Smart Offloading of IoT Big Data for Network Resources Optimization

The Internet of Things (IoT) devices generate massive data that leads to network congestion, propagation delays, and suboptimal resource allocation. Traditional Cloud Computing (CC) offers scalable resources required for that data; however, it has a long delay and communication overhead. On the other hand, Edge Computing (EC) guarantees low latency but has limited computational capacity. In this paper, we propose an intermediate paradigm, Regional Computing (RC), combined with a Fuzzy Logic System (FLS) for dynamic, multi-criteria offloading across edge, regional, and cloud. The FLS takes task size, cost, and computational demand as input metrics. It uses a rule-based inference engine to select the optimal offloading tier for each task. We created real-time data using an Arduino UNO R4 and ran it in our Python custom-built simulator, RegionalEdgeSimPy. It is specially designed to simulate IoT environments. Experimentation results show that the proposed strategy reduces average network latency by 50% as compared to CC offloading. The model also reduces costs by 30% in comparison with EC or CC. The framework enhances scalability and responsiveness in IoT big data applications and is representative of a practical solution for real-world deployment.