Performance analysis of punctured convolutional codes for variable-bit-rate speech transmission

A good alternative of forward error correction for variable bit-rate speech transmission is use of rate R=n=1/n punctured convolutional codes (PCCs). The authors evaluate the performance of different code formats of PCC in conjunction with embedded differential pulse code modulation (EDPCM) and coherent binary phase shift keying (BPSK) for AWGN. Rayleigh and Rician fading channels using either soft-decision or hard-decision decoding. It is found that the punctured codes of rate R=2/3 and R=3/4, which are derived from a rate R=1/2 convolutional codes, are more convenient to implement since the code generators remain unchanged and therefore, the control mechanism for selecting code formats of different code rates is not necessary. In terms of performance of the above system, it is found that the bit error rates are lowered in comparison to the uncoded system in all three different types of channels. On the other hand, to evaluate the EDPCM combined with PCC for speech transmission, the authors evaluate the speech quality as a function of audio signal-to-noise ratio with respect to channel signal-to-noise ratio. With the same voice quality, significant improvements in the coding gain has been found among different code formats.<
>