Simulating attacker and defender strategies within a dynamic game on network topology

Successful military counterinsurgency operations increasingly rely upon an advanced understanding of relevant networks and their topologies. This paper evaluates, via simulations, various attacker and defender strategies within a dynamic game on network topology. The simulation is designed to provide insight into the effectiveness of offensive targeting strategies as determined by various centrality measures, given limited states of information and varying network topologies. Improved modeling of complex social behaviors is accomplished through incorporation of a distance-based utility function. Moreover, insights into effective defensive strategies are gained through incorporation of a hybrid model of network regeneration. Two designed experiments investigate the impact of game features on the superlative offensive and defensive strategies. Results indicate that degree centrality, proximal target centrality, and closeness centrality outperform other measures as targeting criteria given varying network topologies and defensive regeneration methods. Furthermore, the attacker state of information is only significant given a topology conducive to defense. The costs of direct relationships significantly impact effective regeneration methods, whereas restructuring methods are insignificant. These results offer preliminary insight into practical attack and defense strategies utilizing a simulation that can be easily adapted for operational applications.