Many frequency-recovery algorithms for orthogonal frequency-division multiplexing systems are geared to recover offsets up to half the distance of the subcarriers. If the error is larger, an ambiguity of an integer number of subcarrier spacings is left. This paper develops a blind algorithm for ambiguity resolution that works effectively over frequency-selective radio channels. Simulation results show a sizable improvement over existing techniques, expecially at high signal-to-noise ratios.

This paper introduces a blind (i.e., data independent) algorithm for carrier frequency offset recovery in an orthogonal frequency-division multiplexing (OFDM) receiver operating over frequency-selective fading channels. The main idea behind this algorithm is to exploit the time-frequency-domain exchange inherent to the modulation scheme. Due to this feature, a carrier frequency offset has a similar impact on OFDM as a clock timing offset has in a quadrature amplitude modulation (QAM) system. The scheme we propose is a variant of Oerder-Meyr's (1988) feedforward clock recovery. Its performance is assessed by simulation, and the results are compared to those obtained from Van de Beek-Sandell-Borjesson's (1997) frequency synchronizer, which bears comparable complexity. The new scheme is shown to outperform the latter over frequency-selective fading channels, notably at medium to high signal-to-noise ratios. We also evaluated the efficiency of two different (time domain and frequency domain) offset correction strategies embedded in a particular OFDM receiver.

This paper deals with symbol timing recovery for multirate transmission systems employing wavelets as signaling pulses either at baseband or with linear bandpass modulation. Our analysis is twofold. First, we investigate the impact of clock errors on the bit-error rate of the optimum receiver for a wavelet-based multirate signal. Second, we consider a nondata-aided maximum-likelihood clock synchronization scheme implemented as a multiple delay-locked loop. We derive a linearized loop model and assess its performance in terms of tracking jitter variance. The analytical results are contrasted with those obtained by simulation and with the relevant modified Cramer-Rao bound.