ClusFC-IoT: A clustering-based approach for data reduction in fog-cloud-enabled IoT

The Internet of Things (IoT) is an ever-expanding network technology that connects diverse objects and devices, generating vast amounts of heterogeneous data at the network edge. These vast volumes of data present significant challenges in data management, transmission, and storage. In fog-cloud-enabled IoT, where data are processed at the edge (fog) and in the cloud, efficient data reduction strategies become imperative. One such method is clustering, which groups similar data points together to reduce redundancy and facilitate more efficient data management. In this paper, we introduce ClusFC-IoT, a novel two-phase clustering-based approach designed to optimize the management of IoT-generated data. In the first phase, which is performed in the fog layer, we used the K-means clustering algorithm to group the received data from the IoT layer based on similarity. This initial clustering creates distinct clusters, with a central data point representing each cluster. Incoming data from the IoT side is assigned to these existing clusters if they have similar characteristics, which reduces data redundancy and transfers to the cloud layer. In a second phase performed in the cloud layer, we performed additional K-means clustering on the data obtained from the fog layer. In this secondary clustering phase, we stabilized the similarities between the clusters created in the fog layer further optimized the data display, and reduced the redundancy. To verify the effectiveness of ClusFC-IoT, we implemented it using four different IoT data sets in Python 3. The implementation results show a reduction in data transmission compared to other methods, which makes ClusFC-IoT very suitable for resource-constrained IoT environments.