The Geospatial Complex Event Processing Engine

SATWAYS, December 2016

The INSPEC2T Community Policing system aims to increase the collaboration of citizens with Law Enforcement Agencies (LEAs) by opening various digital communication channels and building communities with mutual interests. These communication channels will result in a complex event cloud in the form of reports, posts and comments originated by the INSPEC2T tools (Mobile application, Portal) and/or existing Social Media data that needs to be processed in an automated fashion in order to produce actionable intelligence.

It is recognizable that when dealing with multiple and disparate events the Complex Event Processing (CEP) technology can deliver high-speed event processing, correlation and identification. CEP patterns emerge from relationships between events attributes, cause (causal relation between events), time and aggregation (significance of an event’s activity towards other events).

Geospatial event processing requires the extension of CEP to include location attributes and spatial relations are used to combine location aware events and to create spatiotemporal patterns. In this context, the GEO Spatial Complex Event Processing (G-CEP) is an INSPEC2T backend component responsible for correlating events in time and space from different modules produced by other INSPEC2T backend and frontend components as well as external systems (e.g. legacy Computer Aided Dispatch, Social Media) and provide actionable intelligence to LEA operators.

Figure 1: Complex rule definition process in the G-CEP Configurator application

The main responsibility of this component is the reception of disparate events –objects (Inbound Data Process), their processing as these object streams flow in and the production of (actionable) intelligence (Outbound Data Process). The correlation of these disparate events can occur via pattern matching that will be applied in real time using time windows and custom spatially enabled algorithms. Spatiotemporal patterns in the INSPEC2T domain are defined in the ESPER knowledge base pattern language via the G-CEP Configurator application and sent to a Rule Manager via a Message Broker.

Figure 2: Correlation of spatio-temporal events in the G-CEP Configurator application

These patterns are used by the pattern matching process. As events are processed (by the Event consumer sub-component from the INSPEC2T Data Warehouse) with or without spatial attributes these are matched against the knowledge base of the Rule Manager. Events can be represented by various methods available in Java programming such as Java objects or Map interface implementations or Object-array or XML documents.