Supporting Emergency-Response by Retasking Network Infrastructures

Recent events have demonstrated the susceptibility of conventional network infrastructures to both man-made and natural disasters. The attacks on September 11th, 2001 disrupted many communication channels that were routed through the World Trade Center, and the mass panic that ensued also caused the telephone switching network to collapse [2]. Even more significant disruptions to communication channels occurred when Hurricane Katrina rendered most of the infrastructure components within its wake partially or completely inoperable [4]. This caused great difficulties for both the victims of Katrina and those who were working to save them. Network connectivity is very important in the aftermath of a disaster, as it can be used by victims and rescuers to communicate among themselves, send messages out to unaffected areas, and receive critical information from external sources. In the rescue operation that followed Katrina, it would have been helpful to rescuers if victims had been able to communicate their locations to rescuers, rather than forcing them to search every house. Analysis of the Kobe earthquake also cited a lack of communication as a cause for delayed emergency-response actions and a mis-direction of resources to areas that had less urgent needs than other areas [14]. Thus, it is clear that resilient data networks could have provided great benefits in the aftermath of this disaster, and the many others like it that occur every year. After Katrina, the only significant, operational network in New Orleans was a Wireless Mesh Network (WMN) used to transport data from security cameras in the city [7]. City officials used this network to provide the services normally provided by other networks, such as voice messaging (VoIP) and general police communications. A number of other major cities are now planning to deploy dedicated mesh networks to improve the robustness of the information infrastructure used by government personnel. Independently, many commercial mesh networks are being deployed for various purposes. For example, traditional electric meters are being replaced with advanced meters that have computational capabilities and are often connected to the Meter Data Management Agency (MDMA) using mesh networking [11]. Buildings are also being enhanced with mesh networks for building automation [5]. Mesh networks are more resilient to node failures than other types of networks, which makes them a logical choice for such applications. However, like any other infrastructure improvement, it is often expensive to deploy mesh networks on a wide scale. Even if a mesh network is deployed, the number of nodes it contains must be based on the amount of functionality and value it provides. If a mesh network carries only government communications, the level of value may be relatively low compared to commercial networks, ultimately causing the governmental mesh networks to be smaller than commercial networks. Large, dedicated Emergency-Response Networks (ERNs) are even more difficult to justify, particularly if there is a low probability of a disaster happening in some covered area [10]. Ideally, all significantly-populated areas should be covered by ERNs since disasters can occur anywhere, so other solutions are required. Additionally, as a general principle, rarely-used systems tend to be poorly maintained and less likely to function properly when required. This suggests that the best path is the retasking of existing networks for ERN purposes in times of disaster. In such conditions, the primary purpose of the network may not be necessary anyway, as is the case with advanced electric meters that are not required to transmit measurements when a power outage has occurred. In fact, the economic interests of the network owner may be furthered by supporting emergency response if their revenues are tied to activities in the affected region, since the availability of ERN may permit the affected region to recover more quickly and return to normal business. It may be possible to modify existing networks to provide this service, but we also discuss requirements for future networks that will ensure they can be retasked to support emergency communications when necessary. How is it possible to use networks for emergency communications if they were originally intended to provide a different service while also ensuring that the original service is not adversely affected during ordinary operations? There are at least three primary considerations in answering this question: detection, platform support, and topology. First, it is necessary to establish the policies and mechanisms by which devices within the network will detect the presence of valid emergency conditions and adapt to them. Second, it may be necessary to have special emergency-response hardware and software platform support provided by devices both internal and external to the network. Third, it may be desirable to anticipate the support that will be provided by the fixed network topology itself, to ensure that ERN services are available regardless of the presence or absence of mobile nodes that may provide ad-hoc infrastructure enhancements. Of course, it can be beneficial for an ERN to permit mobile nodes to join the network and offer routing services to extend its coverage and bandwidth, but such nodes can not necessarily be relied upon as emergency service providers, since their locations may be unpredictable. The aim of this paper is to discuss these three challenges with respect to technologies appropriate to ERN retasking. Our primary technical proposal is