Research On The Transport Protocol Based On LT Network Coding In Mobile Ad Hoc Networks

Traditional transport layer protocol is designed for reliable channel, it can workwell in the cable network, but there will be a significant reduction in the MANET. Theemergence of network coding provides a new way to solve the problem oftransmission protocol in MANET. Based on network coding technique, researcherspropose TCP/NC protocol, the strategy improve the performance of TCP protocol inwireless network, but the efficiency of random linear coding is low, which affects theperformance improvement. At the same time, the current video business hasdeveloped rapidly, the performance of the UDP protocol is in focus, therefore, thisarticle use LT(Luby Transform)code to replace the random linear code, and use it inthe TCP/NC and UDP/NC protocol, to further improve the performance of TCP/NCand extend its application range.In this paper, it introduces wireless network transmission protocol and networkcoding research status, and analysis the basic principle and characteristics of LT code,on this basis, using LT code because of its simple coding and decoding method andlow decoding complexity, then propose TCP/NLT and UDP/NLT protocols. Based onTCP/NC, we redesigned the network coding layer of TCP/NC protocol, the design issuitable for the contract and receiving algorithm, and gives a new ACK control andRTT calculation method. For performance comparison, we expand the structure andprinciple of the TCP/NC protocol to the UDP protocol, then design the UDP/NCprotocol.We do the simulation implementations of the several protocols in NS2, andsimulate in the typical chain ad-hoc network scenario, the results show that thedecoding time of LT code is larger than the traditional linear code in order to achievethe decoding condition, waiting for the enough packets arriving, it leads to theaverage throughput of TCP/NLT below the TCP/NC’s, TCP/NLT still need to befurther improved. And for the UDP, compared with UDP/NC protocol, the averagethroughput and decoding efficiency has better performance.