An adaptive dual-threshold chunking framework for performance optimization in fog–cloud systems

Abstract

The widespread adoption of Internet of Things (IoT) applications and distributed computing systems has led to the continuous generation of large volumes of data that require efficient processing with low latency and minimal bandwidth usage. Traditional chunking approaches, particularly single-threshold mechanisms, are not suitable for dynamic fog environments due to fluctuating data arrival rates and heterogeneous node capabilities. To overcome these challenges, this study presents a Hybrid Dual Threshold Chunking (HDTC) framework aimed at improving load distribution, reducing latency, and enhancing Quality of Service (QoS) in integrated cloud–fog architectures. The proposed model combines dual-threshold-based resource management with adaptive, content-aware chunking to achieve efficient data segmentation and balanced workload allocation across fog nodes. A simulation-based implementation is developed in which users are clustered using the K-means algorithm, and fog nodes are assigned based on proximity. The system continuously monitors node utilization through upper and lower threshold limits, enabling dynamic decision-making for task allocation and chunking. Implemented in MATLAB and tested on real-world datasets consisting of text and image files, the framework is evaluated against Dynamic Prime Chunking (DPC) and Improved Dynamic Prime Chunking (IDPC). Experimental results indicate that HDTC significantly enhances performance by reducing chunking time and overall execution time by up to 95% and 85%, respectively. Although a slight decrease in delivery ratio and throughput is observed compared to IDPC, the framework improves load balancing and responsiveness, offering a scalable and efficient solution for real-time IoT and fog-based distributed computing environments.