Scalable, network-assisted congestion control for the MobilityFirst future internet architecture

摘要

MobilityFirst (MF), as a realization of Information Centric Network architecture, places intelligent functionality, such as storage and reliability, inside the network to assist with data delivery. The MF architecture requires effective congestion and flow control to efficiently support data delivery at scale. Traditional end-to-end, window-based congestion control like that used by TCP is unsuitable as it is unable to take advantage of such in-network functionality. We design network-layer assisted congestion control schemes tailored to MF. One approach that works well for hop-by-hop reliable networks is using per-flow queueing and backpressure to alleviate congestion. However, it could become impractical in the presence of a large number of flows, which leads to substantial memory consumption and computational complexity. Building on a more scalable per-interface queueing model, we design congestion control mechanism that embodies traffic source rate control and explicit congestion notification from routers. Sample results show that the proposed scheme is able to achieve similar link utilization and better fairness compared with a per-flow queueing scheme.