Abstract: Terrorist attacks have always been a major threat to human security. While the frequently terrorist attack occurs all over the world, it is increasingly more urgent for researchers to analyze terror activities deeply. While the revolution in the science of game theory has emerged in recent decades, previous research work in the field indicates it is possible to make prediction and reveal the patterns of terror activities through statistical method and behavioral analysis. Dealing with the location problem for the terror response facilities, we take the characteristics of multi-stage and dynamics into account. In this paper, we apply a method called adversarial risk analysis which represents a combination of statistical risk analysis and game theory without requiring common knowledge in the calculation. According to the Bayesian Decision Theory, the idea of sequential game theory and strategic thinking, we construct a multi-stage model of anti-terrorist facilities location based on adversarial risk analysis. Standing in the shoes of both attackers and defenders, we analyze the decision-making influential diagrams respectively and provide a global and comprehensive computing framework. We expand the previous model just including one resource facility, consider fixed number of facilities and aim to minimize the loss of terrorist attack by proactive and reactive measures. This comprehensive framework could also be applied in the case of multiple terror attacks at the mean time. In the end, we conduct several programming simulations in the background of Shanghai districts and compare results towards different parameters and assumptions. The numerical results show that the adversarial risk analysis model is an effective and feasible method to obtain the best location solutions under various resource scheduling.

Key words: anti-terrorist facilities location, adversarial risk analysis, Bayesian decision theory, multi-stage optimizations

摘要： 恐怖袭击一直是人类安全的重要威胁之一。随着当前恐怖袭击在全球的频发,对反恐问题的研究更加急迫。针对反恐设施选址问题,考虑资源分配的多阶段性以及动态性,根据贝叶斯决策理论和序贯博弈思想,构建了多阶段反恐设施的敌意风险分析决策模型。讨论在城市多个设施点离散选址的不同情况下,通过预防性和修复性的资源分配将恐怖袭击的损失降低到最小。以上海市区县网络为例进行编程仿真测试,结果表明,模型可给出不同给定资源设定下的最优选址方案,是一种有效并切实可行的分析方法。

关键词: 反恐设备选址, 敌意风险分析, 贝叶斯决策理论, 多阶段优化理论