Abstract:
Weproposea newapproachtothe analytical estimation of the error burst probabil ity, the probability of erroneous decoding, and the probability of error per bit for convolutional codes with Viterbi decoding in a binary symmetric channel (BSC). Upper and lower estimates of the probability of error per bit and of the erroneous decoding probability are based on active distances and the distance spectrum of active distances for a convolutional code. The esti mates are derived for rate $1/2$ convolutional codes, but they can also be generalized to any convolutional code with rate $1/n$. Calculation of the estimates described here has linear time complexity in the error burst minimal length if code distance properties are known. The compu tational complexity does not depend on the crossover probability of a BSC. Simulation results show that the considered estimates are rather tight, especially for small crossover probabilities.
Keywords:convolutional codes, active distance, bit error rate, code trellis.
Fulltext:
PDF file (398 kB)
