Data Allocation Algorithm Based On Kalman Filter For Mobile MPTCP

Recently,multi-homed host has become increasingly common,which especially in mobile users.In order to take advantage of multiple file transfer access points,IETF has proposed multi-path transmission control protocol (MPTCP).MPTCP,as a TCP extension,is based on the original TCP and divided TCP into two layers:MPTCP layer and TCP subflow layer. MPTCP can use multiple interfaces to transfer data, this approach not only makes full use of the available bandwidth, but also to improve the robustness of the network. However, by using multiple interfaces to improve network performance facing enormous challenges. Multi-homed devices are typically heterogeneous, various networks have different link characteristics. Due to the instability of the heterogeneous network performance, a good description of heterogeneous network’s link characteristics is needed. Kalman filter is an effective recursion method, which not only can evaluate and correct the current system state, and even predict the future by recent state status. As used, the Kalman filter prediction transmission delay, and to evaluate the results as a basis for MPTCP data allocation. This paper also study MPTCP’s delay acknowledgment mechanism and point out TCP’s delay acknowledgment mechanism doesn’t adapt well in MPTCP, and to propose new delay acknowledgment mechanism, in order to optimize the performance of MPTCP. The simulation results in our proposal show the rapid changes for the transmission delay has made good suppression, and can quickly track changes in transmission delay, thus reducing the rearrangement of the package, improved to increase the congestion window mode, and ultimately improve the overall benefit.