Attacker-Defender Strategy Optimization Using Multi-objective Competitive Co-Evolution

Attacker-defender strategy optimization deals with optimizing and deciding on different tactics used by two independent entities working in tandem. Unlike in standard optimization problems, a complete solution of the entire two-agent problem consists of strategies of both agents and evaluation of a solution requires precise information of both strategies. For this reason, a co-evolutionary optimization framework is proposed in this paper to keep two co-evolving populations interacting with each other in tandem to reach their optimal strategies. While co-evolutionary algorithms have been proposed in the past, multi-objective co-evolutionary problems make the optimization task more complex, resulting in a set of Pareto-optimal strategies for each entity. In this paper, we apply a multi-objective competitive co-evolutionary optimization algorithm to a real-world wargame strategy optimization problem. The proposed co-evolutionary algorithm is used to find trade-off sets of competitive wargame strategies for both entities and a novel post-optimization decision-making procedure is also proposed to choose preferred strategies for each entity in tandem, leading to a stable or a cycle of sequential strategies. To the best of our knowledge, this paper marks one of the first-ever applications of multi-objective, competitive, co-evolutionary optimization approaches to a real-world wargame scenario, revealing their impact and importance in practice.