This project is aimed at developing and experimentally evaluating a framework for stateless quality of service routing in IP networks (SQUARN). The need for QoS routing solutions has been recognized in several recent studies and has led to several ongoing efforts to introduce QoS extensions in IP routing protocols. Until now, the work on QoS routing has been carried out under the assumption that the underlying network architecture is reservation based. In such an architecture, routers must maintain per-flow state information and end systems must use explicit signaling messages to convey their QoS requirements. This explicit-reservation model, though attractive from a network control viewpoint, suffers from scalability problems. The proposed SQUARN model is intended to provide stateless QoS routing solutions that can be integrated into reservation-less QoS frameworks (e.g., DiffServ). SQUARN relies on dynamic packet forwarding mechanisms that make novel use of the TOS and TTL fields in the header of an IP packet. It also uses a probing procedure for exploration of viable QoS paths. Link metrics are disseminated in a scalable fashion using a subset of routers that form a "dominating" graph. "Repair" algorithms are used to dynamically update this dominating graph. Protocols and algorithms in support of the SQUARN framework will be developed in the course of this project, with emphasis on computational efficiency and backward compatibility with existing Internet routing protocols. The research agenda also includes an experimental component for evaluating the efficacy of the proposed solutions on a testbed of IP routers.
Besides its intellectual merits, the proposed project is expected to have a profound broader impact. The enabling of stateless QoS routing solutions will lead to successful commercialization of real-time voice (and video) applications, resulting in a significant reduction in long-distance phone rates and wide deployment of real-time multimedia applications (e.g., video telephony over IP). The source code of solutions developed in the course of this project will be publicly released. Various aspects of the proposed project will be integrated into undergraduate and graduate courses at the University of Arizona. The project will be used as a vehicle to promote the understanding of and appreciation for networked multimedia applications among students.
