Static Homogeneous Multiprocessor Task Graph Scheduling Using Ant Colony Optimization

Nowadays, the utilization of multiprocessor environments has been increased due to the increase in time complexity of application programs and decrease in hardware costs. In such architectures during the compilation step, each program is decomposed into the smaller and maybe dependent segments so-called tasks. Precedence constraints, required execution times of the tasks, and communication costs among them are modeled using a directed acyclic graph (DAG) named task-graph. All the tasks in the task-graph must be assigned to a predefined number of processors in such a way that the precedence constraints are preserved, and the program`s completion time is minimized, and this is an NP-hard problem from the time-complexity point of view. The results obtained by different approaches are dominated by two major factors; first, which order of tasks should be selected (sequence subproblem), and second, how the selected sequence should be assigned to the processors (assigning subproblem). In this paper, a hybrid proposed approach has been presented, in which two different artificial ant colonies cooperate to solve the multiprocessor task-scheduling problem; one colony to tackle the sequence subproblem, and another to cope with assigning subproblem. The utilization of background knowledge about the problem (different priority measurements of the tasks) has made the proposed approach very robust and efficient. 125 different task-graphs with various shape parameters such as size, communication-to-computation ratio and parallelism have been utilized for a comprehensive evaluation of the proposed approach, and the results show its superiority versus the other conventional methods from the performance point of view.