Improving the efficiency of reliability-based hybrid-ARQ with convolutional codes

In this paper, we propose and evaluate the performance of a new adaptive HARQ technique that utilizes the correlation among bit errors at the output of a convolutional decoder. The proposed correlated bit-error hybrid ARQ (CB-HARQ) uses error events encountered in a soft-input soft-output (SISO) decoder to capture the correlation in bit errors. Retransmission is requested for selected coded symbols that have the potential to correct a set of message bits associated with these error events. The performance of the proposed technique is compared to conventional incremental redundancy HARQ that uses rate-compatible punctured convolutional (RCPC) codes and reliability-based HARQ (RB-HARQ) that does not utilize correlation between error events. The fundamental difference between CB-HARQ and RB-HARQ is that CB-HARQ tries to correct error events occurring in a decoder whereas RB-HARQ tries to correct individual bit errors. The results show that the proposed HARQ technique based on correlated bit errors can provide much higher throughput