Today’s Internet offers best-effort service, whereby the treatment received by a packet or traffic flow is fully dependent on the network conditions, such as traffic load. However, best effort service is often insufficient for real-time applications that have strict quality of service requirements of network performance. To successfully provide new services and to meet an application’s quality-of-service (QoS) requirements the network must employ effective resource management mechanisms. This paper introduces a new constraint based routing algorithm, termed Active Dynamic Routing (ADR), which employs active network concepts to enhance routemaking decisions within the network. This algorithm is based on cooperation between packet scheduling and routing, two network algorithms, which have traditionally been, viewed as independent resource management mechanisms. In ADR, agents residing in network nodes monitor and record the performance of specific metrics. They then build link metric pdfs whose characteristics are communicated to other nodes.

A longstanding problem encountered with link-state routing is that the information held by the router will always have a degree of inaccuracy. ADR implements a form of constraint based routing that considers only those metrics relevant to the traffic being routed at a specific time. Path decisions are made using metrics that incorporate the traffic conditions at the upstream nodes. However these conditions are continuously changing. ADR increases the longevity of accuracy of information held by a sending router by having the scheduler at an upstream node maintain a packets’ expected relative position within the appropriate queue. This paper compares the performance of ADR to OSPF, a common best effort routing algorithm.