APPA symbol timing recovery scheme for turbo codes

A timing recovery system is proposed based on maximum likelihood estimation, suited to coding systems employing iterative soft-in/soft-out decoding like turbo codes, as this method utilises the soft-output from the decoder to assist in the timing estimation. Timing recovery is combined with decoding through an iterative technique in contrast to the traditional approaches where timing recovery is carried out prior to decoding. The timing estimate is one of the zeros on the S-curve (the log-likelihood function derivative) corresponding to the maximum of the log-likelihood function. It is resolved using an interpolation algorithm based merely on four samples and avoids generating the whole S-curve for every block. The synchroniser has effectively no acquisition period. The performance of the estimator is analysed in terms of the mean and the variance of the timing estimate, and it is shown that it provides a significant improvement over the non-data aided (NDA) method and approaches the results obtained from the data aided (DA) method (which ideally assumes the data is known and available all the time) after 5 iterations.