A New Distance-Based Algorithm for Block Turbo Codes: from Concept to Implementation

List-based algorithms for decoding Block Turbo Codes (BTC) have gained popularity due to their low computational complexity. The normal way to calculate the soft outputs involves searching for a decision code word D and a competing codeword B. In addition, a scaling factor alpha and an estimated reliability value beta are used. In this paper, we present a new approach that does not require alpha and beta. Soft outputs are generated based on the Euclidean distance property of decision codewords. More importantly, such algorithm has very low computational complexity and is very attractive for practical applications. Based on the synthesis result of FPGA (Field Programmable Gate Array) implementations of the new algorithm, significant complexity saving (up to 79%) is achieved compared to commercially available products. In terms of error performance, we observe certain improvement (0.3dB coding gain) for BTCs of large Hamming distance and negligible performance degradation for BTCs of short Hamming distance.