As sensory information becomes more advanced, the government is faced with a continuous and ever-expanding stream of real-time information from which it collects intelligence in support of national security. Thousands of small, seemingly insignificant events happen every day. ObjectFX has created a solution that fits within Complex Event Processing.CEP helps to automate incident detection and is enabling applications in government to streamline processes, while more quickly identifying what's important. Geospatial analysts, working toward the same goal of most efficiently identifying actionable intelligence, are turning to Geospatial Event Processing to correlate space- and time-relevant events to determine when a significant event has occurred. Editor in Chief Joe Francica posed some questions to Steve Panzer, the vice president of ObjectFX's Government Division.

Joe Francica (JF): Explain in some detail how Complex Event Processing is defined and what location-based elements might be included in CEP analysis.

Steve Panzer (SP): CEP is the processing and possible correlation of multiple events with the objective of identifying meaningful events and/or patterns within the mass of incoming data streams. CEP typically looks at real-time data to determine complex patterns or relationships. Within certain markets, including defense and intelligence, the geospatial and temporal aspects of data are key indicators of the relevance of patterns. For example, whether two ships were in close enough proximity and slowed down sufficiently to exchange contraband may make a new ship that came close to a "high interest vessel" in turn a suspect vessel or candidate for inspection.

JF: What organizations are using CEP today to support national security applications?

SP: Defense, intelligence and even law enforcement are all using CEP for analyzing data and looking at patterns that could determine whether one or more people or objects were co-located in space and time. In some cases, a quantitative threshold of events within a certain region could be indicative of activities that deserve further investigation or action. For example, one person on a watch list coming into Washington, DC may be a concern, while four or five people on a watch list coming into DC may be an even greater concern.

JF: Explain more about the rule base employed by ObjectFX's SpatialRules in the context of spatio-temporal analysis. How much is defined by the user and how much is based on the algorithms developed by the software solution?

SP: The spatio-temporal rules engine monitors whether objects or events are inside/outside of a region, whether they have entered/exited a boundary, how close in proximity they came (both spatially and temporally), whether their paths actually crossed or became sufficiently close, and whether a cluster or threshold of events were exceeded. The user has control as to the data sources, boundary conditions (from ingested geo-political boundaries or user defined regions), and rule definitions (including Boolean logic using multiple rules). So one example may be the tracking of DOE HAZMAT vehicles carrying spent nuclear fuel or toxic materials, which are supposed to stay on major interstates, but are allowed to deviate from that route by a certain distance to allow them to make gas stops. Another example is a shipping container whose door is instrumented to send an alarm to a central facility if opened in transit, but not if that occurs within a customs inspection yard.

JF: What type of sensor feeds can be accommodated by SpatialRules, and does it follow any of the OGC specifications as part of the SensorML spec?

SP: SensorML is a spec that deals mainly with the description and discovery of the sensors themselves, not really with the transmission of sensor data. SpatialRules does accept a number of standards based data types, including the OGC's Geography Markup Language (GML), Keyhole Markup Language (KML/KMZ, which have been given to the OGC by Google), GEORSS and others. The inputs and outputs from SpatialRules are very adaptable, so other projects have used the OGC Sensor Web Environment, the Sensor Observation Services (SOS) and the Sensor Alert Service (SAS). Some of these extensions were used during ObjectFX's participation in Empire Challenge 08.

JF: Where do you see SpatialRules fitting in to an enterprise IT stack? Is this a middleware solution or an application that works in conjunction with other GIS solutions?

SP: SpatialRules can be used anywhere in the IT stack, as either middleware or as part of an application. The software may be used as a spatial alerting service either independently or in conjunction with a GIS visualization engine (e.g. ESRI, Google, SpatialFX, etc.). In fact, the SpatialRules engine may be concurrently employed at numerous places within the processing chain, including as a filtering device close to the sensor to only allow the passage of relevant data, in order to make the best use of limited bandwidth.

JF: Which industries do you believe could realize a substantial return on investment by more fully integrating spatio-temporal analysis into an existing mobile resource management (MRM) or field force management solution that uses GIS today?

SP: Any organization managing a field force that wants to maximize the use of personnel and other assets would be a candidate for SpatialRules. Analyzing where there are too many people, vehicles or specialized equipment at one site, versus a lack of the same at another location is a proper use of rules. Determining which service calls or deliveries are taking longer than normal is a way of monitoring and hopefully optimizing operations. That's pretty much the case across the board for government, utilities, transportation/logistics or supply chain management.