Acoustic Model-based Blind Channel Estimation In Shallow Water Environments

Underwater acoustic channel is quite harsh because of its unique characteristics:large delay spread due to time-varying multipath propagation causing significant inter-symbol interference for underwater acoustic communications,limited available bandwidth due to the frequency-selective attenuation,and strong spatial correlation,which make the emitted signals severely distorted.The channel impulse response(CIR)should be estimated and compensated in the receiver,especially for those non-cooperative applications including the recovery of communication signals,source localization and signal classification.The source signal is unknown to the receiver or a priori information is difficult to acquire,then the process of estimating CIR is commonly referred as blind channel estimation(BCE).This paper is focus on blind channel estimation(BCE)for underwater acoustic communications,highlighted on utilizing the properties of the acoustic propagation and the acoustic environmental information.Acoustic model-based BCE methods are proposed to deconvolve the CIR from the received signals and to decode the embedded information.Firstly,we implement acoustic inversion method with genetic algorithm on the received signal to estimate the parameters of the sediments and the bottom,which are inconvenient to measure directly.Then we use these parameters to estimate the CIR and its performance is tested with experimental data.For BCE with low frequency signals,a normal mode-based BCE method is proposed the estimate the CIR and to recover the original signal.This method is implemented by localizing the source with a matched-field processing algorithm firstly,then using the physical model of the ocean waveguides to determine the channel impulse response.Its performance is investigated by making comparisons with the mode-based artificial time reversal and the ray-based synthetic time reversal methods,using the array data measured from the SWellEx-96 experiment at four source-array ranges.Cross-correlation coefficient and normalized projection misalignment are adopted to evaluate the performance of the proposed method,and the simulation results for different multipath channels confirm the effectiveness of the proposed method.Due to the sparse recording array and the high frequency source signal,the performance of conventional spatial matched filtering methods will be degraded tremendously.Moreover,the acoustic model might mismatch with the received field as the high frequency waves are more sensitive to the propagating environments.We develop a novel spatial filtering technique which involves the frequency-difference method to process the high frequency signals with a sparse array.Both propagation simulations and measured results from the KAM11 experiment are considered to evaluate the proposed method,which shows that the proposed method is robust to modeling mismatch.To achieve a high resolution of channel impulse response using a small aperture receiving array from high SNR data,this thesis develops a blind channel estimation method using Sparse Bayesian Learning based on the single-input multiple-output model and exploits the sparse multipath structure.Simulation and the TREX04 experimental results confirm the effectiveness of the proposed method.Last but not the least,the automatic modulation classification(AMC)schemes has been applied to recognize the modulated communication signals.These schemes are used to recognize the modulation type of the signals for further demodulation,playing an important part in non-cooperative communication system.We implement two main classes of AMC schemes:likelihood-based(LB)and feature-based(FB)schemes to underwater acoustic communication signals.Their performance has been tested with simulation and experimental data.Moreover,the LB scheme has been successfully applied to Orthogonal Frequency Division Multiplexing(OFDM)modulated signals.Parameters have been recognized from the OFDM signals in the KAM11 experimental data.