Performance of Hybrid ARQ Schemes with ML Code Combining over a Block-Fading Rayleigh Channel

Abstract

This paper considers two well-known selective-repeat retransmission schemes, namely, hybrid type-I ARQ and hybrid type-II ARQ, using convolutional coding, in conjunction with maximum-likelihood code combining. Our theoretical analysis, based upon the concept of generalized weight distribution, shows that the use of code combining yields a significant throughput at very high channel error rates not only in constant AWGN channels but also in fading channels. To demonstrate this, we consider a widely-used block-fading Rayleigh channel model, in which the channel is assumed to be constant during each block of data and the fading is assumed to be independent from block to block. A key parameter in designing retransmission protocols for delay-limited applications in such channels is the minimum number of retransmissions, needed to achieve error-free decoding at almost all channel conditions (low outage probability). This number can be reduced significantly when code combining is employed.