Research On Single Carrier Underwater Acoustic Communication Technologies In Impulsive Noise Environment

In underwater acoustic（UWA）communication,the transmission signal is often affected by impulsive noise.Impulsive noise is caused by natural or man-made factors,and its presence makes the background noise no longer obey the Gaussian distribution,which will have an impact on all aspects of UWA communication.Common UWA communication and signal processing methods are usually derived under Gaussian noise.The presence of substantial impulsive noise makes the performance of traditional methods severely degraded and has an impact on all aspects of UWA communication.This paper will focus on several key issues such as synchronous signal detection,channel estimation and pre-processing channel equalization techniques in the impulsive noise to realize robust single carrier UWA communication.Firstly,the detection technology of synchronous signal affected by multi-path channel in impulsive noise is studied.A two-step method is proposed to deal with this problem.In the first step,based on the sparse physical characteristics of the UWA channel,the robust orthogonal matching pursuit（ROMP）algorithm is used to estimate the channel structure and reconstruct the received signal based on the compressed sensing theory.If the received signal contains the transmitted synchronization signal,the reconstructed signal matches the received signal;if the received signal contains only impulsive noise,the two are mismatched;in the second step,the channel structure information obtained in the first step is used to design a parametric enhanced log-likelihood ratio type detector and a nonparametric enhanced pseudo-correlation detector.The robustness of the proposed detectors is verified in different impulse noise under multipath channels.Also,the detection probability is significantly improved after using the multipath information.Second,the sparse UWA channel estimation algorithm is studied.In order to improve the performance of channel estimation under norm constraints,a complex domain adaptive penalized least mean square algorithm（CAP-LMS）is proposed in this paper.It groups channel taps,and for large-amplitude taps,no norm constraint is imposed to reduce the estimation error;for small-amplitude taps,the norm constraint exists to improve the convergence speed.Simulation and Arctic measurement data verify the superiority of CAP-LMS over other channel estimation algorithms.To resist the effect of impulsive noise,two robust channel estimation methods are proposed next in this paper,namely,the variable forgetting factor l₁₀ recursive least sign algorithm(VFF-l₁₀-RLSA)and the variable forgetting factor l_{2 0} recursive least sign algorithm（VFF-l_{2 0}-RLSA）.They utilize the ₁l norm of the estimation error as a cost function to attenuate the adverse effects of impulsuve noise,and the mixture l_{1 0} and l₂₀ norms represent the sparsity of the channel.Meanwhile,the time-varying forgetting factor and the regularization parameter can enhance the ability of the algorithm to resist impulse noise.The results based on the impulsive noise under Arctic ice confirm the robustness and effectiveness of the proposed algorithm.Finally,the UWA channel equalization technique based on impulsive noise preprocessing is studied.An important idea of signal processing in an impulsive noise environment is to first pass the received signal through a preprocessor to suppress high-amplitude impulses,and then process it according to the method under Gaussian noise.To resist the effect of impulsive noise,an extended preprocessor is added before the equalizer,which nonlinearly transforms the signal disturbed by impulsive noise,preserving more useful information in the received signal while suppressing the impulse.In this paper,an improved proportional least mean square/fourth（IPLMS/F）equalization for Gaussian noise is proposed.To exploit the sparsity of the equalizer,it introduces a proportional diagonal matrix associated with the taps on top of the conventional LMS/F.The experimental results show the advantages of IPLMS/F in terms of convergence speed and BER.To weaken the adverse effect of impulse noise on the equalizer,this paper proposes improved proportional stabilized fast lateral filter equalization（IPSFTF）based on extended preprocessing.IPSFTF is a fast algorithm of recursive least squares（RLS）,which also combines the sparse characteristics of the equalizer.Simulation and experimental results show that the IPSFTF equalizer after extended preprocessing is robust in the impulsive noise environment and outperforms other preprocessors in terms of BER performance.