Enhancing transient fault tolerance in embedded systems through an OS task level redundancy approach

In numerous safety critical applications, the use of high-reliability or radiation-tolerant equipment may not be a viable option due to the presence of several constraints (such as cost) and the need to utilize Commercial off-the Shelf (COTS) equipment. However, such equipment may not meet reliability requirements, and therefore certain appropriate measures need to be taken to enhance their reliability. In this paper, a fully software-based method is presented to increase the reliability of COTS equipment against transient faults. The reliability of COTS is increased by utilizing a task-level redundancy in operating system. The proposed method is evaluated using a software fault injection method and a full system prototype. The experimental results show that the proposed method increases the fault coverage up to 99.34%. Moreover, the proposed method can be used in embedded systems without any hardware, software, or information redundancy.