基于多指标感知的命名数据网络自适应拥塞控制算法

A multi-metric perception based NDN adaptive congestion control algorithm

为解决命名数据网络拥塞控制算法普遍存在的瓶颈链路缓存队列积压问题,提出一种基于多指标感知的命名数据网络自适应拥塞控制(MACC)算法。在网内节点,根据实时监测的接收端数据接收速率、数据包排队时延、链路带宽等多个指标估计链路负载程度,并通告终端节点。在终端节点,根据Data包携带的链路负载信息,自适应地调整发送窗口,以实现高吞吐、低时延目标。基于ndnSIM的仿真结果表明,该算法能有效实现终端高吞吐量、瓶颈链路低队列,并在实现流间公平性方面具有显著优势。

In order to solve the common backlog issue of bottleneck link buffer queue in named data network(NDN),a multi-metric perception and adaptive based congestion control(MACC)method is proposed.Specifically,intermediate routers notify the data receiver of its estimated link load level by monitoring real-time data receiving rate,data packet queuing delay,link bandwidth and other metrics.According to the link load information carried by the data packet,data receivers adaptively adjust the sending window size to achieve the goal of high throughput and low latency.Simulation results based on ndnSIM show that MACC can achieve good performance in terms of throughput,bottleneck link queuing delay and inter-flow fairness.