Multi-objective algorithms for effective resource management in Edge-Fog-Cloud computing

Edge-Fog-Cloud computing is a platform that facilitates the processing of loT tasks that generate a massive amount of data from edge computing. Small or delay-sensitive tasks should be sent to fog computing, while complex or large-scale tasks must be transferred to the cloud data center due to its enormous capabilities in computation and storage. However, workload allocation remains a critical concern, involving the allocation of sensitive tasks to edge-fog computing and large complex tasks to edge-cloud computing to meet user requirements based on their specific characteristics. The diversity of task attributes, such as input length, computing unit requirements, and sensitivity to delays, presents challenges in distributing workloads across different computing layers, resulting in both load overhead and increased transmission delays. The second crucial issue is task scheduling, which revolves around efficiently scheduling tasks to suitable resources across various computing layers while considering the unique characteristics of each task. Inefficient scheduling can result in increased transmission delays in edge-cloud computing, particularly due to the long distances involved, as well as higher energy consumption in edge-fog computing. The third problem concerns task offloading. When processing massive Edge tasks, computational devices may unexpectedly shut down due to the network's dynamic nature or power issues, leading to the interruption of task execution and incomplete processing. Offloading uncompleted tasks randomly to any computational node for execution can result in inefficient resource utilization and increased energy consumption. There are three (3) main objectives laid out in this thesis to tackle these issues. First, proposed the Non-dominated Particle Swarm Optimization (NPSO) algorithm for workload allocation to reduce transmission delay in edge-cloud computing and imbalance load degree in edge-fog computing. Second, proposed a Multi-objective Grey Wolf Optimizer (MGWO) algorithm for optimizing task scheduling to reduce transmission delay on edge-cloud computing and energy consumption on edge-fog computing. Third, proposed a Multi-objective Firefly (MFA) algorithm for task offloading to maximize resource utilization on edge-cloud computing and reduce energy consumption on edge-fog computing. Simulations were conducted to evaluate the proposed algorithms compared to the PSO algorithm, Cloud-Fog Cooperation Scheduling algorithm, and Task offloading algorithm. The experimental results prove the effectiveness of the proposed algorithms and outperform comparing them. Thus, the NPSO algorithm reduces the imbalance load degree in edge-fog computing by an average of 6% and the transmission delay in Edge-cloud computing by an average of 25%, respectively. In addition, the MLLF algorithm reduces the maximum delay threshold by an average of 11% compared with other related algorithms. Besides that, the MGWO algorithm reduces energy consumption in edge-fog computing by an average of 32% and the transmission delay on edge-cloud computing by an average of 22% compared to another approach. In comparison, the MFA algorithm reduces energy consumption in edge-fog computing and maximizes resource utilization in edge-cloud computing by an average of 23% and 86%, respectively. Finally, this study has several limitations that can serve as avenues for future research. These include the consideration of heterogeneous resources, the incorporation of additional QoS objectives, and the adoption of machine learning techniques for detecting threats within the edge-fog-cloud computing environment and predicting incoming tasks.

Text 120977.pdf Download (12MB) Official URL or Download Paper: http://ethesis.upm.edu.my/id/eprint/18538

Actions (login required)

View Item