Exploration of security threats in In-Memory Computing Paradigms
In-Memory Computing, Cryptography, Hardware Security
Security is critical today for information and communication technologies. It is the basis for obtaining confidentiality, authentication, and integrity of data. Improving the attack resilience of secure devices is a major challenge today due, in part, to the accelerated race among the developers and the attackers, and also to the heterogeneity of new systems and their ever-increasing numbers. The vulnerabilities of electronic devices that implement cryptography functions have been well studied in the last decade for von Neumann computing architectures, designed with CMOS technology. However, there is little evidence that these studies hold true for novel computing paradigms with new technologies. 

In-memory computing paradigm is an emerging concept based on the tight integration of traditionally separated memory elements and combinational circuitry. It allows the minimization of the time and the energy needed to move data across the processor. The most promising solutions for in-memory computing architectures are based on the use of emerging technologies (Spin Transfer Torque RAM and Redox RAM) that are able to act as both storage and information processing unit. Despite the promising nature of the in-memory computing-based architectures many issues related to the devices themselves and to their double usage (storage and computing unit) need still to be solved. In particular, a correct evaluation of security threats targeting systems based on in-memory computing is still missing.

The innovative aspects of this PhD are related to the security aspects to be considered for a In Memory computing implementation. Indeed, the tight coupled memory-computing elements vulnerabilities have to be analyzed in hybrid NV-CMOS technologies. This research direction is very ambitious and can lead to consequent research direction.
Mis à jour le 8 February 2022