A Dynamic Interdomain Communication Path Setup in Active Network

Abstract

An internetwork is composed of many administrative domains (ADs) with different administrative and security policies for protecting their own valuable resources. A network traffic flow between end-to-end stub ADs through intermediate transit ADs must not violate any stub or transit domain policies. Packets may be dropped by routers that detect a policy violation. Therefore, it is necessary for a communication session to set up a communication path in which all constituent routers are willing to serve for the session so that data packets can be delivered safely without being discarded. Moreover, such a communication path cannot always guarantee successful packet delivery if the intermediate routers or links are prone to failure or congestion. This paper proposes a dynamic interdomain communication path setup protocol to address these issues.

The protocol is dynamic in the sense that the path determination strategy is distributed and a path can be reconfigured to bypass a failed or congested router or link. These two dynamic features require the intermediate network nodes to perform some computation and to make some decisions. The implementation of the protocol relies on the computational capability of an active network in which active nodes in the network can provide computational capabilities in addition to traditional communication. Thus, the design of the protocol is based on the assumption of an active network architecture. The protocol will be a useful tool for all connection-oriented applications in active networks.

The research is partially supported by NSA under contract number : MDA-904-98- C-A892.