Congestion Control in Wireless Multi-hop Networks

Wireless multi-hop networks have been envisioned to have many potential applications, e.g., a small stand-alone network for a group of mobile users (Mobile Ad Hoc Networks), a costefficient stub-network to connect to the Internet (Wireless Mesh Networks), or a self-organized community network connecting houses together (Wireless Community Networks). However the preliminary deployment experience of wireless multi-hop networks reveals that despite the advances in physical-layer transmission technologies, limited capacity continues to be a major factor that limits the performance of wireless multi-hop networks and severe congestion collapses are pervasive within multi-hop wireless networks. A congestion control scheme which provides an efficient and fair sharing of the underlying network capacity among multiple competing applications is crucial to the efficiency, stability, and fairness of wireless multi-hop networks. This dissertation presents an experimental investigation into the congestion control problem in wireless multi-hop networks. As the first step, we investigate the interaction between TCP congestion control and the IEEE 802.11 link-layer bit-rate adaptation in a wireless multi-hop network testbed. Based on the observations for the all-wireless paths, we further extend our experiments to paths that consist of both multiple hops in the wireless networks and the wireline Internet. We observe: (i) Although it performs well in single-hop paths, link-layer bit-rate adaptation becomes very inefficient for TCP flows as the length of the path increases; (ii) Poor interaction between TCP and bit-rate adaptation contributes to the inefficiency; (iii) This poor interaction occurs both in pure wireless environments as well as hybrid paths. We develop a simple model that captures the cause for the inefficiency of TCP over autorate links. We then examine several techniques at both the TCP level and the link layer (TCP Vegas, clamping, limiting the buffer size at the wireless routers) to alleviate contention. None of these techniques works for all the scenarios. As part of our experimental evaluation study, we develop IvyNet, a large-scale miniaturized wireless testbed. The validation studies confirm the fidelity of the miniaturized network and also identify the potential problems of the miniaturization technique. The observations from the TCP experiments lead us to develop the idea of channel-aware congestion control protocol for wireless multi-hop networks. Channel-aware congestion control protocols explicitly incorporate the wireless channel information into the congestion control process and as a result, they are able to adapt to the wireless network dynamics quickly, effectively eliminate the congestion collapse within the wireless channel, and significantly improve the efficiency, stability, and fairness of wireless multi-hop networks: 1. We design and implement wireless eXplicit Congestion control Protocol (WXCP), a congestion control protocol for wireless multi-hop networks. WXCP explores the idea