Symbol error probability for convolutional codes (Q2640573)

The authors of this paper propose a new approach for studying the statistic characteristics of the Viterbi type decoders. It is calculated the symbol error probability for convolutional codes and the authors make some observations on the limits of the error probabilities in the conditions of a signal-small noise ratio. This paper contains a presentation of the Viterbi algorithm and a demonstration of two theorems in connection with the error probability of bit whose asymptotic value is determined. The approach of this paper allows for obtaining a non-asymptotic value for the error probability of bit, \(P_ b\), for a certain given code. Consequently for the (7,5) code and for \(p=0,01\) they obtain \(4,81\cdot 10^{-5}<P_ b<4,94\cdot 10^{-5}\). This paper is conceived in a didactic manner and it is richly illustrated by examples that make it easy to understand.