TDO: A Novel Tower Defense-inspired Optimization Algorithm

Metaheuristic optimization algorithms are a powerful tool in engineering disciplines as they can be deployed to solve complex and high-dimensional optimization problems that classical methods cannot efficiently handle.Following the mechanics of tower defense video games which use strategic and adaptive skills, this study proposes a new population-based metaheuristic algorithm, Tower Defense Optimizer (TDO).In TDO, every candidate solution is conceived as a defensive tower that adapts its approach using multiple game-like mechanisms such as positioning of the tower, specialization of types, upgrade system, waves of difficulty, and resources.The algorithm incorporates multi-strategy search behaviour in which towers are sorted into four different typesexploitative, explorative, balanced and adaptiveto achieve a good balance between exploration and exploitation during the optimization procedure.To improve the search robustness, TDO is equipped with elite guidance, local intensive search with a considerable period during the boss wave, and diversity recovery via opposition-based learning.The performance of TDO is assessed on thirty benchmark functions taken from CEC2017 suite and compared with the performance of ten baseline metaheuristic algorithms on similar budgets.The experimental results show that TDO achieves competitive and often better performance than comparable algorithms.This is especially true with multimodal optimization problems where it wins 70% of the time.The analysis found TDO to outperform most other competing algorithms while exhibiting consistent convergence behaviour and well-behaved computation.In light of these revelations, it can be stated that game-inspired optimization paradigms like TDO have great potential to inspire the design of novel robust and adaptable metaheuristic algorithms.