Extracting Intelligence from Sensor Data

Nigel Elkan

Whether the issue is global security or campus safety, our responsibility is to correlate the available information, identify the risks, and respond accordingly. We have the data. We have the sensors. Data are being captured, whether through simple devices (such as RFID tags, GPS devices, or door sensors) or devices capturing raw data that is then parsed by an intelligence engine (such as recognizing a face, or a torpedo, or certain words in a conversation). The transformation of these data into actionable intelligence requires that we recognize the significance of individual or correlated events and ensure that the appropriate escalation and responders are informed. The key to "connecting the dots" is the ability to recognize and leverage the correlation of real-time sensor events with appropriate intelligence knowledge bases.

Until recently, the possibility of integrating data from diverse sensors—whether signals from discrete devices or recognized words or patterns parsed from multimedia conversations— ranked somewhere between "esoteric" and "science fiction." The challenge isn't whether it is feasible to correlate sensor data from any one sensor source with any other set of current or future sensors. The challenge is to ensure that correlated events can be anticipated (that the scenario is predictable) and/or to guarantee a timely and appropriate response by providing real-time correlation of any of these potentially trillions of data events.

As Prof. Alex Buchmann of Darmstadt University of Technology describes it, "Regular programming is like drinking with a straw: this is good when the data is standing. However, if the data is moving, like in event processing, using the same kind of thinking is similar to using a straw to drink from a waterfall."

In other words, the thinking and techniques used for interrogating and extracting intelligence from (standing) databases is completely different from the techniques and thinking required to extract meaning and response from an almost infinite set of combinations and permutations from real-time data waterfalls. Furthermore, sensor data are capable of being applied to applications well beyond the scope and purpose of their initially conceived mission.

As such, the event-based middleware infrastructure required to correlate these real-time events has to be similarly flexible. These event-based intelligence systems need to interface with sensor types that range from the simple, such as a door sensor, to highly complex sensor applications, leveraging geospatial, temporal, probabilistic, pattern-matching and anomaly management. To accommodate these requirements, the event-based middleware must be open, scalable, and flexible. The middleware must, of course, be able to capture and correlate event data from each and every available and relevant source.

For most organizations, the implementation of Events-Based Middleware is not a "Big Bang", it is an evolution. The organization will have some immediate requirement that requires capabilities that are unmet by the existing complement of access control systems, cameras, analytic tools, alert and intrusion systems, etc. So, rather than replacing these systems, the organization can add the specific new capabilities, and integrate these new capabilities with the existing systems. Following this smart evolution, the organization can gradually enable, as priority and budgets allow, the comprehensive integration of each of these existing security, safety, and environmental controls with the next generation products, and integration and compliance with outside entities, such as the Department of Homeland Security.

The way in which we relate to computing systems and expect to be able to interact with information is changing in a fundamental way. The driving force behind the momentum that has resulted in mainstream adoption of smart phones, smart pads, and an ever growing portfolio of real-time communicating devices is the realization and now expectation that we can be kept up to date with the real-time, relevant, and actionable events that may impact our lives. This has shifted from being possible to being a expectation—indeed we feel orphaned when we lose the connection.

Event-based middleware enables real time actionable intelligence. In many respects, this information has qualities that are dramatic improvements on traditional data, as the data that is provided is delivered is personalized and specific to the recipient's current location, needs and interests and that specific moment in time. Connecting to and living within our real time event-driven world does not replace or demean traditional databases and knowledge repositories. Indeed the value of these repositories is increased enormously by our ability to recognize and integrate the real-time event stimuli with the knowledge bases that enable and inform our ability to respond appropriately.

In any given situation we, as intelligent animals, will gather all available information that our senses will allow, and apply that data to our experiences and knowledge. We will then respond accordingly. Our interpretation of events, and our experiences and knowledge bases differ and, as such, different people will respond to different events in different ways. In a security, safety, and environmental management environment, some of these correlations are clearly defined and indeed some of the responses may be mandated by law. As we build our portfolio of sensors and our ability to establish and maintain the correlations between these sensors, we will find it ever easier to adopt best practices—to propose, test and build the science behind the correlations and, as such, recognize and respond to situations (and anomalies) that could be predicted to represent a risk or an opportunity.

Event-based middleware enables the controller of the events dashboard to make the most informed decisions. These decisions will take into account information and behaviors that are not in the response scenarios. Yet.

Nigel Elkan is VP of Business Development for
Knowledge Vector International LLC, Cary, NC; 919-569-5258.

Suggested Articles

Hydrogen refueling stations are limited in the U.S., restricting interest in use of fuel cell electric cars

Silicon Labs is providing the BT module needed for detecting proximity with another Maggy device

Test automation won't fix everything, but can help, according to an automation engineer. Here are five problems to avoi to improve chances of success