Algorithm with Quality-Runtime Tradeoff Parameter for Attack Policy in Attack Graphs for Network Security

Finding the optimal policy in the attack graph is known to be an NP-hard problem. Correlated Stackelberg Equilibrium (CSE) approximates Strong Stackelberg Equilibrium (SSE), which finds the optimal policy profile for two players, a defender and an attacker. This thesis investigates the application of iterative algorithms to reduce the computation time of CSE algorithm. Iterative algorithms allow improving lower bound to the optimal attack policy. It can be used to direct exploration of the attack graph and by that to reduce the complexity of the searching space. Proposed techniques look for a tradeoff parameter between algorithm’s quality and its runtime.