Equalization of Frequency Selective Channels Using Multirate Signal Processing

Abstract

Multirate signal processing techniques for the equalization of frequency selective communications channels are proposed. It is shown that a multirate equalizer (MRE) is substantially less complex compared to the conventional channel equalizers. This feature makes single-carrier transmission feasible in channels having much longer delay-spreads or wider bandwidths than it was possible earlier. The proposed algorithms do not require transmission of a cyclic prefix. They have both linear and nonlinear versions, wherein the latter employs successive interference cancellation similar to the decision-feedback equalizer (DFE). It is possible to perform decoding prior to interference cancellation. Thus, successive interference canceling (SIC) MRE does not suffer from error propagation if operating on an achievable rate. The SIC-MRE can provably achieve the independent identically distributed information rate of the channel if a Gaussian input alphabet is used. For finite input alphabets, numerical results demonstrate that the single-carrier transmission with the SIC-MRE outperforms multicarrier transmission in terms of throughput. This result holds not only for low-order modulations and relatively narrow bandwidths but also for high-order modulations and wide bandwidths.