Blind Equalization Method Of Underwater Acoustic Communication System Based On Extreme Learning Machine

The limited bandwidth characteristics of the underwater acoustic channel restrict the communication rate,and the complex channel characteristics make the signal transmission vulnerable to interference and absorption attenuation,causing signal distortion,which reduces the communication quality.The use of equalization technology with high frequency band utilization and effective signal recovery and modulation technology with strong anti-interference ability is an important guarantee for achieving stable and reliable underwater acoustic communication performance.The blind equalization technology is a channel compensation technology that only uses the prior statistical information of the received signal and does not need to send the training sequence to recover the original signal,which effectively compensates for the limited bandwidth of the underwater acoustic channel.Orthogonal Frequency Division Multiplexing(OFDM)technology is widely adopted.On the one hand,it divides the frequency selective fading channel into multiple orthogonal flat fading channels to realize fast data transmission.On the other hand,it allows the frequency of the transmission data stream to be superimposed,so that the spectrum utilization rate of the channel has been qualitatively improved.Extreme Learning Machine(ELM),as a new type of feedforward neural network,is widely used in channel equalization with its fast learning rate and superior nonlinear processing capabilities.In this regard,this paper studies the blind equalization method of the underwater acoustic communication system based on the extreme learning machine,and the main work is as follows:(1)On the basis of in-depth study of the extreme learning machine network structure and learning algorithm,it is proposed to use the extreme learning machine as a network equalizer to realize the blind equalization of the underwater acoustic channel,and construct a blind equalization algorithm for the underwater acoustic channel of the extreme learning machine based on the traditional cost function.That is to say,the error function of the Constant Modulus Algorithm(CMA)and MultiModulus Algorithm(MMA)is combined with the constraints of the extreme learning machine to construct the cost function,and the iterative weighted least square method is used to output the weights are trained in batch and online respectively,the principle of the algorithm is explained in detail,and simulation experiments are carried out.The results show that the constructed OS-ELM-SVR-CMA/MMA has a superior balance compared with ELM-SVR-CMA/MMA performance.(2)The principle of OFDM underwater acoustic communication system is studied.Aiming at the blind equalization of OFDM sub-carriers,the sub-carrier blind equalization algorithm based on normal mode signals and multi-mode signals uses the nonlinear processing capabilities of the extreme learning machine and uses it as a blind equalizer to construct an OFDM underwater acoustic communication based on the extreme learning machine The system sub-carrier blind equalization algorithm is simulated by setting relevant condition parameters and the result analysis is given.