Enhancement of Metaheuristic Algorithm for Scheduling Workflows in Multi-fog Environments

  • Omed Ahmed

Student thesis: Doctoral Thesis


Whether in computer science, engineering, or economics, optimization lies at the heart of any challenge involving decision-making. Choosing between several options is part of the decision- making process. Our desire to make the "better" decision drives our decision. An objective function or performance index describes the assessment of the alternative's goodness. The theory and methods of optimization are concerned with picking the best option. There are two types of optimization methods: deterministic and stochastic. The first is a traditional approach, which works well for small and linear problems. However, they struggle to address most of the real-world problems, which have a highly dimensional, nonlinear, and complex nature. As an alternative, stochastic optimization algorithms are specifically designed to tackle these types of challenges and are more common nowadays. This study proposed two stochastic, robust swarm-based metaheuristic optimization methods. They are both hybrid algorithms, which are formulated by combining Particle Swarm Optimization and Salp Swarm Optimization algorithms. Further, these algorithms are then applied to an important and thought-provoking problem.

The problem is scientific workflow scheduling in multiple fog environments. Many computer environments, such as fog computing, are plagued by security attacks that must be handled. DDoS attacks are effectively harmful to fog computing environments as they occupy the fog's resources and make them busy. Thus, the fog environments would generally have fewer resources available during these types of attacks, and then the scheduling of submitted Internet of Things (IoT) workflows would be affected. Nevertheless, the current systems disregard the impact of DDoS attacks occurring in their scheduling process, causing the amount of workflows that miss deadlines as well as increasing the amount of tasks that are offloaded to the cloud. Hence, this study proposed a hybrid optimization algorithm as a solution for dealing with the workflow scheduling issue in various fog computing locations. The proposed algorithm comprises Salp Swarm Algorithm (SSA) and Particle Swarm Optimization (PSO).

In dealing with the effects of DDoS attacks on fog computing locations, two Markov-chain schemes of discrete time types were used, whereby one calculates the average network bandwidth existing in each fog while the other determines the number of virtual machines existing in every fog on average. DDoS attacks are addressed at various levels. The approach predicts the DDoS attackā€™s influences on fog environments. Based on the simulation results, the proposed method can significantly lessen the amount of offloaded tasks that are transferred to the cloud data centers. It could also decrease the amount of workflows with missed deadlines.

Moreover, the significance of green fog computing is growing in fog computing environments, in which the consumption of energy plays an essential role in determining maintenance expenses and carbon dioxide emissions. The implementation of efficient scheduling methods has the potential to mitigate the usage of energy by allocating tasks to the most appropriate resources, considering the energy efficiency of each individual resource. In order to mitigate these challenges, the proposed algorithm integrates the Dynamic Voltage and Frequency Scaling (DVFS) technique, which is commonly employed to enhance the energy efficiency of processors. The experimental findings demonstrate that the utilization of the proposed method, combined with the Dynamic Voltage and Frequency Scaling (DVFS) technique, yields improved outcomes. These benefits encompass a minimization in energy consumption. Consequently, this approach emerges as a more environmentally friendly and sustainable solution for fog computing environments.
Date of Award10 Oct 2023
Original languageEnglish
SupervisorJoan Lu (Main Supervisor) & Qiang Xu (Co-Supervisor)

Cite this