Research On Sparse Adaptive Turbo Equalization Technique For Underwater Acoustic Communication

With the growing demand for ocean exploration,underwater acoustic(UWA)communications play an important role in it.Thus,extensive research has been conducted in this field.Because of the complex UWA environment,the uneven medium of transmission,and the long delay of the channel,serious inter-symbol interference(ISI)occurs in received signals.Besides,with the band-limited of the channel,the length of training sequences is limited,so obtaining accurate channel state information(CSI)is difficult.To solve the problems above,the research is conducted on the sparse adaptive Turbo equalization.The main contribution of this article is as follows.1.Firstly,the physical characteristics of the UWA channel are introduced,utilizing them establishes the model of the UWA channel,and the influence of its parameters on the channel coding performance is explored.Then,the Turbo equalization based on minimum square error(MMSE)is introduced to solve the ISI problem in UWA coherent communications.However,the traditional Turbo equalization directly assumes that the symbols obey the Gaussian distribution,ignoring the inherent distribution of the transmitted symbols,thus the expectation propagation(EP)frequency domain Turbo equalization(FDTE)with interference cancellation(IC)(EPIC-FDTE)is proposed.On the one hand,to improve the performance of equalization,EP is utilized to estimate the true a posteriori distribution.On the other hand,to effectively reduce the amount of calculation,it converts the received signal to the frequency domain.Finally,according to the results of the simulation,the extrinsic information transfer(EXIT)charts show the EPIC-FDTE has the faster convergence than exact linear equalization(Exact-LE)and other FDTE,and the BER curves show EPIC-FDTE has 1.5d B gain compared with Exact-LE and soft interference cancellation equalization(SICE),and 2d B gain compared with soft interference cancellation FDTE(SIC-FDTE)and FDDF-FDTE when the BER is10^-3.2.Due to the long delay of the UWA channel,traditional channel estimation has relatively high computational complexity due to the operation of inversion,while the adaptive algorithm has a simple structure with low computational complexity.Moreover,utilizing the norm constraint effectively improve the accuracy of channel estimation for the adaptive algorithm.So this article applies selective zeros attracting improved proportional normalized least mean square algorithm(SZA-IPNLMS)to UWA communications.Different from the traditional hard threshold,using the tap proportional to constrain the small tap coefficients can improve the channel estimation accuracy.When the BER is10^-3,here is a 1d B performance gain compared with the traditional norm-constrained channel estimation.Because of the narrow bandwidth of the UWA channel,the length of the training sequence is not too long,for this reason,soft iterative channel estimation is introduced.Because EP can provide reliable symbol a posteriori estimation,a soft iterative channel estimation based on EP a posterior estimation is designed.In addition,to avoid error propagation,a threshold-based channel estimation update scheme is proposed.It is verified by simulation that when the BER is10^-3,there are 1.5d B and 0.8d B performance gains compared with the training sequence-based and traditional soft iterative channel estimation.3.In the UWA multiple-input multiple-output(MIMO)high rate communication,with the ISI and the co-channel interference(CCI)with different sequences,the equalization performance is further challenged.For this reason,a scheme based on EPIC-FDTE called BI-EPIC-FDTE is proposed to obtain the bidirectional gain of equalization.Using 3D-EXIT graph and bit error rate curve analysis,BI-EPIC-FDTE has a faster convergence speed and better reception performance compared with EPIC-FDTE,although a certain amount of computational complexity is sacrificed.In summary,this thesis studies Turbo equalization technology in the UWA environment,optimizes Turbo equalization,and soft iterative channel estimation to improve signal reception performance.With low computational complexity and good performance,the proposed scheme is very suitable for underwater applications in underwater sensor networks,underwater vehicles,and other applications.