Controlling Real-Time Tasks Schedule Using the Value Parameter

Modern real-time applications are very dynamic and cannot cope with using worst-case workload. Therefore scheduling algorithms able to deal with overload situations are required. In this context the Value-Based Scheduling theory has become useful to add flexibility to such systems. This paper presents a comparative study among different real-time schedulers during overload, including traditional algorithms and the proposed DMB (Dynamic Misses Based). The main goal is to define the most suitable algorithm to be used with TAFT (Time-Aware Fault-Tolerant) scheduler in order to achieve a predictable system behavior. Obtained results show that DMB reached the most promising results because of its ability to control tasks degradation in a determined way.