CPS: Medium:Embedded Fault Detection for Low-Cost, Safety-Critical Systems

Fault tolerance is vital to ensuring the integrity and availability of safety critical systems. Current solutions are based almost exclusively on physical redundancy at all levels of the design. A fault tolerant system must include the logic and algorithms for fault detection, fault diagnosis, fault containment, and reconfiguration to continue operation in face of failures. The standard approach to fault detection is to detect faults using a voting scheme on physically redundant measurements of the same signal. The the use of physical redundancy, however, dramatically increases system size, complexity, weight, and power consumption. Moreover, such systems are extremely expensive in terms of both the design and development, as well as the unit production costs.

Our main research focus will be on one specific—but imperative—sub-problem: the development of algorithms and computing architectures which enable the detection of faults without relying on physical redundancy. In this context, we aim to address both detection of faults in the physical domain (sensor, actuator, and environment faults) as well as faults in the logical—cyber—domain (software and processor faults).

The broader impact of the proposed project promises to have several positive influences on industrial practice and education