Kernel Adaptive Filtering Algorithm And Its Application Of Noise Cancellation And Channel Equalization

Linear adaptive filtering algorithm has some advantages in dealing with some problems.However,in practical applications,the performance of linear algorithm is not satisfactory when dealing with nonlinear problems,for example,in the application of noise cancellation and channel equalization.Kernel methods were introduced into the nonlinear domain with support vector machine.The kernel methods provide a technological path to study the nonlinear problem.The nonlinear relation in the original space is solved by using the linear algorithm in some appropriate high feature space.Kernel methods have been widely used in the fields of pattern recognition and image processing.At present,there are a lot of linear adaptive filtering algorithms.Therefore,it has an important theoretical significance and value how to use these achievements to develop more and effective nonlinear filtering algorithms.Kernel method is an effective technique for deriving nonlinear algorithms from linear algorithms.The performance of least mean square(LMS)algorithms is not ideal when dealing with nonlinear problems.In this paper,two nonlinear adaptive filtering algorithms based on kernel methods are researched,called KNLMS algorithm and KLMF algorithm.The primary work of this paper is summarized as follows:(1)Based on the principle of kernel method,we research two kinds of kernel adaptive filtering algorithms: kernel normalized least mean square(KNLMS)algorithm and kernel four order error signal minimization(KLMF)algorithm.The results of numerical experiments are presented.(2)The application of the researched KLMF algorithm in noise cancellation is discussed.The simulation results show the availability and effectiveness of the algorithm.(3)The application of the researched KLMF algorithm in nonlinear channel equalization is discussed.The simulation results show the availability and advantage of the algorithm,and the influence of different parameters on the performance of the algorithm is analyzed.