This week twenty of us met at NIST to discuss situation awareness during emergencies. The centerpiece of the conversation was the NG911 system, or Next Generation 911. NG911 supports better interaction between call centers, and uses policy-based security to let other local call centers, or even centers in other regions take calls as circumstances and policy require. Private call centers, run by alarm services, can be full peers if local policy allows. Even buildings, and building systems, might act as 911 operators. The conveners asked me to lead an effort to develop a security model for these systems.
Interoperability at this level requires standardization of security, of policy management, and even of building semantics. 911 centers will accept even the most basic calls, the narrative that includes the confused description of flames from a building from the confused homeless person without a proper address. Such calls require further guess work and human attention before they can be dispatched. Without standards, other calls are no better once they hit the system.
During a major disaster, the local 911 center may be swamped, or even destroyed. The communication lines to the 911 center may be cut off. When NG911 is deployed, calls to 911 can be automatically re-routed to nearby call centers, which will be able to dispatch calls to first responders just as does the local center. If the operator captures all information accurately, and the local policy permits, the NG911 can dispatch the call irrespective of the originating operator's location. Key elements include the type of emergency, the full address of the emergency, and the geo-location of the emergency.
Private alarm monitoring companies originate many 911 calls. Alarms go off, monitoring systems are checked, owners and tenants are contacted, and, if appropriate, 911 is called. If local policy allows it, this call can go directly from the Private alarm monitoring company's dispatch system into the NG911 network, perhaps even into automatic dispatch; no critical time would be lost in re-entering data. With proper standards, this extra information can be shared through the NG911 system.
Intelligent buildings, in concept, can initiate their own 911 calls. They have the advantage over the indigent of knowing their own address and geo-location. They may know what type of emergency is underway, and where in the building the problem occurred. With standards, they can supply the supporting detail very well; they may have to strain to provide the basic narrative.
911 centers want to be able to call back the initiator, to get more detail. Buildings do not always answer the phone. Call operators could access the building systems directly, but only if the information is made more useful and accessible than we find in the typical building control system. The requirements for creating those semantic standards are similar to those needed for transactive energy management. These standards will rely on descriptive information found in building information models (BIMs) such as NBIMS and buildingSmart.
We want standards for whoever comes to the site wants the same sort of access. This access may be a simple as floor plans or intimate as access to video surveillance. The responder may want to understand what the building systems are telling him and use them to clear smoke from the 4th floor or to shut off additional air. The responder may want to find where hazardous material is stored, or even the last known location of building occupants. Making this information and control simple and easy to access, no matter what brand of system in the building, will require abstraction, standards, and service oriented architectures in the building.
These interactions require careful service definitions, security interactions, and policy. In normal times, the building owner cannot accept passers-by able to see and interact within the building. Access to the video surveillance network for the third floor during an incident does not warrant access on other floors or to archives from the day before.
Good security is always about situation awareness. In an emergency, the situation awareness of the emergency includes awareness of its context within the building, and awareness of the informational context of the building. Awareness of the person accessing these systems will include federated identity management, as it might include the 911 operator, the local police, and even the fire department from the next town over.
I am heading up a group tasked by NIST with defining a security framework for this context in January. We have a good group working on this, but this type of work needs many different perspectives. Drop me a line, or comment here, and let me know what you think.