اسم الباحث :
Maged A. Alkhawlani Ibrahim Ahmed Al-Baltah
Abstract Scheduling is a critical process to accomplish real-time application tasks in distributed systems. In the literature, there are a lot of algorithms that attended real-time tasks scheduling in distributed system. Yet, further work need to be done to improve the performance of real-time applications in distributed systems environment. Therefore, this paper aims to propose a two-phase scheduling algorithm, to improve the scheduling of real-time applications. The proposed algorithm that named enhanced EDF(eEDF)will play an important role in tasks scheduling of real-time distributed systems, this algorithm must be managing tasks based on two factors, earliest deadline first and the size of message, this approach must be enhancing sorting for tasks inside the network. Keywords: eEDF, Real-time systems, Task scheduling, Distributed systems Distributed system is basically containing a set of tasks interrelated through a computing network environment. In real-time applications of distributed system, there are some challenges with respect to tasks scheduling in order to perform tasks without having complexity in terms of allocating tasks in the computing nodes and the processing of these tasks. According to [1] when comparing single processor and distributed processors, there is an immanent difference between the two types of systems in their scheduling. In addition, making a decision regarding which task should be executed is not straightforward in distributed systems, because processors should be selected beforehand to take responsibility of running the required task. Unfortunately, any suggested algorithm will certainly suffer from networking issues, such as synchronization, delay and other issues that impact on the performance of the system.[1]As a matter of fact, researchers have proposed a noticeable number of techniques with the goal of improving real-time scheduling, but it has not reach its maturity due to some restricted constraints which are imposed by the nature of real-time devices Real-time applications are not allowing preempting task by another; thus, we need to use non-preemptive mechanisms to these applications. One of the predominant benefits behind using non-preemptive mechanisms is that, they are efficient to reduce overhead caused by switching between tasks[3], while, Lombardi et al. proposed another study for non-preemptive scheduling for Robust scheduling of task graphs under execution time uncertainty, they addressed some issues in multitask applications [4] Earliest Deadline First (EDF) is a scheduling algorithm which is used in tasks sorting through deadlines; it gives a high priority to the tasks that have shortest deadlines. EDF is probably considered to be one of the best dynamic scheduling schemes in real-time systems[5]. From another point of view, [6] proposed a scheduling algorithm that aims at reducing the delay in the communications of the real-time systems. Earliest deadline first scheduler makes the processor more serviceable up to 100 percent and enhanced the dealing with overload issue. EDF alternatively give the capability for processor to achieve the tasks as a very well. [7] Nonetheless, TOPS is an approach that breaks scheduling mechanism into two phases, sorting phase and scheduling phase. The first phase includes three policies for sorting. The second phase contains two policies for scheduling and deciding the start time for the tasks in the distributed system. [8]. According to [8], TOPS’s policies and algorithms are SORT-LIFO that adapts the idea of stack structure, SORT-FIFO that depends the idea of queue structure, and SORT-EDF that relies on the idea of getting task with the earliest deadline is given the highest priority. The reminder of this paper is organized as follow. Section 2 presents a sufficient review with respect to the topic of this research. The propose algorithm is discussed in section 3. Finally, section 4 concludes this paper and highlighting some future work.
سنة النشر :
2017
ملخص البحث :
