Design Methods for trusted and secure embedded computing in SOC.

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

The distributed, wireless, mobile computing revolution moves information gathering and processing into the human environment. This has a profound impact on security. Traditional security applications, such as firewalls and VPN's, focus on protecting the communication channels between computers against attacks. This is done with security protocols and encryption algorithms running on the powerful processors of physically-protected servers. In an environment of small embedded distributed, wireless connected devices, this assumption is incorrect. The embedded device itself is vulnerable to attacks, and a hacker will select the method of attack that breaks the weakest link in an entire system including the embedded device as well as its communication channel. On top, the embedded device has limited computing and energy resources, and security is expensive (in terms of extra processing, memory, energy and development cost). Therefore, the embedded system is typically divided into secure and non-secure operations. The objective of embedded security is to thwart attacks at whatever abstraction level they happen. Security partitioning thus needs to address all design abstraction levels, software and hardware. The objective of this project is to develop design and validation methods that support a systematic security partitioning in a SoC (System-on-chip) integrated circuit design flow. Our design methods will help the embedded systems designer to cope with security constraints next to the existing constraints such as memory footprint, silicon area, performance, and power consumption.