A multi-objective fitness dependent optimizer for workflow scheduling

Workflow scheduling is a significant challenge due to the large scale of workflows and heterogeneity of cloud resources. The vast size of the cloud makes execution times higher, leading to high computational and communication costs. Workflow scheduling is an NP-hard problem, thus, creating meta-heuristic algorithms is one of the best options for finding optimal solutions. This paper models workflow scheduling as a multi-objective optimization problem that considers execution time and communication cost. Optimization efforts are accomplished by proposing a Fitness-Dependent Optimizer (FDO) inspired by bee reproductive behavior. However, it has many drawbacks, including being a single-objective problem. To improve this, we present a Genetic Algorithm-based multi-objective FDO, eliminating many of the previous algorithm's issues. The proposed algorithm takes advantage of both the Genetic Algorithm and FDO. Moreover, it does not show signs of sticking to a local optimal solution. The proposed algorithm is compared with the Genetic Algorithm (GA), Particle Swarm Optimization (PSO), GA-PSO, and FDO, where it shows its effectiveness by performing better on both parameters.