Multi objective adaptive task offloading at the edge using fuzzy logic for time sensitive applications in industry 4.0

Abstract

The rapid growth of Industry 4.0 has increased the demand for real time data processing and ultra low latency communication to support time sensitive Industrial Internet of Things (IIoT) applications. While edge computing reduces latency by executing tasks closer to data sources, the mixture of devices and dynamic workloads often leads to imbalanced task distribution, making efficient offloading a persistent challenge. Traditional static strategies such as Priority Based Preemptive Scheduling (PPS), Minimum Completion Time (MCT), and Round Robin (RR) fail to address multiple conflicting objectives including latency, resource utilization, and task priority under strict time constraints. To overcome these limitations, this research proposes FAST EDGE, a multi objective adaptive task offloading algorithm based on a Fuzzy Expert System (FES). The FES evaluates device suitability using key metrics like processing power, bandwidth, load and distance from the server while modeling uncertainties through fuzzy inference. The system is implemented in EdgeCloudSim and evaluated across diverse IIoT workloads and device configurations. Performance is compared against PPS, MCT, RR, and a non adaptive baseline using latency, completion time, makespan, accuracy, and overhead as metrics. Experimental evaluation in EdgeCloudSim shows that FAST EDGE reduces latency by 28.8 % (3.46 at 2.46 ms) and completion time by 27 % (4.13 to 3.01 ms) while lowering makespan by 29 %, improving decision accuracy by 15%, and reducing overhead to 0.010 equal to 25 % drop compared with the best baseline (PPS). These quantitative gains demonstrate the model’s practical ability to meet 10 ms Industry 4.0 timing requirements under heterogeneous IIoT loads.