Symbol synchronization algorithms for M-PSK operating at one sample per symbol

Carrier-dependent and carrier-independent clock synchronization algorithms requiring only one sample per symbol are considered. The connection between symbol synchronization and "blind identification" of the impulse response of a discrete-time channel, driven by an unknown independent identically distributed (i.i.d.) data sequence, is analyzed. It is shown that this approach leads to a couple of algorithms already proposed. As carrier-independent synchronizers (for M>2) are twice as noisy as carrier-dependent ones, the interaction between an algorithm for carrier phase estimation, well suited to digital receiver implementation, and carrier-dependent symbol synchronization is analyzed. Much of the loss can be recovered, for all values of M, without harmful interaction between the clock and phase estimators