Air Combat Tactics Optimization Using Genetic Algorithms (ACTOGA)

We propose to investigate the feasibility of using stochastic genetic algorithms (GAs) for the development and optimization of air combat tactics for large MvN engagements. Stochastic GAs, which encode each optimizing parameter as a Gaussian region rather than a fixed value, offer several benefits over conventional GAs when the problem under study contains a large number of parameters, as is the case in a large MvN engagement. We plan to use the well-established MIL-AASPEM simulation model as the test & evaluation environment, to bypass a simulation development effort and concentrtate our Phase I study on creating a framework for GA-based tactics optimization. We plan to use a hierarchial concent that builds large formation tactics from small conventional fighting units, to provide a straightforward development path for large group tactics development, while ensuring that we deal with tactics that are compatible with how fighter pilots learn and implement their techniques. We propose a five-task development effort, consisting of: 1) problem scope and requirements definition; 2) development of ACTOGA software libraries and performance measures; 3) integration of ACTOGA prototype with MIL-AASPEM; 4) demonstartion of system performance; and 5) requirements specification for full-scope implementation and commercialization.