Swarming Pattern Analysis to Identify IED Threat

At a tactical level, insurgents planning attacks with Improvised Explosive Devices (IEDs) are constrained in their choice of target by the specific location of their safe house or weapons cache, the geographic context in which they operate, and the pattern of potential targets as it presents itself at a given time. Geographic profiling in law-enforcement already takes advantage of similar constraints to identify possible origin locations of serial offenders. We show how geographic profiling of past IED events can significantly enhance our ability to identify areas at risk for future attacks. Specifically, we introduce three tightly coupled swarming pattern analysis models (profiling, clustering, forecasting) that refine each others' conclusions dynamically and point to systematic evaluation experiments that confirm the research hypothesis.