Cycle-slip detection using soft-output information

To realize their full coding gain potential, iterative decoders need to operate in low signal-to-noise ratio (SNR) regions where timing recovery devices experience cycle-slips. This paper develops a binary Bayesian cycle-slip detector by using soft-output information. A practical decision rule is derived utilizing a series of approximations. Approximate analytic expressions for the probabilities of cycle-slip detection and false alarm are obtained. They guide us in choosing detector parameters that will meet a desired performance specification. The applicability of this method is shown on artificially generated waveforms.