Fog-Cloud Interpolation of Urban Monitoring Data Collected by LoRaWan Networks

The increasing deployment of distributed sensor networks for environmental monitoring introduces challenges in data collection. Traditional Cloud-based architectures often struggle with high latency, bandwidth constraints, and scalability issues, especially when handling spatially dense and high-frequency sensor data. To address these challenges, we propose a Fog-based interpolation and Cloud aggregation framework that distributes computational tasks across Edge, Fog, and Cloud layers to enhance the efficiency of smart sensing applications. Our approach exploits localized spatial interpolation methods (e.g., Inverse Distance Weighting and Radial Basis Functions) directly at Fog nodes, allowing for fast computation and reduced data transmission to the Cloud, which remains only in charge of refining and merging the interpolated datasets. The proposed hierarchical data processing strategy minimizes network usage and enhances scalability due to local processing at the Fog layer, making it ideal for large-scale smart computing systems. The paper focuses on the analysis of the trade-offs between computational overhead, accuracy, and data transmission efficiency through an experimental validation conducted using real-world vehicles' paths. Results demonstrate that Fog-assisted interpolation can reduce latency, achieving up to 81% improvement in some cases, while maintaining accuracy comparable to a global approach in the Cloud.