on reliable communication in high-mobility scenarios such as vehicle-to-vehicle communication and unmanned aerial vehicles, including in the Terahertz bands. In such doubly dispersive channels, Orthogonal Time Frequency and Space (OTFS) modulation is expected to provide good performance as it is resilient to high Doppler shifts. The standard channel estimation for OTFS is performed in the Delay-Doppler domain by a fixed thresholding method wherein the threshold requires the knowledge of noise variance. As 6G would demand low latency communication, accurate noise estimation on the fly could be difficult. Hence, our motivation is to develop a non-parametric thresholding technique that is independent of the knowledge of noise variance and channel statistics. In this paper, two less complex adaptive thresholding techniques namely (a) the knee-point method and (b) the one-sided trimmed mean (OSTM) method are proposed. They estimate the channel parameters from the received noisy data symbols by utilizing the sparse nature of the channel matrix in the Delay-Doppler domain. Simulation results show that the OSTM method performs on par with the conventional fixed thresholding method which requires accurate knowledge of noise variance. The knee-point method performs similar to the fixed thresholding method only for high pilot SNR. The OSTM adaptive thresholding technique is efficient and has no performance degradation compared to fixed thresholding, and is best suited for embedded pilot-based channel estimation for OTFS, one of the good contenders for waveforms for 6G.