Real-Time Application Processing for FPGA-Based Resilient Embedded Systems in Harsh Environments

摘要

Real-time embedded systems nowadays get employed in harsh environments such as space, nuclear sites to carry out critical operations. Along with the traditional software based (CPU) execution, FPGAs are now also emerging as a bright prospect to accomplish such routines. However, these platforms are often get plagued by faults generated due to the high radiations in such environments. As a result, the real-time applications running on the platform could also get jeopardized. Thus, efficient execution of a set of hard real-time applications on reconfigurable systems with anomaly detection and recovery mechanism is inevitable. This work aims at tackling such problem with a “healing” approach for extreme environments. Initially, the applications are intelligently partitioned for hardware and software execution, then attempts have been made to schedule hardware applications with intermittent preemption point. Upon detecting any abnormality on such distinct points, our approach orchestrates a healing mechanism to remediate the scenario without hampering the pre-determined schedule. Experimental validation of our proposed method reveals its effectiveness.