Security as an Important Ingredient in Neuromorphic Engineering

IEEE Computer Society Annual Symposium on VLSI, pp. 314–319

Abstract

Mimicking brain functionality has been challenging as today's emerging technologies, systems, and the exploitation of systems are inefficient to meet a brain's energy budget. On the other hand, security has become a design goal that sometimes results in a contrasting outcome for the other objectives such as energy, area, and performance. Incorporating security counter-measures can make attaining the desired design goals even more difficult. Therefore, selecting the right components for a neuromorphic platform meets the desired performance and energy goals while maintaining secrecy. Furthermore, the emergence of novel technologies adds another design point to be explored for neuromorphic platform engineering. Firstly, this work reviews the technology for next-generation neuromorphic systems from a security perspective. We discuss various attack vectors and countermeasures for neuromorphic platforms. Secondly, we discuss the applicability of emerging technologies for security primitive designs. The purpose is to identify the right technology adoption for secure neuromorphic platforms. Eventually, we discuss both the aspects of neuromorphic engineering, “secure-by-design” and “design-for-security,”

Authors 1

1. Farhad Merchant corresponding Aachen Institute for Communication Technologies and Embedded Systems

   RWTH Aachen University

   Affiliation as printed

   Institute for Communication Technologies and Embedded Systems, RWTH Aachen University,Germany

   Institute for Communication Technologies and Embedded Systems, RWTH Aachen University, Germany