Hybrid fuzzy clustering with Golden Eagle Optimization Algorithm for fault tolerant load balancing in fog computing environment
Fog Computing (FC), when integrated with emerging 5G technologies, provides significant potential to reduce latency and enhance Quality of Service (QoS). Nevertheless, current scheduling strategies tend to lack the in-depth fault-tolerance and load balancing provisions as virtual machine (VM) health indicators, including CPU utilization, memory status, and reliability are not specifically taken into account during task assignment. To overcome these difficulties, this paper will propose a Hybrid Fuzzy Clustering with Golden Eagle Optimization Algorithm (HFC-GEOA). The framework incorporates capacity aware VM clustering into three levels (HCC, MCC, LCC), fuzzy-based task prioritization and localized metaheuristic optimization, which allows efficient mapping of tasks to VM under resource constraints and adaptive inter-cluster migration. Extensive scaling simulations in iFogSim2 using five task-VM scenarios (1,000-20,000 tasks; 20-1,500 VMs) of 30 independent runs demonstrate that HFC-GEOA is a strong and consistent scale performer. In large scale scenarios HFC-GEOA is used to reduce the average turnaround time by up to 78% over FGELB, 71% over ACO-LWC, and 85% over FCLB. The greatest scenario shows improvement of up to 87% of the average wait time over FGELB and 81% improvement over ACO-LWC and 91% improvement over FCLB. The consumption of energy has been competitive as it has remained 11 to 16% less than ACO-LWC over S3-S5. The stably maintained system reliability at 0.71-0.74 across all scenarios more than twice that of EWOA (0.31-0.33) and the failure rates are kept in control under 2.88 to 3.81%. The fault tolerance score increases progressively when compared to and exceeding EWOA (0.77) and FCLB (0.71) at large scale, and as the system resilience to scale improves. Overall, HFC-GEOA offers a scalable, health-conscious, and fault-tolerant scheduler solution that fully leverages the best latency performance at scale with health-conscious energy usage and stable reliability in heterogeneous fog computing systems.