Improvement And Application Research Of LT Code

With the development and popularization of network technologies, there is an increasing number of services to distribute and broadcast the mass-data, as well as a growing demanding for highly efficient and reliable data transmission. It has been one of the crucial research topics of modern communication systems how we can transmit the information efficiently and reliably with the limited bandwidth resource. In the audio or video transmission network, a large deal of data in different priorities needs to be transmitted. Moreover, the channel qualities vary with time. Therefore, the unequal error protection(UEP) based on variable bitrate Erasure Codes is the best choice to satisfy the requirements.As one kind of the Erasure Codes with distinctive performance in erasure channel, the code rate of Digital Fountain Code can change in the transmission according to the qualities of channels. Being the first realization of Digital Fountain Code, LT code has the merits of low complexity and good performance. However, the error diffusion and error floor is very obvious in the decoding of LT, for which some improvement of the decoding algorithm of LT code is given in the dissertation. What’s more, in order to boost the transmission performance for the realization of UEP transmission, we also study the method to integrate the LT coding with the network coding which is superior in promoting the transmission efficiency.The major work in the paper is as follows.After introducing the principle of coding and decoding, we propose the Improved Belief Propagation(IBP) based on redundancy due to a mass of structural redundancy. If redundancy exists in the decoding, we use the redundancy to make assumptions and then test the assumption. In this way, some wrong decoding can be corrected which reduces the transmission probability of errors in the decoding. Compared with the original BP, the simulation results demonstrate that IBP is superior to the original in theerror rate.Due to the inferior performance of contemporary schemes to integrate the LT code with the network code, we propose the degree distribution network code based on Expending Window(EWDNC) which fulfills the UEP transmission by introducing and optimizing different initial degrees. The simulation result shows that the decoding performance gets some performance if EWNDC is used by the unit of data sending。 Furthermore, we also put forward the optimization distribution scheme in fixed network structure and make an analysis of the network structure, code probability and other factors.