An Adaptive Maximum Likelihood Sequence Estimation Technique for Wideband HF Communications

Performance is presented for a non-real-time (NRT) software coherent binary phase shift keyed (BPSK) receiver using both simulated and recorded HF signals at a bit rate of 2.4 kbps. The receiver consists of: (1) an adaptive matched filter, (2) a maximum likelihood sequence estimator (MLSE) incorporating a modified Viterbi algorithm (MVA), (3) a maximum likelihood (in nature) channel estimator (MLCE) algorithm based on a discrete multipath channel model. For all simulated fixed and time-varying channels, performance is presented in terms of bit error rate (BER) versus average Eb/No. Examples of channel acquisition and tracking are also presented. For the recorded HF signals, performance is presented in terms of BER and availability measures. The implementation of a 2.4 kbps modem suitable for military HF communications applications based on maximum likelihood principles is given preliminary consideration.