Abstract
Joint Encoding and Decoding Optimization of LT Codes over Noise Channels
Highlights
Fountain codes known as rateless codes, which were originally designed for reliable transmission over the binary erasure channel, exhibit excellent performance over time-varying channels
We present simulations of bit error rate (BER) and frame error rate (FER) performance of the proposed Regularized Variable-Node (RVN) Luby Transform (LT) Codes with Cyclic Redundancy Check (CRC)-BF scheme in comparison with conventional LT Codes and RVN LT Codes with classical Belief Propagation (BP) decoder
The complexity overhead of the proposed scheme compared with conventional LT Codes is mainly due to CRC-BF scheme in decoding process
Summary
Fountain codes known as rateless codes, which were originally designed for reliable transmission over the binary erasure channel, exhibit excellent performance over time-varying channels. Improving the decoding performance of LT Codes and lowering its error floor are of great importance for the application of fountain codes in practice. In addition to the degree distribution and the random encoding method, the belief production (BP) algorithm is the major reason for the poor decoding performance of LT Codes in noise channels. Most attention has just been paid to the optimization of degree distribution and encoding method for lowering error floor of LT Codes in noise channels. When the CRC detection fails, by some post-processing according to the characteristics of the encoding structure and decoding algorithm, the performance of channel decoding could be further improved. We propose a joint encoding and decoding optimization scheme, which combines the RVN encoding and CRC-aided BP decoding to lower the error floor of LT Codes.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have