Statistical Bit Synchronization in Digital Communications

A Kalman filtering algorithm has been applied to the problem of bit synchronization in an M -ary communication system. No synchronizing signal is assumed present; the approach is arranged to determine timing information from the transitions occurring in a pseudorandom sequence of symbols, generated with a fixed but initially unknown bit rate. Procedural steps subdivide naturally into an acquisition phase (block data processing) and a track mode (recursion). The method is supported by sample binary frequency-shift keying (FSK) simulation results, obtained from the output of a 4-pole Butterworth digital filter fed by a random MARK/SPACE sequence plus additive Gaussian noise. Results demonstrate accurate determination of both the bit phase reference time and the bit period. The scope is restricted to high SNR digital communication systems, for which acceptable error rates are obtainable without sophisticated decoding schemes.