Exact analysis of bit error and burst error probability for 2-state soft decision Viterbi decoding

Although convolutional codes with soft decision Viterbi decoding are widely used in practical systems, only a few studies have been made for exact ana lysis of error probability. Most results are based on the bound of error pr obability or on the computer simulation. In this paper, we propose a new method for the exact analysis of bit error probability of 2-state convolutional code which is applicable to soft decis ion Viterbi decoding, The method employs an iterative calculation of probab ility density function (pdf) of path metric per one transition. Numerical r esults of bit error probability show that 4 or 5 iteration gives the bit er ror probability with sufficient accuracy, and that the proposed method is e ffective even in the region of low signal-to-noise power ratio (SNR). The e xtension to burst error probability analysis is also presented.