Leveraging Fog Computing for Security‐Aware Resource Allocation in Narrowband Internet of Things

Abstract

Narrowband Internet of Things (NB‐IoT) is LPWAN operating using narrowband spectrum in IoTs, requiring low data rates and long battery life. Since NB‐IoT does not support handover, the requirement to sustain network connectivity during mobility may result in fake base station connections, hence applications are limited to stationary use‐cases. Researchers propose extending NB‐IoT in mobile applications, owing to higher signal quality and battery life, despite DoS attacks due to low bandwidth. As NB‐IoT is resource‐constrained, resources must be allocated based on application with data processing offloaded to optimise performance. Cloud servers, being centralised and memory‐intensive, may result in increased computational delay, lower throughput, and DoS attacks in NB‐IoT. Hence, in this paper, we implement fog computing, a decentralised technology, providing scalability, reduced bandwidth, and enhanced privacy, to provide distributed processing. Additionally, we develop secure handover protocols for private and service‐provider‐controlled fog networks under normal and cell‐splitting conditions to minimise fake base station attacks and provide seamless handover. We introduce reputation‐based mechanisms to determine device integrity and differentiate faulty behaviour and attacks. Further, we implement real‐time application‐aware resource allocation and QoS‐based load‐balancing using deep learning to distribute data processing between devices, fog and cloud servers. We simulate and prototype protocols on iFogSim2 and Raspberry Pi 4. Security of the fog computing framework is validated against various attacks and formally verified using Scyther. Evaluation shows that our approach consumes 12% and 43.75% lower power and communication overhead and approximately 6 and 16 times lower execution time and memory compared with existing solutions, thus making our approach lightweight.