Topology Discovery in Autonomic Networks

The network Management Research Group (NMRG) introduced their own version of autonomic networks based on the viewpoint of the Internet Society and following the definition provided by IBM of autonomic systems. NMRG focused on self-optimizing, self-configuring, self-protecting, and self-healing capabilities in the proposed design model of autonomic networks. Later the Autonomic Networking Integrated Model and Approach (ANIMA) working group of the Internet Engineering Task Force (IETF) designed protocols to support the goals set by NMRG. The proposed autonomic network mitigates the human administration influence as much as possible and make the nodes dependent on themselves and the communications with their neighbors. Therefore, autonomic nodes will act as a network management entity that depends on the information they receive/send from/to their surroundings and their knowledge about themselves. In network management, knowing the network’s topology gives nodes a great advantage toward becoming more autonomic. Knowing the topology can help nodes with management tasks such as link failure recovery, routing, and imposing policy. Topology Discovery (TD) is the process of collecting the neighboring information of all nodes and distributing the processed information among them. Topology Maintenance (TM) takes place after the topology map is generated during the TD process. TM updates all nodes upon the changes in the topology map. The TD and TM can be heavy tasks on the network since they require collecting information from all nodes and distributing it among them. We focus on supporting the benefits of autonomic nodes knowing the network’s topology and suggest efficient methods to collect and maintain the topological information of an autonomic network. Our goal is to minimize the bandwidth consumption by reducing the number of exchanged messages for TD or TM purposes. There have been many approaches proposed to improve the performance of TD and TM. There has been thorough research on TD methodologies but not all the proposed solutions can be applied to autonomic networks. In this thesis, we review different methods for TD and discuss their compatibility with the proposed autonomic network guidelines. We then propose two new solutions. Our first solution is based on a clustering algorithm that allows the autonomic nodes to join clusters and limits the message passing to intra-cluster communications and inter-cluster communication between clusterheads. The second proposed solution is based on taking advantage of the secure boot-strapping protocol (BRSKI) for autonomic nodes to generate the topology map of the autonomic network