Fault-tolerant scheduling of real-time tasks having software faults

This paper investigates the problem of fault-tolerant scheduling of a set of real-time tasks where each task has primary and alternate implementations. Similar scheduling problem has been studied before, however, we propose an enhanced scheme for scheduling real-time periodic tasks with software faults. Alternate-primary recovery (APR) based scheduling employs a special backup-primary that can replace the primary when it fails often. The new scheduling technique saves the CPU time wasted on executing of unsuccessful primaries again and again. APR scheduling is implemented and tested for a TRC (TCP-to-RS232 converter) embedded system that connects Ethernet to serial-RS232 devices. It is also compared with an existing fault-tolerant scheduling method to verify the proposed enhancement