Adaptive local exponential bases for estimation of linear time-varying channels

A new non-parametric method is presented for the estimation of linear time-varying (LTV) channels. A basis of smooth local exponentials, which are approximate eigenfunctions of underspread LTV channels, is chosen to select time intervals during which the time variation of the channel is small. The time-varying transfer function is expanded in the basis of smooth local exponentials to obtain an estimate of the time-varying impulse response. Simulations for mobile radio channels demonstrate robust tracking of the time-varying channel for a large range of mobile speeds