Analysis of support vector machine based decoding algorithm for bose-chaudhuri-hocquenghem codes

Abstract

In modern communication systems data is exchanged over unreliable noisy channels that cause the data to be altered by errors. To combat the problem, specially designed error correcting codes are applied to transmit the message over the channel. Bose-Chaudhuri-Hocquenghem (BCH) is a class of codes that are widely applied in practice. Apart from the classical decoding algorithms developed in the latter half part of the 20th century, a number of alternative approaches have been proposed. Among other things, it has been suggested to apply Support Vector Machine (SVM), machine learning classification and regression technique, to decode BCH codes. In this paper characteristics of the SVM based decoding approach is analyzed. Properties of two its modifications are considered.