Differential elasticity for network resilience

Networked computers and systems are by nature vulnerable to malicious attacks such as denial of service attacks. Many approaches have been developed for creating resiliency within networks through redundancy in functionality and communication. The underlying principle behind such redundancy is that should one node or one path become unavailable, other nodes or paths can be used in its stead filling the same role. A form of redundancy that has been proposed in the literature consists of k-connectivity where k is a natural number; the higher k, the higher is the redundancy and the more resilient is the network. K-connectivity is the property by which the network will stay connected even after any k-1 nodes have become incapacitated. This type of redundancy can be seen as adding “thickness” to the network. The network is effectively a tissue with k-layers whose integrity will take more than k attacks to be disrupted. While effective, this approach is costly. In this paper, we build on the concept of k-connectivity by choosing to make the network, 1. Combining thickness with elasticity through an adaptive and dynamic concept of connectivity. 2. Using selective directions of elasticity. Not all nodes are subjected to the same pressures and not all tears are equally critical. We use critical paths in the network to select directions of elasticity. Initial simulation results will be used to validate this approach.