Dynamic Risk Assessment (DRA) for mass gathering events

By  Carlo Dambra and Alex GralewskiPROPRS Ltd.
In the management of mass gathering events, Law Enforcement Agencies (LEAs) are collecting weak signals from many different heterogeneous sources:

physical sensors (metal detectors, explosive sniffers, etc.), CCTV-based sensors (intrusion detection, crowd monitoring, car number-plate readers, etc.), humans as sensors (public participating to the event, stewards, policemen in the crowd, etc.), intelligence warnings, cyber-threat intelligence.

The collected weak signals are extremely complex to be interpreted as threat precursors in low probability high impact events (e.g. terrorist attacks) due to: their heterogeneity and numerosity, the intrinsic uncertainty of the sensors and the related processing, the distribution in both time and space.

LETSCROWD has proposed an innovative methodology to process collected weak signals to dynamically assess risks for the crowd by

  • Ranking them according to the Credibility of the detector (usually a trained steward can be considered more credible than a teenager in detecting an abandoned object), the Reliability of the sensor and related processing (a CCTV-based vehicle detector can be misleaded by shadowing in the scene) and the Time Distance between the detection and the event itself (a truck in a forbidden area can be considered differently if it is happening 3 days before or during the event).
  • Grouping them into Suspicious Patterns to be considered as threat precursors according to space-based, time-based and experience-based rules allowing also the operator to group them dynamically.
  • Showing them to the operator on a time-dependent GIS integrated with crowd modelling tools to allow him to take risk-aware decisions and implement mitigation actions.

The proposed DRA methodology will be tested, tuned and validated at LEAs demonstration sites.