Per-symbol ICI mitigation for low-dimensional doubly dispersive channels using pilot restoration

In high-mobility scenarios the channel behavior is Doubly Dispersive with time variation reaching up to multiple cycles over an OFDM symbol period. The latter warrants pilot placement at the individual symbol level, which results in reduced throughput. Even though multiple cycles of fluctuation over a symbol period is a severe case, the dimensionality of the description of the time variation is manageably low. Given knowledge of a bound on the speed of objects affecting the channel a parsimonious basis expansion model (BEM) approximation (based on prolate spheroidal sequences) of the channel behavior can be attained. Leveraging the low dimensionality, an optimization problem is formulated with the objective of restoring the per symbol pilots to their original values, with the design variables being the BEM coefficients embedded in a zero-forcing equalization matrix. Our proposed Pilot Restoration (PR) method provides significant improvement over the least squares estimation method when a small set of subcarriers is allocated for pilot clusters.