An Agent-Based Approach for Dynamic Load Balancing Using Hybrid NSGA II

To date, there have been many observations about load balancing on different machines. Many researchers identified load balancing as a key component for scheduling problems, as strongly NP-hard. Techniques to solve scheduling problems are frequently unfeasible. Metaheuristic techniques are more generic and applicable to solve wider range problems. We choose NSGA II in genetic algorithms because they are known for parallelization, use probabilistic selection techniques, and multiobjective evolutionary algorithm enriches the dynamic performance. This analysis sheds light on using agents to perform decision-making and scheduling operations. In this paper, we propose and implement ANSGA II which we believe is the first of its kind, an agent-based hybrid model which provides Pareto front solutions whose individual solutions will satisfy multi-objectives. This multi-agent-based approach fastens the algorithm performance. Computational experiments and results for our proposed agent-based model thrive in results toward optimality, in scheduling.