A novel receiver architecture for single-carrier transmission over time-varying channels

Abstract

Abstract-In this paper, we present a single-carrier transceiver for rapidly time-varying channels, where the equalization step is implemented in the frequency domain. When the channel abides with both fast fading and severe inter-block interference, our equalizer relies on a band approximation of the frequencydomain channel matrix to maintain low complexity. We will show that the band approximation error can be associated in the time domain to a critically-sampled complex exponential basis expansion modeling error. Based on this property, we propose a novel receiver architecture that extends the original data model by inserting zeros at the receiver. The resulting effective channel can be characterized by an oversampled complex exponential basis expansion model, which has a considerably reduced modeling error compared to the critically-sampled one. In other words, the band assumption that is essential to the equalizer will be made more accurate and thus the equalization performance can be improved.

Low-complexity frequency-domain turbo equalization for single-carrier transmissions over doubly-selective channels

“…This is caused by the larger band approximation error in SC systems with respect to OFDM systems, and by the approximations introduced to simplify the proposed equalizer. In addition, further performance improvements for SC systems could be obtained by employing channel extension techniques [10]. …”

Equalization of Time-Varying Channels

“…6.2(b), in which case any of the fast linear equalization techniques described in Section 6.3.3 can be used to estimate a from y. For example, in [TL08], Tang and Leus proposed a method to equalize a single carrier system using the OFDM fast LMMSE technique [RBL05].…”

Equalization for multi-scale multi-lag OFDM channels

“…A second approach can be described as follows. Method 1 is adopted by many works [3]- [5], [14] for narrowband time-varying OFDM systems, for which the corresponding channel matrix is characterized with a relatively small ICI that is at the same time located in the vicinity of the main diagonal. In that case, Method 1 renders a very tight approximation.…”

“…For instance, a least-squares (LS) equalizer requires an inversion of the channel matrix, which incurs a complexity that is cubic in the number of subcarriers. In a narrowband system, where a similar situation of time-varying OFDM channel occurs, the channel matrix is assumed to be banded to enable a lowcomplexity equalizer [3]- [5]. In a wideband system, a sparse approximation might be more suitable to simplify the channel matrix, and accordingly, we propose to use the conjugate gradient (CG) algorithm for equalization [6], [7], which can benefit maximally from the channel sparsity.…”