In this deliverable D5.1 “Security and Trust Models”, we discuss the hardware-supported FE schemes studied in WP5 of FENTEC and describe the security and trust models for the computing platforms executing computations required by the FE schemes. That is, in this deliverable, we solely focus on the implementation layer of the FE schemes and all discussion on efficiency as well as security and trust models are about the implementations rather than the theoretical schemes or algorithms per se. We define a generic computing platform modeling the large variety of computing platforms covered by FENTEC and its use cases. We employ this generic model to define adversarial, security and trust models that cover different types of adversaries as well as layers of security and trust in various possible deployments of FE.
We study multiple proposals for FE and provide high-level discussion about the critical aspects when they are deployed in actual implementations. The general computing platform also allows discussion of how to optimally map different parts of the FE schemes into the computing platform. For example, we discuss which parts of the FE schemes should be delegated to specific HW accelerators (co-processors) or trust anchors for improved performance or security, respectively. We also map the use cases studied in FENTEC to the models described in this deliverable to provide practical examples of how the models fit practice. Finally, we identify specific requirements for hardware development environments (development kits) to be used in implementation work in the three implementation tasks of WP5.