Multiple-Attempt Decoding of Convolutional Codes Over Rayleigh Channels

Abstract

The Viterbi algorithm (VA) is a maximum-likelihood method used to decode convolutional codes with the minimum average number of bit errors. In the case of decoding errors, the receiver can request a retransmission of a message by means of an automatic repeat request (ARQ) scheme until it receives the message correctly. Here, we present an alternative approach to reduce decoding errors by deploying a scheme of multiple-attempt decoding (MAD) at the receiver over fading channels. This procedure is deployed only if the frame is found erroneous in the first attempt. It starts with dividing the received sequence into multiple sections and erasing a symbol from each section. The resultant sequence is then decoded and checked through cyclic redundancy check (CRC). If CRC fails, this algorithm resumes by deleting a different set of symbols and retaining the previous set of deleted symbols. These procedures stop if the CRC passes or the maximum number of attempts has been exhausted. The proposed method requires no change in specifications of the transmitter and the hardware at the receiver. Significant gain margins are observed in the frame error rate (FER) at the cost of only a slight increase in software complexity.