The Research And Mplementation Of LDPC Coding And Decoding Techniques Based On Multi Core Architecture

With the development of high speed communication,the expansion of the communication is proposed.At the same time,the rapid development of multi-core architecture and the gradual maturity of the parallel computing model technology are caused concerned in recent years.In this paper,a parallel computing model is established based on the multi-core platform architecture.The key technology of multi-core and multi-thread is used on implementation of the LDPC(Low-Density Parity-Check)encoding and decoding system.For the establishment of multi core platform parallel computing model,this paper describes the implementation mechanism of OpenMP parallel computing model and the comprehensive use of multi thread programming synchronization technology,and then simulates the performance of the parallel model.For the LDPC coding and decoding technology,the standard of IEEE 802.16e code,whose bit rate is 1/2 and the expansion factor is 64,is used.In this paper,it improves and optimizes the traditional coding and decoding algorithm.In the encoding algorithm,it analyses the fast encoding algorithm based on the DFS(Depth-First-Search)algorithm,improves it as the fast encoding algorithm based on end DFS.In the decoding algorithm,the minimum-sum-algorithm based on BP(Error Back Propagation)algorithm is discussed,and increment-minimum-sum algorithm optimizes this algorithm by speeding up its convergence.At last,the optimization of encoding and decoding algorithm is verified respectively by the time cost and the number of convergence.In addition,this paper discusses the application of LDPC in the deep space communication,researches the two synchronous modes,and verifies the LDPC can indeed enhance the transmission efficiency remarkably in the deep space communication.For the implementation of LDPC encoding and decoding system,this paper describes the design and implementation of the system in multi-core platform.On the new multi-core platform,the OpenMP parallel computing model is established,and based on the in-depth study of the mechanism of shared memory,the data synchronization is performed using multi thread locking mechanism.By addition,the static scheduling strategy is invoked to achieve load balancing,and then the LDPC code algorithm is calculated based on all above.Finally,with the operation of LDPC code carried out on the new platform,this paper simulates the rate of LDPC encoding and decoding,and achieves the expected rate performance improvement based on the number of cores,completing the performance index of Mbps level on LDPC encoding and decoding.