Research On Unequal Error Proctection Based On Fountain Codes

The rapid development of wireless communication technology and the increasing bandwidth of wireless network make multimedia transmission possible in wireless environment, especially multimedia broadcast and multicast technology, which is increasingly becoming the core communications business. In recent years, the fountain code technology has received wide attention from the coding industry, with its advantages like non-fixed rate, low encoding and decoding complexity, no feedback etc. Moreover it has been selected as the component of Multimedia Broadcast/Multicast Service (MBMS) of the third generation mobile communication by 3GPP, and has been wildly used in the communication field.Firstly, we introduced recent research on the Unequal Error Protection scheme, and outlined the development, advantages, research status and prospects of the digital fountain technology. We also introduced two practical LT codes, Raptor codes and the UEP scheme based on them. By detailed simulation, we analyzed the factors which impact the performance of LT codes and Raptor codes, furthermore pointed out the strengths and weaknesses of the existing UEP schemes.Secondly, we proposed a novel UEP scheme named D-EWF (Duplication-Expanding Fountain Window) codes based on poor performance of UEP scheme upon the LT codes and Raptor codes. By introducing the expansion factor, the virtual expansion degree distribution, and combining them with the window technology of EWF (Expanding Fountain Window) code, we divided the virtual expansion data into different windows with all the windows using the robust soliton distribution as the degree distribution of LT code. After all these we realized the unequal error protection of the important data. The simulation results showed that the proposed D-EWF algorithm had better UEP performance and improved the error-robustness of less important bits, thus enhance the overall performance, compared with other UEP-LT codes.Finally, we presented a new UEP strategy based on Raptor codes. We adopted the partially regular LDPC Codes as the pre-encoded codes and then got the middle symbols. Then we divide these middle symbols into different windows and implement the second-level coding using the EWF code. By detailed theoretical derivation, we got the density evolution (DE) formula over the binary erasure channels (BEC). Using the DE formula, we optimized the degree distribution of the proposed UEP-Raptor code. Simulation results showed that the proposed algorithm had a better UEP performance than the one proposed in literature [45].