Low-complexity time-varying frequency-shift equalization for doubly selective channels

This paper deals with the problem of designing linear time-varying (LTV) finite-impulse response equalizers for doubly selective channels. Specifically, we consider the frequencydomain representation of the LTV minimum mean-square error (MMSE) equalizer, which relies on the complex exponential basis expansion model of the time-varying channel impulse response and can be implemented as a parallel bank of linear timeinvariant filters having, as input signals, different frequency-shift (FRESH) versions of the received data. We show that such a FRESH formulation allows to derive an effective low-complexity iterative version of the optimal LTV-MMSE equalizer, based on the steepest-descent method. Simulation results show that the proposed iterative equalizer ensures a very satisfactory tradeoff among complexity burden, convergence speed, and performance.