SBIR-STTR Award

21CT and CUBRC propose CEDAR (Complex Event Discovery, Analysis, and Ranking), a robust framework to detect and analyze indicators of complex activities, such as an insurgent ambush observable via myriad of simple events in multiple sensor streams. CEDAR will provide a mapping from complex behavior, such as conducting an insurgent ambush or rebuilding trust in local communities, to simple event indicators detectable in intelligence, open-source, blue force, and population sensor streams. Examples of such events include increased chatter in blue or red force networks, changes in population sentiment, or curfew movement. CEDAR will also provide an integrated process to execute event queries that leverages our team’s abilities in performing approximate pattern matching over multi-dimensional data. This capability enables us to mine sensor feeds at scale, such as detecting motion in video, non-verbal audio cues, such as gunfire, sentiment in text sources, and changes in human network activity. Finally, CEDAR will provide a suite of event projections to facilitate discriminating aggregate events that indicate complex activity. By combining event projections with semantic event scoring, we can assess if detected events are progressing on a “good” or “bad” vector and exploit this information to choose actions to modulate complex event behavior.

Benefit:
CEDARs extensible framework for detecting and analyzing aggregated simple event indicators of complex event activities will improve the ability of analysts to understand activities with respect to insurgents, the local population, coalition forces, and the environment. By capturing analyst perceptions of perceived event intent, impact, and visibility, we can construct robust projections of event results to discriminate the trends of aggregate behavior in an area towards favorable or unfavorable outcomes. Such a system for detecting and analyzing complex event behavior is of immediate value to intelligence agencies seeking to suggest coalition actions based on multiple sensor inputs. In addition, the complex event analysis abilities of CEDAR will be of interest to government and commercial agencies seeking to combat fraud, understand how events affect logistics chains or other business processes, or to understand the influence of multiple events with respect to offensive and defensive actions in the cyber domain.

Keywords:
Event Ranking, Event Ranking, Complex Event Processing (Cep), Event Chain Analysis, Complex Event Analysis, David Kilcullen, Approximate Graph Pattern Matching, Simple Event D

Successful intelligence operations require the ability to anticipate enemy actions in time to prevent or mitigate their impact. These actions are often part of a complex set of events; for example, an assault on a specific place can involve planning and logistics activity as well as the coordinated movement of multiple actors. Complex event processing (CEP) techniques are a promising technology to enable the detection of emerging threats by monitoring intelligence, surveillance, and reconnaissance (ISR) data for combinations of simple behaviors which are indicative of enemy attack planning. By recognizing emerging complex events in this way, ISR data can be interpreted to extract timely actionable intelligence. 21st Century Technologies, Inc. with academic partner Calspan-University of Buffalo Research Center (CUBRC), is developing the Complex Event Discovery, Analysis, and Ranking (CEDAR) system to meet this operational need. In the STTR Phase 2 effort, we propose to enhance CUBRC’s existing CEP technology to operate in the asymmetric threat domain, where enemy tactics are highly variable and dynamic. We will also perform research to leverage the simple behaviors detected from multiple INT sources. By the end of Phase 2, we will have produced a prototype system that can detect emerging threats on operational data.

Keywords:
Machine Learning, Machine Learning, Entity Resolution, Threat Analysis, Intelligence Analysis, Data