A study on the channel equalizer noise of softmodem with Haar transform and neural network

Issue title: ICMIT '01

Article type: Research Article

Authors: Lee, Seok Won^a | Nam, Boo Hee^b

Affiliations: [a] Yountel Corporation, IT Venture Tower B-5F, 78 Garakbondong, Songpaku, Seoul, Korea. E-mail: [email protected] | [b] Department of Electrical and Computer Engineering, Kangwon National University, Chunchon, Kangwondo, Korea. E-mail: [email protected]

Abstract: In this paper, we propose a new structure for a binary data transmission using the neural network and Haar transform to reduce the noise and distortion in the channel. To solve the problems such as signal distortion and nonlinearity during the transmission and reception of the signal, we use the neural network to design an equalizer to adapt the change of the channel characteristics. Before transmitting digital information, two adjacent binary signals are Haar transformed and are sent to the channel which is assumed to corrupt the signal by an additive white Gaussian noise (AWGN), and then we receive the noise-reduced signal through the inverse Haar transform. We analyze the probability of error and simulate the model to show the validity.

DOI: 10.3233/JAE-2003-281

Journal: International Journal of Applied Electromagnetics and Mechanics, vol. 18, no. 1-3, pp. 139-147, 2003