Solving Simultaneous Continuous Multi-Objective FlipIt Games Using Co-Evolutionary Computation

We present a novel extension to the game of FlipIt, introducing infrastructure costs and their impact on the at-tacker's success rate. This extension results in a simultaneous, continuous, multi-variable, multi-objective game, which we alge-braically solve in the case of periodic strategies. We then propose a novel fitness criterion for Co-Evolutionary Algorithms suitable for estimating Pure Strategy Nash Equilibria for such types of games. Afterward, we estimate solutions to the extended FlipIt game when game outcomes are evaluated both via analytical expectancy expressions and as the average of game simulations. The results demonstrate the effectiveness of the proposed framework for deterministic games and also hint towards its applications in stochastic games like FlipIt and even in broader multi-agent settings.