Abstract Malicious insiders are one of the most serious threats to an organization’s information assets. The threat is also extremely difficult to mitigate: an insider can be more knowledgeable than an external attacker about the target system and is, therefore, more effective at defeating security controls that mainly defend against external attacks. A promising technique for addressing the insider threat is to accurately predict an insider’s moves and identify the optimal defense strategy. To this end, we propose a game-theoretic model for the insider problem, which we call an “insider game”. An insider game is built on a stochastic game, a game played in a non-deterministic state machine that can describe most computing systems. The model captures other key properties, especially the system administrator’s uncertainty about the system state due to the insider’s hidden action. The efficacy of the model is demonstrated using a real-life incident. Equilibrium strategies are computed to predict the insider’s actions and identify the best way to respond to them.
[1]
A. W. Tucker,et al.
Contributions to the Theory of Games
,
1953
.
[2]
J. Filar,et al.
Competitive Markov Decision Processes
,
1996
.
[3]
H. W. Kuhn,et al.
11. Extensive Games and the Problem of Information
,
1953
.
[4]
Dawn M. Cappelli,et al.
Management and Education of the Risk of Insider Threat (MERIT): System Dynamics Modeling of Computer System Sabotage
,
2008
.
[5]
Thomas Bozek,et al.
Research on Mitigating the Insider Threat to Information Systems - #2
,
2000
.
[6]
A. Arora,et al.
Optimal Bidding in Sequential Online Auctions
,
2003
.