AORTA: Autonomic network control and management system

Abstract

This paper reports on an autonomic network management architecture based on the concept of "evolution". A management methodology is developed which is relying on the ideas from evolutionary science, virtual networks, and autonomic networking. We argue that any communication network could be modeled as an evolved topology based on survivability and performance requirements. The evolution is in the direction of decreasing the chance of congestion and increasing the network robustness. We describe the architecture of our network management system in detail and tie it to the theory of evolution. We evaluate the "betweenness centrality" of network topologies and build our robust routing algorithm to manage the transport of packets in the network based on it. This routing scheme is at the heart of out proposed network management system.

Centrality-Based Network Coder Placement for Peer-To-Peer Content Distribution

“…Betweenness centrality [5] measures the degree that a node stands on the shortest paths between other nodes, which has been applied in different contexts such as routing and cache placement [24,25] to place a network function in a set of nodes. Since we are interested in distributing data from the source to all peers, all shortest paths under consideration are from the source to other nodes.…”

Managing the future internet through intelligent in-network substrates

“…For instance, in the process of shifting traffic away from highly utilized links, the different reacting nodes can re-direct traffic flows towards the same links, as depicted in Figure 3 potentially causing new congestion. In the example, source nodes n1 and n2 both contribute to the load of link l [5][6] , (Figure 3(a)) which becomes the most utilized link in the network. If both ingress nodes react by performing reconfigurations locally, more traffic will be routed towards link l [3][4] (as alternative paths to reach their original destinations) which can then become overloaded (Figure 3(b)).…”

Comparison of network criticality, algebraic connectivity, and other graph metrics

“…A final result is that in general by changing an ELG-like topology to a LAG, we do not lose too much, as a matter of fact the average node betweenness even decreases, while the network criticality (effective resistance) slightly increases. The average node betweenness has an important role in developing traffic engineering algorithms for communication networks [8].…”