Iterative estimation of doubly selective underwater acoustic channel using basis expansion models

An iterative channel estimation scheme is proposed for orthogonal frequency division multiplexing (OFDM) systems over doubly selective underwater acoustic channels. The channel estimator is developed based on the basis expansion model (BEM) with two different types of basis functions, complex exponential (CE) basis and discrete prolate spheroidal sequences (DPSS). Considering the different Doppler characteristics for each cluster, we use cluster-specific parameters in the BEM for the most significant taps, such that the number of unknowns during channel estimation can be considerably reduced. The frequency domain received samples are expressed in terms of the model coefficients of the most significant taps. The channel estimator operates in an iterative, decision-directed fashion. At the first iteration, it utilizes only pilot symbols. After the first iteration, the estimator also uses the symbol decisions produced by a linear minimum mean squared error (LMMSE) block equalizer, in addition to the pilot symbols. We verify the bit-error-rate (BER) performance of the OFDM system over a simulated underwater acoustic channel based on Monterey-Miami Parabolic Equation modeling. It is shown that the LMMSE equalizer with the BEM-based channel estimators and the one-tap equalizer with perfect channel state information have similar BER performances. Both the proposed BEM-based channel estimation schemes perform much better than the basic matching pursuit method, a classic sparsity-aware channel estimation technique.