Session 1C: A Distributed Greedy Auction Algorithm to Mitigate Public Mass Shootings

Session Number

Session 1C: 1st Presentation

Advisor(s)

Henry Hexmoor, Southern Illinois University Carbondale

Location

Room A151

Start Date

28-4-2017 8:30 AM

End Date

28-4-2017 9:45 AM

Abstract

We introduce a distributed, greedy auction algorithm for an Internet of Things decision support system to evacuate civilians during public internal mass shootings. The public building environment is abstracted into a network of doors and rooms. During the emergency, door agents bid against neighboring doors using a safety score that considers crowd congestion, number of exit doors, and proximity to shooter. The score is weighted by a distance decay function of node count to nearest exit door. Room agents decide auction winners by highest score, and users are then alerted to the safest doors through which to evacuate. The greedy system is dynamic, avoids the attacker, prevents crowd congestion, and minimizes evacuee panic. We present the protocol and study its application in a sample environment, which attests to its effectiveness.

Share

COinS
 
Apr 28th, 8:30 AM Apr 28th, 9:45 AM

Session 1C: A Distributed Greedy Auction Algorithm to Mitigate Public Mass Shootings

Room A151

We introduce a distributed, greedy auction algorithm for an Internet of Things decision support system to evacuate civilians during public internal mass shootings. The public building environment is abstracted into a network of doors and rooms. During the emergency, door agents bid against neighboring doors using a safety score that considers crowd congestion, number of exit doors, and proximity to shooter. The score is weighted by a distance decay function of node count to nearest exit door. Room agents decide auction winners by highest score, and users are then alerted to the safest doors through which to evacuate. The greedy system is dynamic, avoids the attacker, prevents crowd congestion, and minimizes evacuee panic. We present the protocol and study its application in a sample environment, which attests to its effectiveness.