Fault-Tolerance in Real-Time Systems

Abstract

Abstract With ever-increasing reliance on digital computers in embedded systems such as in space, avionics, manufacturing, and life-support monitoring/control applications, the need for dependable systems that deliver services in a timely manner has become crucial. Embedded systems often interact with the external environment and operate under strict timeliness and reliability requirements. Fault tolerance and real-time requirements on a system often influence one another in subtle ways, for example, the requirements on a highly—available system, such as an air traffic control system, may derive the timing constraints imposed on certain critical tasks. In cases where a set of interacting tasks operate under strict timing constraints, missing a deadline may result in a catastrophic system failure, which was termed a dynamic failure in (Shin, Krishna, and Lee, 1985). This paper argues that fault- tolerance and real-time requirements are not orthogonal and it addresses some of the challenges that confront the designers of fault-tolerant real-time systems. These challenges include formal specification of reliability and timing requirements, appropriate language and operating system support for providing fault-tolerance in a time-critical system, new scheduling theories which consider multiple resources and fault-tolerance, tradeoff between time and space redundancy, and predictable redundancy management in the presence of faults.