Puncturing is well-known with rate-compatible punctured convolutional codes [1]. Ha et al. successfully show that it is also possible to design capacity-approaching rate-compatible punctured LDPC codes at very long block lengths [2, 3] and good practical rate-compatible punctured LDPC codes at short block lengths [4, 5]. Since [4, 5], some extensions to the work in [4, 5] have been devised to mitigate the performance losses at high code rates [6, 7, 8]. The extensions take structural advantages of specific types of LDPC codes. That is, the work in [6, 7] utilizes the dual-diagonal structure of repeat accumulate codes and quasi-cyclic LDPC codes, respectively.
Although puncturing is a promising technique to realize rate-compatible LDPC codes, it has a few problems such as performance losses at high code rates and slow convergence of the belief-propagation (BP) decoding [9].The convergence speed is also an important issue in real applications since the slower convergence needs more iterations which result in more power consumption or longer decoding latency. However, to our best knowledge, there has not been any study on how to cope with the slow convergence problem with punctured LDPC codes.