Research On The Design Of Coder-decoder&Algorithm For Fountain Code

The data in the transfer process may be lost in a variety of channels because ofa variety of issues, which cause that the packet can not reach the receiving end, orthe packet can not be decoded because of the check code does not match. Both casescan be viewed as lost, so any channel has its own probability of deletion and can beseen as a Erasure Channel in a broad sense. In order to improve the reliability of thetransmission channel, we can use the Automatic Repeat Request technology, but itwill cause a large delay. We also can use the error-correcting code, but its ability toadapt to the changing channel is weak because of the fixed code rate, in themeantime, the algorithm has a high complexity. Digital fountain code can adapt tothe changing channel very well because its rateless characteristic, it also can avoidthe feedback retransmission and has a low comp lexity. The packets received can beerror-free decoded in the receiver as long as enough encoding packets have beenreceived, and the number of it should larger than that of the source packets. Modernwireless communication pursues high-speed data transmission, the fountain codecan effectively ensure that the transmission is reliable. Not only in wirelessmultimedia transmission, the fountain code also has many considerable prospectsfor application in mass storage and deep space communications.This thesis studies two fountain codes which cause the most attention-LTcodes and Raptor codes. At first, this paper completed the design of coder-decoderfor LT codes with the use of two distribution algorithm: Ideal Soliton Distributionand Robust Soliton Distribution, compared their influences on the performance ofLT codes. On this basis, cascaded LDPC codes are used as the precoding, Raptorcodes. Simulation results show that Raptor codes have a better performance than LTcodes, the decoding failure probability of Raptor codes is lower than that of LTcodes. But the decoding time increased exponentially, they are suitable for thechannels which have limited capacity. This thesis discovered that the degreedistribution have a great influence on the encoding and decoding by analyzing theperformance of the two different codes. Robust Soliton Distribution algorithm hasbeen improved in this thesis. First, the encoded packets whose value of degree islarge are removed, this will reduce the complexity of encoding and decodingalgorithms. Then the encoded packets whose value of degree are one are added, thismakes that the number of the decoding open key increases. At last, the sourcepackets which are not covered in the code generation matrix are found and assignedvalues to. The source packets are all covered in this situation, because of that, thedecoding success probability increases. The simulation results show that the performance of improved algorithm is better than that of the the original algorithm,The decoding sucess probability of the improved algorithm increases, and thecomplexity of encoding and decoding decreases.The improved algorithm workseven better when the number of the soure packets k is small.