Securing the “brains” of countless systems — from MP3 players and traffic lights to plant floor process controllers — is the aim of Dr. Yunsi Fei’s recent National Science Foundation Early Career (CAREER) Development award. An assistant professor of Electrical & Computer Engineering, Dr. Fei will apply her five-year, $405,000 award to her research aimed at developing superior computer architectures and designs for secure processing in embedded systems.
She explained that embedded systems are designed to perform one or more specific tasks, not the full range of tasks performed, for example, by a personal computer. While many embedded systems carry critical information and perform vital functions in communication, commerce, industrial operations and entertainment, she said they often operate in an insecure environment prone to tampering.
Dr. Fei explained that her CAREER research will focus on a common computing model for embedded systems that emulates secure microprocessor and insecure memory and peripherals. She said that as data flows between the microprocessor and the system memory, it can become compromised in the memory sector and trigger security attacks during processing within the microprocessor.
Existing safeguards hamper the efficiency and effectiveness of embedded systems, according to Dr. Fei. “Software approaches introduce significant execution time overhead, making them too expensive for resource-constrained embedded systems. Hardware countermeasures are normally customized, losing flexibility in security policy enforcement and often are circumvented by evolving security threats.”
Dr. Fei and her collaborators aim to enhance security at a fundamental level — the processor architecture for run-time secure processing. “The architectural support will significantly strengthen the inherent system design and implementation process,” she explained, thereby providing effective security measures. “Since it can be made transparent to upper-level software, it may not be easily circumvented by new software security attacks, and requires little recompilation of legacy code. Meanwhile, a hardware-oriented mechanism tightly coupled with processors for security enhancement will be much more efficient than pure software implementations.”
Another facet of the work will involve the up-scaling of architectural support from a single-core environment to more sophisticated and increasingly popular multi-core platforms. Dr. Fei anticipates the development of a suite of CAD design tools that may be distributed free of charge to embedded system designers so that individuals may secure the accurate and reliable functioning of their devices.
Dr. Fei earned her M.A. and Ph.D. degrees in electrical engineering from Princeton University in 2001 and 2004, respectively. Her research interests encompass embedded system and integrated circuit (IC) design automation, power analysis and optimization of ICs and systems, mobile computing systems, high performance and low power computer architecture, and hardware/software co-synthesis. She currently directs the Efficient and Secure Embedded Systems Laboratory and is a member of the UnderWater Sensor Network Laboratory.