Congestion control is an enabler for making the Internet a shared resource. TCP is a good example of it and without it, the Internet would not be able to scale as today in terms of the number of users. TCP is built on top of the end-to-end paradigm, an architectural principle that holds the Internet architecture; it relies on only the end-point feedback to adjust transmission rates at the senders and trusts the end-points (senders and receivers) to make correct and fair decisions in regulating their transmission rates. However, the Internet has evolved tremendously since the inception of TCP congestion control algorithm in 80s. Letting alone greatly different and diverse operating conditions of the network, the groups of stakeholders become so diverse that we cannot trust end-points to behave like they were several 10s years ago. End-points may use their own transport protocols, not conforming to TCP. A good example is the arrival of P2P applications which may use any number of flows for any duration to get a single file transfer. Multi-media streaming can use a transport that does not employ any congestion control. The current architecture of the Internet does not prevent these trends as the E2E argument trust all the complex decision making mechanisms to end-points. The time has come to modify this paradigm and re-design congestion control that can support and withstand the dynamics of tussles among diverse interest groups. In this talk, I will discuss challenges for developing such congestion control and propose several potential approaches for those challenges.

Injong Rhee is Associate Professor of Computer Science at North Carolina State University and runs Networking Research Lab (NRL). He works primarily on network protocols for the Internet. His major contributions in the field include the development of congestion control protocols, called BIC and CUBIC. Since 2004, these protocols have been the default TCP algorithms for Linux and are currently being used by more than 40% of Internet servers around the world and by several tens millions Linux users for daily Internet communication. He also has invented several multimedia streaming and multicast technologies licensed to companies for commercial applications. He started a company based on these technologies in 2000 where he developed and launched the world's first video streaming products and push-to-talk (PTT) VoIP products for cell phones. His recent research topics include mobile ad hoc networks, delay/disruption tolerant networks, and P2P systems. He received NSF Career Award in 1999 and NCSU New Inventor's award in 2000.