Concept of Information Inversion in Forward Error Correction-A Supervised Learning Algorithm for Decoding

Abstract

The present work aims at proposing a binary forward error correction scheme, where, encoding of aninformation word depends on its weight (Number ofnon-zero elements it possesses). The structure of the overheads of the code word is based whether the weightof the corresponding information word is Even or Odd.The decoding is implemented using the principle of Supervised Learning, where, the code words arelabelled (Label-1 and Label-2) depending on thestructure of the individual’s overheads. The concept ofinformation inversion proposed to constitute the parity word(overheads) of the codeword made the entireprocess of encoding simple, unlike the conventionalBlock coding schemes, where the encoding process is associated with complex mathematics. The concepts of Machine learning, specifically, Supervised Learning used simplified the process of error detection andcorrection also. Generation of codewords using a simple inverter results a great reduction in the complexity of the hardware realization of the encoder and the proposed decoding process facilitates a simple detection scheme, relative to conventional process.