Dynamic allocation of security resources for protecting public events

Large scale public events are attractive targets for terrorist attacks. It is of great significance to intelligently allocate limited security resources to protect such events. In most public events, the impact of an attack at different targets changes over time. For instance, in marathon, the impact of an attack at different locations changes over time as the participants move along the race course. In addition, the police can relocate security resources among potential attacked targets at any time during the event and an attacker may act at any time, thus the strategy spaces of both agents are continuous. Furthermore, a certain kind of public events, e.g., the Olympic Games, is usually held infrequently. Thus the attacker does not get an opportunity to conduct surveillance and respond to a distribution of defender strategies. In this paper, we aim to address the security resource allocation problem in public events domain with time-critical payoff, continuous strategy spaces, and low frequency.