Security Games with Restricted Strategies: An Approximate Dynamic Programming Approach
bookPart
In this chapter we consider a security game between an agent and an intruder to find optimal strategies for patrolling against illegal fishery. When patrolling large areas that consist of multiple cells, several aspects have to be taken into account. First, the current risk of the cells has to be considered such that cells with high risk are visited more often. Moreover, it is important to be unpredictable in order to increase the patrolling effectiveness countering illegal fishery. Finally, patrolling strategies have to be chosen in such a manner that they satisfy given operational requirements. For example, the agent might be required to patrol some cells more often than others imposing extra restrictions on the agent strategies. In this chapter, we develop a dynamic variant of the security game with restrictions on the agent’s strategy so that all these requirements are taken into account. We model this game as a stochastic game with a final penalty to ensure that the operational requirements are met. In this way, strategies are formed that both consider past actions and expected future risk levels. Due to the curse of dimensionality, these stochastic games cannot be solved for large scale instances. We develop an approximate dynamic programming algorithm to find approximate solutions.
Topics
TNO Identifier
810245
ISBN
978-94-6265-245-3 (print)
978-94-6265-246-0 (online)
978-94-6265-246-0 (online)
Publisher
T.M.C. Asser Press/Springer
Source title
NL ARMS Netherlands Annual Review of Military Studies 2018 Coastal Border Control: From Data and Tasks to Deployment and Law Enforcement
Editor(s)
Monsuur, H.
Jansen, J.M.
Marchal, F.J.
Jansen, J.M.
Marchal, F.J.
Place of publication
The Hague
Pages
171-191
Files
To receive the publication files, please send an e-mail request to TNO Repository.