Research And Design Of Underwater Acoustic Channel Equalizer Based On Kernel Function

High speed underwater acoustic communication is very important for China's economic construction and national defense construction.However,due to the interference factors such as multipath effect,Doppler effect and complex background noise in underwater acoustic channel,efficient and stable underwater acoustic communication is still an urgent problem to be solved.Traditional underwater acoustic communication mainly uses equalization technology to eliminate inter symbol interference,such as least mean square(LMS),recursive least square(RLS),etc.However,due to the existence of training sequence,the performance of adaptive equalization algorithm is not ideal.Therefore,based on the in-depth analysis of the characteristics of underwater acoustic channel,this paper proposes a joint blind equalization algorithm based on kernel function,and realizes the equalizer design based on FPGA.Aiming at the nonlinear characteristics of underwater acoustic channel,this paper introduces the idea of kernel function to solve the problem of underwater acoustic communication.The signal sequence in the original low-dimensional input space is transformed into a highdimensional feature space,and the inner product operation displayed inside is estimated and replaced by the reconstructed kernel function,thereby solving the nonlinear channel equalization problem by means of linear equalization.This paper analyzes the Kernel Least Mean Square(KLMS,Kernel Least Mean Square)algorithm under Gaussian noise and the Kernel Least Mean Norm algorithm(KLMP,Kernel LMP)under Alpha stable distribution noise,and proposes a variable step size Kernel Minimum Average Norm algorithm(VSSKLMP,Variable step size KLMP)based on the deformed Gaussian function.The simulation shows that the convergence speed of the variable step size algorithm is obviously accelerated,the steady-state error is reduced,and the equalization performance of underwater acoustic communication is effectively improved.In order to solve the influence of training sequence on underwater acoustic communication,this paper introduces a decision mechanism based on the KLMS algorithm,and proposes a decision kernel least mean square algorithm(DDKLMS,Decision Directed KLMS),which eliminates the training sequence process and saves communication resources.On this basis,the mixed kernel function is used to replace the commonly used Gaussian kernel function,and the decision mixed kernel least mean square algorithm(DDMKLMS)is obtained.At the same time,MMA+DDMKLMS is obtained by combining the MMA algorithm with the DDMKLMS algorithm by adopting the joint blind equalization algorithm of the parallel structure.Simulation analysis shows that the MMA+DDMKLMS blind equalization algorithm can effectively reduce the mean square error and bit error rate.Finally,the design of blind equalizer based on FPGA is realized.Divide the modules according to their functions,analyze the blind equalization algorithm logically,and use the channel transmission data collected through the anechoic pool to test.The results show that the joint nuclear blind equalizer can also achieve faster convergence on the hardware platform,and the rear signal is basically close to the expected signal,achieving effective restoration.