Near-Optimal Noncoherent Sequence Detection for Doubly Dispersive Channels

We propose a scheme for near-optimal sequence detection (SD) of uncoded block transmissions over unknown doubly dispersive (DD) channels. Starting with a noncoherent maximum likelihood (ML) metric that leverages a basis expansion model (BEM) for the channel´s time-variation, we propose an efficient noncoherent SD strategy based on suboptimal tree search with a fast metric update. Our scheme yields performance within a fraction-of-a-dB from ML sequence detection with genie aided channel estimates, and maintains complexity that is only quadratic in the block length.