A Fog Computing Architecture for Disaster Response Networks

Abstract

In the aftermath of a disaster, the impacted communication infrastructure is unable to provide first responders with a reliable medium of communication. Delay tolerant networks that leverage mobility in the area have been proposed as a scalable solution that can be deployed quickly. Such disaster response networks (DRNs) typically have limited capacity due to frequent disconnections in the network, and under-perform when saturated with data. On the other hand, there is a large amount of data being produced and consumed due to the recent popularity of smartphones and the cloud computing paradigm.

Fog Computing brings the cloud computing paradigm to the complex environments that DRNs operate in. The proposed architecture addresses the key challenges of ensuring high situational awareness and energy efficiency when such DRNs are saturated with large amounts of data. Situational awareness is increased by providing data reliably, and at a high temporal and spatial resolution. A waypoint placement algorithm places hardware in the disaster struck area such that the aggregate good-put is maximized. The Raven routing framework allows for risk-averse data delivery by allowing the user to control the variance of the packet delivery delay. The Pareto frontier between performance and energy consumption is discovered, and the DRN is made to operate at these Pareto optimal points. The FuzLoc distributed protocol enables mobile self-localization in indoor environments. The architecture has been evaluated in realistic scenarios involving deployments of multiple vehicles and devices.