Iterative Decoding of K/N Convolutional Codes based on Recurrent Neural Network with Stopping Criterion
Jwh Kao, Stevan Mirko Berber · UTS ePRESS (University of Technology Sydney) · 2007
This paper outlines a novel iterative decoding technique for a rate K/N convolutional code based on recurrent neural network (RNN) with stopping criterion. The algorithm is introduced by describing the theoretical models of the encoder and decoder. In particular this paper focuses on the investigation of a stopping criterion on the iterating procedure in order to minimize the decoding time yet still obtain an optimal BER performance. The simulation results of a rate 1/2 and 2/3 encoders respectively in comparison with the conventional Viterbi decoder are also presented