HBRP Publication Page 1-6 2020. All Rights Reserved Page 1 Journal of Advancement in Communication System Volume 3 Issue 1 DOI: [To be assigned] System Swarm for Equalization of Nonlinear Channels Padma Charan Sahu 1* , Sunita Panda 2 , Ratnakar Dash 3 1 Gandhi Institute of Engineering & Technology University, Gunupur, Odisha, India 2 Gandhi Institute of Technology and Management, Bangalore, Karnataka, India 3 National Institute of Technology, Rourkela, Odisha, India *Corresponding Author E-Mail Id: padma.tvc@gmail.com ABSTRACT In order to improve the performance of digital communication system, here the author proposes equalization of nonlinear channels by using ANN trained with PSO. Broad recreations introduced during this paper shows that, when contrasted with other ANN based equalizers additionally as Neuro-fuzzy equalizers the proposed equalizer performs better by and large condition Keywords: ANN, PSO, channel equalization INTRODUCTION In order to retrieve the binary information from digital channels Adaptive channel equalizers assume is the best method which plays a significant job in recuperating computerized data. Preparta had proposed [1] a straightforward and appealing plan for dispersal recuperation of computerized data dependent on the Discrete Fourier Transform. In this way Gibson et al have detailed [2] a productive nonlinear Artificial Neural Network structure for recreating binary signals, which have been gone through a dispersive channel and adulterated with added some unwanted signal (Additive Noise). The ideal preprocessing procedures for reproduction of binary signals from a dispersive correspondence channels has been represented in [3]. Touri et al have created [4] deterministic most pessimistic scenario system for ideal reproduction of discrete information data from advanced correspondence channels. Preparta had recommended [1] a basic and alluring plan for dispersal recuperation of computerized data dependent on the Discrete Fourier Transform. Thusly Gibson et al have announced [2] a proficient nonlinear Artificial Neural Network structure for recreating advanced signs, which have been gone through a dispersive channel and defiled with added unwanted signal (Additive Noise). In later past new versatile equalizers have been recommended utilizing delicate processing devices, for example, Artificial Neural Network (ANN), PPN and the FLANN [5]. It has been accounted for that these techniques are most appropriate for nonlinear and complex channels. As of late, Chebyshev Artificial Neural Network has additionally been proposed for nonlinear channel evening out [6]. The disadvantage of these strategies is that during training, it is very difficult to compute the local minima. Thus Genetic Algorithm (GA) has been recommended for preparing versatile channel equalizers [7]. The principle fascination of GA lies in the way that it doesn't depend on Newton gradient techniques, and thus there is no requirement for count of subordinates.