Dr Li Yang PhD
Brief biographical informationLi_yang@adp.com
A Study of Iterative Capacity-Approaching Codes And Their Optimal Decoding Algorithms
Since the invention of turbo codes in 1993 and the rediscovery of Gallager's LDPC codes in 1999, Shannon's determination of the capacity of memoryless channels has become touchable and achievable by adopting these codes iterative decoding scheme. Theoretically, turbo codes and LDPC codes have essentially revolutionised the coding field and become the most focused research topic in recent years. Although these remarkable codes have demonstrated an asymptotic performance close to the Shannon limit, there exists a fact that the practice still lags behind the theory by some margins, for instance, the code constructional diculty, decoding complexity and the hardware or other implementation issues. In terms of the near optimum decoding performance, it still seems infeasible in practice.Dr Li Yang
This research work endeavours to fill some of these gaps concerning the design, analysis, application and algorithm optimisation of these simple but good codes, which aims to provide a near optimum decoding performance with much less computational complexity.
After the study of these codes and their iterative decoding scheme, we introduce two hybrid decoding arrangements for the erasure channel and the AWGN channel. Both decoding arrangements are designed to achieve the near optimum or sub-optimal performance with much less computational complexity compared to the maximum likelihood decoder.
The second main contribution is to introduce an ecient algorithm by exploring the codes tree representation to help analyse the codes weight spectra and stopping sets. Furthermore, an extended decoding method based on the state-of-art tree search is proposed to ensure the optimum decoding performance for sparse structural codes in moderate codeword length.
Director of studies: Professor Martin Tomlinson
Other supervisors: Dr Ambroze Marcel, Dr Ahmed Mohammed
Decoding low-density parity-check codes with error-floor free over the AWGN channel
Extended optimum decoding for LDPC codes based on exhaustive tree search algorithm
A tree-based ML decoding algorithm with adaptive thresholding
Comparison of decoding turbo Gallager codes in hybrid decoding arrangements with different iterative decoders over the erasure channel
4 Conference papers
4 publication(s) - all categories.