Research On Real-time Task Scheduling In Multi-core System

All the time, the task scheduling problem in the operating systems is a hot issue that researchers have been concerned about. With the development direction of the processor, more and more experts and scholars have been put into the research on the task scheduling problem based on multi-core processor. In this paper, we mainly study the scheduling algorithm of real-time tasks based on heterogeneous multi-core processors. Because it has a crucial role in the task scheduling algorithm on the system performance. In this paper, we hope that the algorithm can be reasonably designed to find the scheduling sequence with shorter task scheduling length. While ensuring that each real-time task can be run successfully, it can reduce the time of task scheduling and improve the working efficiency of the systems.According to the reference, the real time task scheduling problem in heterogeneous multi-core systems is a NP(Non-Deterministic Polynomial) problem. And the artificial intelligence algorithm is usually used to obtain the approximate optimal solution for this kind of problem. Therefore, this paper chooses ant colony algorithm to solve the task scheduling problem. Aiming at the two problems, which are slow convergence and easy to fall into local optimum, this paper has made some improvements in several aspects. So that it can get the approximate optimal solution and speed up the convergence speed. On the premise of some basic assumptions, the model of heterogeneous multi-core system is established, which includes three parts: the task model, processor model and task scheduling model. The model considers the amount of communication between tasks, inter core communication bandwidth problems, provides close to the actual system environment for the implementation of the algorithm. At the same time, the improved ant colony algorithm based on the actual situation of heterogeneous multi-core system, make adjustments to the probability selection formula of choosing task and choosing processor. And two strategies are adopted to improve the update method of information. On the one hand, the maximum amount of information and the minimum amount of information on the path are set. On the other hand, according to the speed of convergence and the evolution of the algorithm in the implementation process, flexibility to adjust the value of single information increment Q and the information evaporation factor ρ, to enhance the search ability of the algorithm and speed of convergence.In order to test the improved ant colony algorithm, the improved ant colony algorithm is implemented in Microsoft Visual C++ 6.0. The improved ant colony algorithm is proved to be feasible and effective, and compared with the same type algorithm. The experimental results show that the improved ant colony algorithm can get a shorter time scheduling sequence. And the performance of the task scheduling length is better than the same type algorithm which compared with in the average task scheduling length and the optimal solution.