Iterative APPA Symbol Timing Recovery for Turbo-coded Systems

Abstract

This paper examines the problem of symbol timing recovery and decoding in turbo-coded systems. The large coding gain of Turbo-codes enables reliable communications at very low signal-to-noise ratio (SNR). However, the application of turbo coding exacerbates the problem of timing recovery due to its very low operating SNRs. Therefore we propose a new concept: ‘a-priori probability aided’ symbol timing recovery, which is suited to coding systems employing iterative soft-in/soft-out (SISO) decoding such as Turbo Codes, where the log-likelihood ratio obtained from a Maximum A Posteriori (MAP) decoder (as used in turbodecoding) is used to aid estimation in an iterative synchroniser/decoder. This method jointly solves the timing recovery and decoding problem and the synchroniser has effectively no acquisition period. We illustrate this in detail in a turbo-coded BPSK system on AWGN channel. We analyse the performance of the estimator in terms of the mean and the variance of the timing estimate, and demonstrate a substantial performance advantage over conventional non-data-aided (NDA) method. It is also shown that this method can approach the performance of Data-Aided (DA) method without using additional pilot symbols.