Hybrid Laplace-Gaussian Differential Privacy to Secure Data Aggregation in Edge-IoT Systems
Paritosh Kumar Yadav, Sudhakar Pandey, Prerna Singh, Parth Pandey · 2024
The conventional cloud-enabled Internet of Things (IoTs) framework has seen several new sensing applications and techniques introduced by the blooming Internet of Things (IoTs) devices. As a result, the new network architecture gradually turns bidirectional, allowing IoT devices to do light processing in addition to serving as data harvesters. By moving some processing back to the “local,” edge computing, a novel architecture that is on the horizon, dramatically reduces latency and reliance associated with standard cloud-enabled networks. But when edge computing returns data and models to IoT devices, new security threats appear. Big data is being produced by rapidly expanding smart city applications like smart delivery, smart communities, and smart health, and it’s being shared freely online. Systems for the Internet of Things (IoT) are central to smart city applications, as conventional cloud computing is unable to provide the essential needs of intelligent Internet of Things systems. Owing to the sensitive nature of smart city applications, several privacy-preserving techniques, including homomorphic encryption, differential learning, and anonymity, have been used over the years. In addition, the resource consumption for maintaining data privacy in edge computing environments—which have fewer resources than cloud data centers—has received little attention. Differential privacy (DP) in particular has proven to be a successful privacy-preserving technique in the context of edge computing. This paper focuses on implementing differential privacy in edge-enabled IoT systems to address these privacy concerns. Differential privacy (DP) in particular has proven to be a successful privacy-preserving technique in the context of edge computing.