J. Oh, H. Bahn, C. Wu, K. Koh, and B.-R. Moon (Korea)
Scheduling, Real-time System, Dead line, Multiobjective Genetic Algorithm
A new method for real-time task scheduling in multiproc essor systems is presented. Its objectives are to minimize the number of processors required and the tardiness for each of the tasks. The minimization is carried out through a multiobjective genetic algorithm, because the problem has non-commensurable and competing objectives to be optimized. Experimental results showed that when com pared to five methods used previously, such as list sched uling algorithms and a specific genetic algorithm, the per formance of our algorithm was comparable or better for 175 out of 180 randomly generated task graphs.
Important Links:
Go Back
