Fault Tolerance in Fog Computing: Hybrid Approach

Fog computing enables low-latency services for Internet of Things (IoT) applications, but its distributed and resource-constrained nature makes it highly vulnerable to failures. Existing fault tolerance strategies such as replication, resubmission, and migration either increase overhead or lack adaptability under dynamic workloads. This paper presents a hybrid fault-tolerant scheduling framework that integrates fuzzy logic-based task criticality assessment with a dual backup resubmission mechanism. The fuzzy module classifies tasks using deadline tightness, node reliability, queue pressure, and data size, while the scheduler selectively creates backups for high critical tasks and resubmits low-critical tasks upon failures. This design aims to minimize energy overhead while maintaining high reliability. The framework was implemented on the iFogSim simulator using a three-tier IoT fog cloud architecture and evaluated under varying failure rates (1–10%) and workloads (1000–5000 tasks). Experimental results show that the approach improves task reliability by up to 20%, reduces makespan by 30% and energy consumption by 25%, and achieves 220 tasks/s throughput at peak load. These findings confirm that fuzzy-guided hybrid scheduling effectively balances reliability, performance, and energy efficiency in fault prone fog environments.