Self-Adaptive Discrete Firefly Algorithm For Minimal Perturbation In Dynamic Constraint Satisfaction Problems

Many real-world problems such as scheduling, planning and resource allocation can be represented and solved as Constraint Satisfaction Problems (CSPs). The main challenge when tackling these applications is the fact that they occur in an evolving environment. That is, constraints might change over time and this can affect the feasibility of the solution found so far. These changes can be captured with the Dynamic CSP formalism that has been proposed and investigated in the literature. More formally, a Dynamic CSP corresponds to a series of static CSPs, each resulting from a change in the previous one as a result of the evolving world. This change corresponds to either a constraint addition or retraction. In this paper, the focus is on constraint addition (also called constraint restriction) and the goal is to search for the most similar solution satisfying the old constraints and the new ones. In this regard, we propose a new method based on the Firefly algorithm for solving this particular problem with minimal perturbation. To assess the efficiency of new technique, we conducted several experiments on randomly generated dynamic CSP instances. The results achieved clearly demonstrate the efficiency of our algorithm, over other known exact and approximation techniques used in the literature for solving these problems.