Self-Organization and Topology Control of Infrastructure Sensor Networks

Abstract

Infrastructure networks are complex, interconnected, and inter-dependent systems on which modern society has become almost totally dependent. They provide for cost-effective and efficient distribution of energy, communications, and transportation, yet are increasingly fragile and rapidly propagate failure caused by natural or man-made hazards. Despite our reliance on these networks and our awareness of their risks, an understanding of their survivability and methods for mitigating the risks inherent in their spatial and topologic organization has been lacking.

Infrastructure sensor networks are coupled with infrastructure for health, performance, or surveillance monitoring. They detect the precursors of hazards and allow response to prevent cascading failure. These co-located, dependent sensor networks are themselves susceptible to disruption and require control methodologies to maintain surveillance capability (i.e, survivability) should disruption occur.

This research quantifies the risk and vulnerability associated with dependent sensor networks and investigates the role of topology control and self-organization behavior in mitigating that risk. Simulated random and targeted attacks are performed on spatial and lifeline infrastructure topologies. Spatial topologies are shown to exhibit attack resistance, while lifeline topologies undergo percolation sooner and more frequently resulting in significantly higher vulnerability.

Topology control, or dynamic reconfiguration of the network in response to disruption, is shown to significantly mitigate the vulnerability of infrastructure sensor networks. Its application, however, is limited based on the critical spatial density of nodes placed around infrastructure networks. The critical density of dependent sensor networks is computed and a framework for the self-organization of infrastructure sensor networks is discussed.