Hybrid Bacterial Colony Optimization and Particle Swarm Optimization for load balancing in fog computing

Area of Science:

• Computer Science
• Distributed Systems
• Artificial Intelligence

Background:

• Fog computing processes data near the source to reduce latency and bandwidth usage.
• Effective workload balancing is crucial for fog computing due to dynamic, resource-limited nodes.
• Uneven task distribution leads to bottlenecks, underutilized resources, and decreased Quality of Service (QoS).

Purpose of the Study:

• To introduce a hybrid metaheuristic load balancing algorithm for fog computing.
• To enhance load balancing by combining Bacterial Colony Optimization (BCO) and Particle Swarm Optimization (PSO).
• To improve makespan reduction and virtual machine (VM) utilization.

Main Methods:

• Developed a hybrid BCO-PSO algorithm in Python.
• Utilized original BCO and hybrid modules alongside a standard PSO executable.
• Tested the algorithm on a synthetic task-VM dataset (100-10,000 tasks) generated by CloudSim 6.0.

Main Results:

• The hybrid BCO-PSO algorithm demonstrated significant makespan reduction and increased VM utilization.
• Outperformed individual BCO, PSO, and Adaptive Inertia Weight Particle Swarm Optimization (AIW-PSO) algorithms.
• Achieved faster convergence, especially in high workload scenarios, reducing makespan by up to 35.79%.

Conclusions:

• The proposed hybrid BCO-PSO algorithm is an effective and adaptive solution for task allocation in fog computing.
• It addresses the challenges of workload balancing in dynamic fog environments.
• The algorithm shows promise for fog-inspired computational scheduling scenarios.