Design, fabrication, assembly, test, and debug of integrated circuits and systems have become distributed across the globe, raising major concerns about their security and trustworthiness. Further, increased complexity and functionality of system-on-chips (SoCs) has resulted in increased attack surfaces and vulnerabilities. Such systems are prevalent is many critical-mission infrastructures, in which they require long and secure lifecycle. In this talk, we will provide a high-level overview of the newly funded DARPA program called AISS, with focus on RTL level security assessment of SoC designs, SoC security verification, development of security IPs, and establishing security engines to provide an end-to-end security throughout SoC lifecycle.
Design and test are arguably the two most important phases to deliver any system. The recent DARPA AISS
Program aims to create an automated chip design flow that allows security to scale consistently with other
design metrics such as area, delay, power, testability, and manufacturability. Many of the hardware security
research and practice in the past several decades will be implemented and integrated into this platform.
Like building other systems, the assessment of such automated secure silicon implementation platform is
critical. In this talk, we will outline the goals and general approaches of the AISS IV&V team led by the
University of Maryland. Then we will discuss the challenges in assessing security and trust of hardware design.
The large attack surface of commodity operating systems motivated academia and industry to develop novel security architectures which provide strong protection for sensitive applications in enclaves that only require trust in the underlying hardware and minimal software components. However, the enclave architectures proposed by industry often lack important features, such as secure I/O, and assume threat models which do not cover important cross-layer attacks, such as microarchitectural attacks. Thus, recent works in academia have proposed a new line of enclave architectures with distinct features and more comprehensive threat models, many of which were developed on the open RISC-V architecture. In this paper, we present a brief overview of the RISC-V based enclave architectures proposed in academia, discuss their features, limitations and open challenges which we tackle in our current research using our security architecture CURE.