On Fault Recovery of Firm Real-Time Computer Systems with Communication and Resource Requirements

Fault tolerance and fault recovery are integral parts of real-time systems. The literature addresses the issue of fault recovery via two main methods. One is hardware redundancy, and the other is achieved through task replication. Although lots of research has been done in this area, most of the work fell within one of the two streams, or as a combination of both. One point that did not have enough attention in the literature is real-time applications with both resource and communication requirements. Hou and Shin addressed fault recovery of systems with resource requirements, but considered resources as a natural extension to the application. Balaji et. al, considered resource requirements and showed that the overheads due to the non-compliance of resource constraints are alarmingly high. In this paper, a model for rm real-time systems is proposed; bridging the gap between the theory and the implementation. The model more accurately represents the current trend in the industry towards rm non life critical real-time systems. Various Quality of Service \QOS" levels are considered. A fault recovery scheme for rm real-time computer systems with resource and communication requirements is proposed. The proposed approach does not require any extra hardware, nor does it degrade the utilization of the system. The proposed approach should be fairly easy to incorporate into real-life applications without reconstructing the application, or adding new hardware.