On the distribution of error in convolutional codes

This paper applies error exponent analysis the derivation of the distribution of error events in a convolutional code at the output of a Viterbi decoder. Even if the mean error behaviour of convolutional codes is a well known topic, however a precise analysis of the error events at the output of Viterbi decder is still lacking. The classical approach to derivation of the error events distribution after decoding of a convolutional code is by using the weight distribution of the code. However this approach is specific to each code and the complexity involved can be quickly tremendous when the memory of the code goes larger. In this paper we apply the error exponent theory to derive bound on the distribution of error events and we show that these bound are sufficiently tight to be used in practice.