Grid Evolution for Doubly Fractional Channel Estimation in OTFS Systems

TL;DR

This paper introduces a grid evolution technique for channel estimation in OTFS systems that enhances performance and reduces complexity by adaptively refining the delay-Doppler grid.

Contribution

It proposes a novel grid evolution method that transforms a coarse uniform grid into a dense non-uniform grid for improved channel estimation.

Findings

Improved computational efficiency in channel estimation.

Achieved better trade-off between performance and complexity.

Effective in resolving co-located channel responses.

Abstract

In orthogonal time-frequency space communications, the performances of existing on-grid and off-grid channel estimation (CE) schemes are determined by the delay-Doppler (DD) grid density. In practice, multiple real-life DD channel responses might be co-located within a same DD grid interval, leading to performance degradation. A finer grid interval is needed to distinguish these responses, but this could result in a significantly higher CE complexity when traditional methods are used.To address this issue, a grid evolution method for doubly fractional CE is proposed by evolving the initially uniform coarse DD grid into a non-uniform dense grid. Simulation results show that our proposed method leads to improved computational efficiency, and achieves a good trade-off between CE performance and complexity.