Don’t hack my memory: Towards efficient, ubiquitous memory protection

The most secure execution environment offered by hardware is encapsulated in the concept of a “Secure enclave”: a collection of techniques developed over the past 15-years, by seminal contributions in the security field, to safely encrypt program, memory, and execution.

This proposal focuses on memory protection that is becoming more critical as with the expansion of Non-Volatile memory technologies, a large amount of sensitive information is left exposed in mobile or Internet-of-Things devices, even when they are switched off.

Today, implementations of a small set of fundamental concepts for secure , protected memory are commercially available but it is widely recognized that they come at a great performance and energy cost (e.g., over 3× slowdowns).

For this reason, the applicability of memory protection is limited and reserved for specialized situations.This proposal aims to address this critical deficiency by exploring a range of architectural techniques to ameliorate the cost of protecting memory, with an ultimate goal of making memory protection accessible to a wider range of systems and applications.

To this end, this proposal aims to: i) reduce the cost of memory protection by innovating in security-aware, efficiency-enhancing architectural techniques and ii) explore alternative approaches to memory confidentiality and integrity, tailored for the embedded/IoT design space.