A novel error detection scheme for turbo coded hybrid ARQ

In this paper, a new error detection scheme is proposed for a type-I turbo coded hybrid automatic repeat request (ARQ) system. Trellis termination of the recursive systematic convolutional (RSC) constituent code of a turbo code is modified to adopt the proposed error detection scheme. Given the same number of overhead bits for error detection, the equivalent CRC generator polynomial in the proposed scheme has higher degree than the CRC generator polynomial in the conventional CRC error detection scheme. The equivalent CRC generator polynomials which provide the smallest undetected error probability are obtained by computer search. It is shown that a turbo coded hybrid ARQ system with the proposed error detection scheme achieves smaller frame error rate (FER) than that with the conventional CRC error detection scheme.