Research On Network Coding In Transport Layer

Network coding is a frontier technology in the field of network communication. It is widespread concerned by the academia and industry when it is proposed. Network coding is a information exchange technology which integrated the coding and routing technology. Based on the traditional store and forward routing method, intermediate nodes encode multiple packets receiving from other nodes to increase the amount of information in a single transmission. In this way, network coding can improve overall network performance. Network coding can be applied to Ad Hoc networks, sensor networks, P2P content distribution, distributed file storage and network security and so on.In wireless networks, using network coding can reduce the impact of link failure. As the node's processing power and energy is limited, using random linear network coding,the decoding time is very long for a large scale block. Therefore we need to using dividing block technology to reduce the size of data blocks, thereby reducing the decoding time. As the computation is very large by binomial distribution solving the probability of successful decoding, in this paper, a simple and effective method proposed for solving the probability of successful decoding through the normal distribution to approximate the binomial distribution. In different network conditions, we compare the decoding probability of random network coding, block random network coding and the joint block random network coding. Considering the transmission times and time complexity, we find the optimal data block size to make the cost of netwrok coding mimimum.Application of TCP protocol may result in unnecessary retransmissions because of the feedback transmision failure. In this paper, using network coding in the transport layer can greatly reduce the retransmissions by modifying the feedback mechanisms of TCP protocol. Theoretical proved that relative to the random linear network coding, the successful decoding probability of joint block random network coding is increasing, but the decoding time is reduced. Finally, network coding proved to greatly reduce the number of retransmission by comparison test.