Active Dynamic Routing (ADR)

The advent of high-speed networking has introduced opportunities for applications such as video conferencing and medical imaging. These applications have stringent performance requirements in terms of delay, delay jitter and loss rate. Subsequently a central issue in the design of modern communication networks is the provision of performance guarantees. To facilitate the use of real-time communications in the Internet new characteristics and services must be added. In traditional routing, packets are delivered using a route based on their source and destination addresses, while in Quality of Service (QOS) based routing the traffic requirements are also taken into account. The paper describes a novel approach, termed “Active Dynamic Routing” (ADR) that allows routes to be selected based on a range of criteria. Furthermore these criteria (bandwidth, delay, jitter, cost etc) can dynamically vary over the lifetime of a connection, subsequently changing the optimal route. Often the number of hops is the metric used to make the routing decisions but with ADR, customers can use different criteria, and different rules to effectively select an optimal path to their destinations. ADR is based on the concepts of programmable and active networks, whereby the functionality of network elements can be changed or modified by end users or third parties. The ADR approach aims to provide a “preventative” measure against congestion, and is achieved by using agents placed at chosen routers through the network. Selecting a path that provides QoS, requires additional information. In operation, this information, including the traffic profile, can be held and exchanged by the communicating agents. Each agent will have access to read and update a multi-dimensional matrix that identifies the individual weighted costs to move between different specified routers for each class/ type of traffic. A novel aspect of this approach is that the agents take account of the packet scheduling used, along with current status of the queues in each router, in order to derive an optimal route. The matrix is then used to determine the most appropriate path for a given packet or a stream of packets given its/their specific needs