Cyber-physical systems (CPS) and Internet of Things (IoT) devices are increasingly used in our daily lives. Generally speaking, IoT refers to the connection of our everyday objects with a network like the internet. Each of these devices is usually equipped with different kind of sensors to observe its environment, making the device a smart object. In combination with embedded systems, IoT promises to increase the quality of our daily lives by taking over simple tasks like controlling the room temperature and cooking coffee (Nest Labs, 2016). On the other hand, smart objects like wearables are becoming more and more interesting for adversaries due to their increasing functionalities like internet capabilities, cameras, microphones, GPS trackers and other senor devices.
Furthermore, such smart objects are often deployed in unsupervised and untrusted environments raising the question about privacy and security to a crucial topic. Thus, a robust and secure software design is required for the implementation of cryptographic communication protocols and encryption algorithms. Moreover, tamper-proof solutions like secure elements and trusted platform modules are necessary to securely calculate cryptographic functions and to store confidential data or cryptographic keys. While cryptographic protocols and secure hardware architectures are well discussed and subject to further research activities, the issue of provisioning the initial confidential device setup is widely uncovered. However, the protection of this initial setup is as important as the protection of the confidential data during the time in use. Especially the protection of master keys is essential, because otherwise all security measures, which are based on such keys, are futile.
Due to the high quantity of produced chips – e.g. 8.8 billion secure elements for smartcard chips in 2014 (IHS Markit, 2014) – it is obvious that automatic methods are required to generate the trusted data needed for each chip. Otherwise an economical and practical production is infeasible. On the one hand, the system creating this data has to be designed flexible since every product can support different cryptographic protocols and thus, require different keys. On the other hand, the personalization system needs to fulfil high security requirements to prevent the risk that the generated data leaks during the production process to an operator – or even worse – to an adversary.
As revealed in 2015 by Edward Snowden, the secret master key of SIM cards, securing the 3G and 4G mobile communication channels was subject to such an attack (Begley & Scahill, 2015; Scahill, 2015):