On the estimation of rapidly time-varying channels

Relying on the basis expansion model (BEM) for rapidly time-varying channels, we propose a novel training-based BEM channel estimation approach. While the existing approach applies a rectangular window to the received sequence and employs a BEM with a period equal to the window length, the novel approach applies a general window to the received sequence and employs a BEM with a period equal to a multiple of the window length. Simulation results show that these extensions can significantly improve the channel estimation performance.