Design, Implementation Of Multipath Transport Protocol In Wireless Networks

With the rapid development of wireless network technology, Internet and the popularity of mobile devices, it has become reality for people to access to the Internet anywhere, which is also a common demand for human beings. However, both bandwidth and latency of wireless network are difficult to meet the needs of users. Although many mobile devices have more than two network interfaces, but existing transport protocols supports only single path, only one of these interfaces is used, which can't make full use of hardware resources. In recent years, multipath becomes a research hotspot.Research point of this project is a multipath transport protocol in wireless networks. Bottleneck link problem is the main constraint in current multipath transmission, which is worse in wireless networks, that is to say, poor links will become the bottleneck of the network and influence better ones, the throughput of the whole network degrade. First, this paper describes the development of a wireless network, and then introduces the network coding technologies, introduces three mainstream network coding styles, they are random linear codes, fountain codes, system codes, details some multipath transport protocols. Finally, we propose a multipath TCP transport protocol called NCMPTCP based on random linear network coding(RLNC), which introduces network coding into multipath to weak the impacts among multiple paths, the goal is to achieve aggregate throughput of TCP. All data will be split into blocks, and then encoded into symbols before transmission. Each block encoded independently and any symbol isn't unique. As long as the receiver gets sufficient number of symbols, it can recover the block successfully. This characteristic helps solve the bottleneck problem in heterogeneous networks. Besides, we design a scheduling algorithm based on estimated delivery time, which schedules symbols based on the link status of the path, aims to make full use of all available bandwidth, achieve maximum throughput, maximize resource utilization.