Investigating the direct derivation of device identity from low-level hardware devices

The security and privacy are one of the top priorities in the growing field of Internet of Things (IoT), as industries are expanding to incorporate wireless technology to generate valuable data. Adversaries attack the vulnerable node in the system to gain a wider access into the network. This thesis utilises the premise of ICMetrics technology to introduce a novel technique to secure low - level hardware used to build IoT applications. Previously, the technique has been used on off - the shelf IoT devices to generate keys and use it for authentication.

Building on the work of ICMetrics technology, this thesis aims to cover a research gap by using cost - effective, low - resource embedded devices like Raspberry Pi and the basis of the ICMetrics technology i.e., using internal characteristics of the devices and utilise it to identify the devices uniquely. The work performed in this thesis aims to explore the operational features (memory - based) from the devices to generate the device identity.

Based on the Literature Review, Raspberry Pi has served as a gateway for several large - scale attacks by using its vulnerabilities. Hence, we aim to exploit the uniqueness shown by these devices based on the working, environment where these devices are deployed and the application/processes running on the devices by exploiting dynamic operational features from them. The key contribution of this thesis is exploring features that are multimodal in nature and by utilizing the relationship between the modes, a novel classification technique is built which takes in account the multimodal nature of the features to gain higher accuracy in identifying devices.

Hence, making the widely used Raspberry Pi devices more secure and in case of any attacks, the characteristics of the devices change prevents the critical data leak from the devices & network.