ML-Based Iterative Sequence Estimation Without Symbol Timing Recovery

This paper presents a novel iterative scheme of maximum-likelihood (ML) sequence estimation in the absence of timing information. The ML estimates of the symbols are obtained from the oversampling samples of the matched filter output without timing estimation or sample interpolation, eliminating the need for a separate synchronizer. With unsynchronized samples, the detection problem is treated as ML estimation from incomplete data, and the expectation-maximization algorithm is applied to find an iterative solution. The proposed scheme is compared with conventional techniques under different modulation orders. The simulation results indicate that acceptable bit error rate performance can be achieved for short package transmission.