Virtual Privacy-Preserving IoT Data Analytics Platform with Applications in Smart Metering
University of New Brunswick
Internet of Things (IoT) devices have become popular in consumer applications that can be remotely monitored using smartphones. The staggering volume of data that is generated by IoT devices can be harnessed to provide valuable insights. To facilitate this, specialised high-performance computing resources are required. However, integrating a third-party cloud server in this environment builds a complex, heterogeneous system with security risks. This thesis proposes an end-to-end privacy-preserving framework for Cloud-IoT data analytics. Specifically, an application of this framework is realised to build a private smart metering data analytics system. To ensure data privacy in transit, and in storage, a lightweight cryptographic scheme, ASCON is applied before sending data over an unsecured channel. The server performs certain analytical computations over encrypted data using the fully homomorphic encryption scheme, TFHE. To demonstrate this application, privacy-preserving versions of common statistical algorithms have been implemented with optimisations to accelerate homomorphic computations.