Novel Methods for Correcting Closed-Chain Errors of Product Codes

Product codes have many advantages such as simple encoding procedure, low decoding complexity and simple row/column interleaving structure. The iterative soft-input soft-output (SISO) decoding of a product code has been shown that the BER performance can approach to Shannon’s limit. However, SISO decoding is complex. In comparison, hard-input hard-output (HIHO) decoding is simpler but still maintains a reasonable BER performance. It has been indicated that the performance of the iterative HIHO decoding scheme of product codes is degraded severely by closed-chains error patterns (CCEPs). In this proposal, we try to deal with the CCEP problem. The proposal is classified as two parts. In part I, a novel CCEP correcting method for binary product codes is developed. The concept of the new method stems from the fact that bit-flipping decoding corrects more errors than erasure decoding with a high probability for high signal-to-noise ratio. Then, the CCEP problem for Reed-Solomon (RS) product codes is considered in Part II. We first try to develop different RS decoding algorithms for different erasure modes. Then using these decoding algorithms, we try to design a CCEP correcting method for RS product codes.

Project ID:PB10408-5755
External Project ID:MOST104-2221-E182-009