An improved EDF scheduling algorithm based on fuzzy inference being suitable for embedded soft real-time systems in the uncertain environments

EDF is a classic dynamic embedded real-time multi-task scheduling algorithm. In an embedded soft real-time system, the deadline missing ratio is an important metric to evaluate system performance. When an embedded soft real-time system is overloaded, EDF algorithm is not effective. In addition, considering the unsteadiness and unpredictability of a practical task running environment due to the unsteadiness of network communication and the time estimation deviation, it is necessary to introduce fuzzy concept and theory to the scheduling field of embedded soft real-time application systems. In this paper, we proposed an improved fuzzy EDF scheduling model based on fuzzy inference which was more suitable for embedded soft real-time systems in an uncertain environment. In our scheduling model, all task are periodic and a task's criticality and deadline distance are described with fuzzy set. In our scheduling algorithm, a task's scheduling priority is gotten by looking up the inference rule table with its fuzzy deadline distance and fuzzy criticality patterns. Tasks with shorter fuzzy deadline distance and higher fuzzy criticality are scheduled first. The simulation test shows that our scheduling model has less deadline missing ratio than traditional EDF algorithm and the important tasks have less deadline missing ratio than that of others tasks in an overloaded uncertain embedded soft real-time system.