Tracking performance of ML-oriented NDA symbol synchronizers for nonselective fading channels

In this paper we investigate the tracking performance of two maximum-likelihood (ML)-oriented nondecision-aided (NDA) symbol synchronization algorithms, operating on a noisy M-PSK signal (M > 2) transmitted over a time-varying nonselective Rician fading channel. The first algorithm is nonchannel-aided (NCA): timing is acquired independently of the channel gain estimation. The second algorithm is channel-aided (CA): it makes use of (an estimate of) the instantaneous channel gain. As the channel gain estimator requires timing information, use of the CA algorithm implies that the timing and the channel gain must be acquired jointly. Both algorithms are suitable for fully digital implementation, and have a similar complexity. For Es/N0 values of practical interest, we show that the NCA algorithm is to be preferred over the CA algorithm.