One Bit Is Enough: a Framework for Deploying Explicit Feedback Congestion Control Protocols

Since the bandwidth-delay products of Internet links keep increasing, issues with TCP's convergence time, throughput, and amount of queuing in these environments are becoming more pressing. As we move toward the next-generation Internet, it is important to start experimenting with congestion control protocols for high BDP networks as soon as possible. Load-factor based congestion control approach holds great promise as it requires moderate changes in routers (only to monitor and insert current load into packets) and endpoints (to use a congestion controller that uses router feedback). However, existing proposals either require substantial changes in the IP header (XCP and most recently, MLCP) that make deployment difficult, or have slow convergence time and fairness issues (VCP). In this paper we describe UNO, a framework for fine-grain explicit feedback congestion control protocols that uses only 1 or 2 existing ECN bits, thus making algorithms that use more than 2 bits for encoding the load factor and the RTT immediately deployable. UNO accomplishes this task by changing the way load and RTT information is encoded in packets. UNO takes advantage of the IP identification field and trades-off small amount of time (to send several packets) for space to emulate the existence of several extra bits within the IP header. The results from extensive ns2 simulations over various bandwidth and delay scenarios are encouraging. By using only one ECN bit we achieve substantially lower convergence times and better link utilization than the existing deployable protocols, with similar low queue size and negligible packet loss. With 2 ECN bits, we achieve very good fairness for flows with different RTTs, while keeping all the good characteristics of the 1-bit protocol.