摘要

We present a new design approach to construct generalized concatenated codes (GCCs) based on polar codes (PCs). It is already known that PCs can be considered as a special class of the GCCs with polar outer and inner codes. We show how density evolution (DE) can be used to develop a channel-specific method to design outer codes of length L under maximum-likelihood (ML) decoding. Once a bank of outer codes of length L and different rates have been designed, we develop a rate-allocation algorithm to allocate rates to the outer codes of the GCC with the goal of minimizing the overall block error rate for a given overall rate of K/N. To maintain the low complexity of the SC decoder, we only design very short outer codes (L = 8). We show that, at this outer code length, it is possible to design GCC-PCs that can result in up to 0.5-dB coding gain while reducing decoding complexity.

  • 出版日期2017-4