The Analysis Of Buffer-based Distributed LT Codes And Its Application In Cloud Storage

In the traditional communication networks,the relays are only responsible for storing and forwarding the information,they don't do any operation and just send them to the destination directly when receive information.Network coding can process the data in the relays so as to improve the network throughput and the transmission efficiency,which has important significance for how to realize the effective transmission of the amount of data information in the limited bandwidth.Fountain codes have the advantages of rateless,no feedback and the complexity of encoding and decoding is low,using it in the relay encoding can guarantee the reliability and validity of information transmission.LT code is the most effective one of the fountain codes,but it can only be used in the end-to-end(source--destination)network and it is not suitable for multiple sources and multiple relays communication network.When compared with the LT code,SDLT codes are not only suitable for multiple sources and multiple relays communication network but be allowed to process data in the relays.This thesis mainly studies the key technology and application of the DLT codes from the following aspects:Firstly,this thesis introduces a most practical fountain codes--LT code,whose encoding and decoding algorithm and the design of degree distribution function are also analyzed in detail,after that we introduced the model and the encoding and decoding algorithm of DLT codes on the basis of LT code,then we carry on the analysis of and-or tree and draw the iterative formula of probability of LT code and SDLT codes when decoding successfully;Secondly,according to the SDLT codes' defects,this thesis explores a method of encoding and decoding of the Buffer-based DLT codes and combines it with UEP scheme,simulation results show that this algorithm can let the decoding failure rate reduce to 10-3 level and get better system performance;Thirdly,this thesis designs an actual network model applied in cloud storage network,including 8 users 3 relays and 1 destination,the scheme of encoding and decoding is BBDLT code which combined with UEP.Finally,the simulation results show that the user's(with the strongest protection)decoding failure is about 0,which is lower than the rate of EEP scheme and the other users' decoding failure,so it verifies the effectiveness of the proposed scheme.