The Research On Optimal Encoding And Decoding Algorithm And Application Of LT Code

The Digital Fountain code (DFC) is a network coding algorithm that has the leastrelevant to the channel, which is very potential for the deep space and networkcommunication. Because of the DFC, the executing network transport mechanism and theinformation transforming efficiency may be improved. However, many problems likeencoding packet redundancy, decoding delay, stop sending notify, the complexity of nodealgorithm and the degree distribution will make the DFC unsuccessful in actual network.When proposed, the DFC is just an idea, which hasn’t an actual encoding algorithm. LT(Luby Transform) code is a mainly kind of DFC code, it has the basic characteristics of theDFC, and so the construction and feature of DFC could be studied by LT. In order to improvethe encoding and decoding algorithm and provide a simulation platform, this paper researchesthe LT under actual program, including:1. A basic theory of DFC is given out and analyzed, and then the network code and DFCare combined to make certain research about their characteristics, which support theapplication investigation. At the same time, the degree distribution and decoding failureprobability are researched deeply. With the understanding of its process, simulation are thenarranged to produce the result. An application and suitability investigation of the environmentare done and many key problems are discussed.2. The paper investigates a mathematical limiting algorithm into the LT code to solve thebad degree distribution reference. Also, the average degree and algorithm complexity arereduced. The paper mainly researches the formula and works out certain conclusions underlimiting condition. Based on the conclusions, the performance difference is very large withdifferent parameters. Through the rational design, the advantage of LT code is maximized andthe storage is optimized.3. On the basic discussion above, for the data selection problem of the source, a variableframe length LT (VFLLT) code scheme is proposed, whose core idea is encode the sourcedata separately and improve the performance by different design. Furthermore, a softwareexamination platform is designed for the optimal LT algorithms above. The paper works theplatform under Microsoft Visual C++environment, and achieves the process diagram and programming to realize the function of LT code.4. For the deficiency of large storage memory and low decoding probability about the LT,an optimal generator matrix storage scheme is proposed to reduce the decoding delay. For thecolumns of the matrix can be assumed as packets, and so the proposed entropy encodingalgorithm can be introduced into the matrix to encode the columns, which greatly reduces thedata amount for transforming by maintaining the LT characteristics, and improves thedecoding accuracy. The theory research and simulation results show that when the generatormatrix is optimized, the storage will also be reduced heavily.5. Because the delay will exsiting in the LT code, which has encoding delay,transforming delay and decoding delay. And if with large delay, the receiver couldn’t get theencoding packets in time that influences the judgment of the channel and the performance ofLT code seriously. The paper focuses on the delay of the encoding and transmission of onepacket. On the base of VFLLT code above, the paper also improves the networkcommunication process with LT, gives out the formula deduction. On some special condition,the paper analyze the performance of the so called fast LT code and its basic mode ofcommunication.Moreover, the delay performances of LT and network code are simulatedunder different transforming strategy.6. Because of the wideband restriction and network environment and the unsuitability,many systems need the transmitting node to forecast the channel conditions and a lot offeedbacks to ensure the reliable receive, which reduce the reliability of the communicationsystem, the paper discusses the LT code performance in wireless sensor network andunderwater acoustic network, analyzes their basic topology mode with LT code that improvesthe data transmission performance.