Research On Iterative Receiver Algorithms For Underwater Acoustic Communication

Efficient high speed data communication in underwater acoustic channel is still verychallenging due to its long multipath delay and large Doppler shift. Originating from TurboCode, iterative algorithms achieve excellent performance in various communication systems.In this thesis, iterative receiver algorithms for underwater acoustic communication areresearched to combat the severe inter symbol interference (ISI) caused by multipath delay inacoustic propagation.This thesis focuses on the application of turbo equalization in underwater acousticcommunication, mainly studies on the soft input soft output bidirectional decision feedbackequalization algorithm and its application in underwater acoustic communication. Thisalgorithm can effectively combat the error propagation in classic DFE, which is the reasonthe turbo DFE’s performance inferior to LMMSE turbo equalizer. In this thesis, bidirectionalsoft DFE are extended to multilevel modulations, and applied in underwater acousticcommunication. Simulation and pool experiment verified that this new turbo equalizer canachieve excellent performance in severe ISI channels, especially, the low complexityimplemented soft bidirectional DFE is a good tradeoff between performance andcomputational complexity, which could approach the performance of exact implemented softbidirectional DFE with the complexity increasing linearly.This thesis studies the relation between performance degradation of soft bidirectionalDFE and the size of the modulation constellation. Simulations under the condition ofunderwater acoustic channel are conducted to verfy the performance of iterative algorithm inunderwater acoustic communication. EXIT chart are introduced in this thesis to study theconvergence behavior of iterative algorithms. Several turbo equalizers are compared inaspects of performance and convergence speed, which shows that soft input soft outputbidirectional DFE both in exact implementation and low complexity implementation couldapproach the performance of turbo equalizer based on MAP.In this thesis, soft bidirectional DFE are based on the estimated channel, which makeschannel estimation algorithm part of the research of this thesis. Convex regularized RLSalgorithms are applied in sparse channel estimation. Simulation and pool experiment resultsshow that sparse RLS algorithm could achieve performance enhancement in the situation of sparse channel, even in the nonesparse channel, these algorithms do not degrade inperformance.