针对多路传输控制协议联合拥塞控制未考虑丢包率对吞吐量的影响和传统静态路径选择方法降低并行数据传输鲁棒性的问题,提出一种以提升吞吐量为优化目标的基于马尔可夫决策过程的动态路径选择方法.综合考虑往返时延和丢包率对吞吐量的影响,根据最优策略动态地选取不同的路径进行数据传输,数据传输时不减少路径数量,在不影响并行数据传输鲁棒性的基础上提升吞吐量.仿真结果表明,在路径参数变化时,所提出的路径选择方法可以更有效地均衡流量,提升吞吐量.
To solve the problem that the coupled congestion control mechanism of multipath transmission control protocol( MPTCP) did consider the impact of loss rate to the throughput and the traditional static path selection methods did decrease the robustness of parallel data transmission,a dynamic MPTCP path selection method with improving throughput as an optimization target based on Markov decision processes was proposed. This method takes into account the impact of both round trip time and packet loss rate to throughput,chooses different paths based on optimal policy to transfer data but not reduce the number of paths so that it promotes the throughput without reducing the robustness of the parallel data transmission.Simulation shows that the proposed path selection method balances the data stream more effectively and improves the throughput of MPTCP.