Pre-processing of Convolutional Codes for Reducing Decoding Power Consumption
Wei Shao, Linda Brackenbury
Abstract
This paper presents a class of pre-processing approach for convolutional codes that can provide the adaptive capability to a standard Viterbi decoder. By identifying and generating the decoded data from the zero Hamming distance path, all computations in the Viterbi decoder can be avoided. This makes it possible to stop the Viterbi decoder as long as no error occurs in the received code words; the Viterbi decoder is restarted to correct errors otherwise. With this approach, significant power, i.e. 97% of a standard Viterbi decoder power dissipation, can be saved when the Eb/No is as low as 13dB. Further improvement of power saving is also possible by pre-correcting some of the errors in the received code words path. This enables the power saving starting 2dB lower than the original pre-processing approach.