| As the Error correction coding theory becomes more systematic, coding schemes and architectures of low-complexity, strong correction capability and high efficiency have been realized. The concatenated Reed-Solomon (RS) Trellis coded Modulation (TCM) scheme is adopted by ATSC and ADTBT standards (systems) not only because it can correct both burst errors and random errors but also it is bandwidth efficient; Meanwhile, for approaching Shannon limit and low encoding/decoding complexity, low-density parity-check (LDPC) codes have always been the heat topic in the coding field and industrial field. And based on the observation above, this thesis is mainly dedicated to solving three issues: proposed an iterative decoding algorithm and architecture for the RS concatenated TCM systems; proposed the Coefficient adaptive decoding algorithm based on the concept of Generalized Mutual Information; and finally in the shortening area, introduced the concept of using the known bits for shortening and then proposed a modified shortening algorithm.The commonly used decoding scheme for the RS concatenated TCM systems is a combination of Berlekamp algorithm and Viterbi algorithm for the RS and TCM decoding respectively. Now, this conventional scheme is challenged by LDPC and Turbo codes in terms of the correction performance. For this reason, many researchers proposed some improved algorithms based on the soft decision decoding for RS codes, which suffer from huge complexity and thus are not applicable. In this thesis, based on the conventional low-complexity decoding architecture, we propose a novel iterative decoding algorithm which introduces a―soft value modification‖ technique. The proposed algorithm feeds the error position information back to the input of the Viterbi decoder, and then uses a―Minimum Euclidean distance‖criterion to modify the soft values which come from channel output. On one hand, the proposed algorithm has the similar complexity with the conventional algorithm, which is feasible for hardware implementation; On the other hand, the simulation results show that the proposed algorithm improves the conventional algorithm greatly in terms of BER threshold. Additionally, based on pingpong-RAM architecture with cyclic read-write mechanism, we design a low-complexity iterative decoder for the proposed algorithm. Through analysis, we find out that the proposed architecture for iterative decoder consumes less hardware resources compared with the traditional pipeline architecture.As for the decoding of LDPC codes, the mostly used algorithms include the Offset BP-based and the Normalized BP-based algorithms, which subtract and multiply a certain scaling factor respectively. However, they suffer performance loss because of setting the scaling factors constant. In this thesis, we first deduce the mathematical representation of Generalized Mutual Information for both Offset BP-based and Normalized BP-based algorithms, and then, based on the density evolution theory and the criterion of―maximum GMI‖, we propose the GMI Coefficient Adaptive algorithm (GMI-ACA). Simulation results show that the proposed algorithm has better performance.To shorten LDPC codes using known information in order to achieve multi-rate transmission is a creative thinking in the application perspective. Meanwhile it also involves the theoretical problems in terms of degree distribution, correction ability. In this thesis, we first analyze the shortcomings in the conventional LES shortening algorithm, which is based on the―Lowest column-degree‖priority, and then we propose an improved SRVP shortening algorithm. The Exit Chart analysis and a number of simulation results show that the proposed SRVP algorithm is better than the LES algorithm in terms of BER performance. Meanwhile, the results also indicate the relationship between the proportions of shortened information columns with the performance of shortened codes, which, we believe, provides a good reference for the practical use. |
PDF Full Text Request