Research On Practical Low-Density Parity-Check Codes To Increase The Reliability Of Communication Systems

Advanced channel coding technology is a necessary part of broad wireless communication and power line communication with an aim to reduce error ratio and to increase the reliability of the systems. LDPC codes have most excellent performance close to the Shannon limit in the world. They can be widely used in broad wireless communication and power line communication system. Frame synchronization, decoding and construction technology for structure LDPC codes are explored and investigated in this paper. Main content and results are shown as follows.Two LDPC code-aided frame synchronization algorithms based on maximum method and threshold method are proposed. The frame synchronization error ratios for both algorithms are derived. Using part of LDPC decoder hardware resource, the algorithms don't need additional hardware resource and a full iterative decoding process. Compared to other code-aided frame synchronization algorithms, favorable synchronization performance and decoding performance using the algorithm based on maximum method that approaches the performance of ideal frame synchronization are presented by simulations. A frame synchronization threshold for the synchronization algorithm based on threshold method is proposed and its performance is checked by simulations. Although the frame synchronization performance of the algorithm based on threshold is not as good as the algorithm based on maximum method, it has obvious advantage in synchronization search speed.A BP decoding algorithm joint with check pattern under PN disturbance is proposed for the system adapting PN codes to solve the frame synchronization ramp. An improved double layered BP decoding which computes two check row grouping simultaneously to reduce the number of sub-iterative computation and thus to increase decoding speed is proposed. The initialization of BP decoding for amplify-and-forward and decode-and-forward cooperative relaying is solved. Simulation results show efficiency and validity of all the improved BP decoding algorithms.When the information is exchanged, a connecting law of basic switch units in Reverse Banyan network for structre LDPC decoder is discovered and proved. Then a non-blocking permutation structure based on presetting routing algorithm is designed, which are suitable for structure LDPC codes decoders. Compared to Benes exchange structure and Reverse Banyan exchange structure, the novel structure increases the exchange speed for information cyclic shift and occupies less hardware resource and area. Finally, an output converting unit is designed, which is adaptable for all kinds of switch structures.Based on fast gauss elimination algorithm, a constructing method for structure LDPC codes with multiform structure and rational column weight distributing is proposed. The fast gauss elimination based on basic grouping operation eliminates the whole permutation matrix considered as a unit, which reduces the complexity of the algorithm. The condition forming length-four loop and length-six loop are analyzed and check methods for loops with small length are proposed. Compared to other structure LDPC codes, simulation results show better decoding performance of S1-LDPC codes with 1/2 and 2/3 code rate by the first constructing method, especially in the performance of frame error ratio.Based on multi-permutation matrix, another constructing method for LDPC codes with sparse generation matrix is proposed. The conception of multi-permutation matrix is given firstly, than important characteristics of the matrix are explored and proved. It is analyzed that the generation matrix of the structure LDPC codes take on structure and sparse characteristics applicable of joint encoding and decoding for cooperative relaying based network coding. By simulation comparison, the second kind of LDPC codes named as S2-LDPC codes show dissatisfied decoding performance when the code rate is 1/2. When code rate is 2/3, S2-LDPC codes show favourable performance only inferior to S1-LDPC codes.