—An improved concatenated code structure, which generalizes parallel and serially concatenated convolutional codes is presented and investigated. The structure is ideal for designing low-complexity rate-compatible code families with good performance in both the waterfall and error floor regions. As an additional feature, the structure provides a unified analysis and design framework, which includes both parallel and serially concatenated codes as particular cases. We derive design criteria for the generalized class of concatenated convolutional codes based on union bounds for the error probability and extrinsic information transfer (EXIT) charts for the decoding threshold.
Alexandre Graell i Amat, Lars K. Rasmussen, Fredri