Process scheduling using ant colony optimization techniques*

Author(s)

Nery, Bruno Rodrigues ; de Mello, Rodrigo Fernandes ; de Carvalho, Andre Carlos Ponce de Leon Ferreira ; Yang, Laurence Tianruo ; Guo, M ; Yang, LT ; DiMartino, B ; Zima, HP ; Dongarra, J ; Tang, F

• Document type: Journal article
• Source: Lecture Notes in Computer Science; v. 4330, p. 3-pg., 2006-01-01.

Abstract

The growing availability of low cost microprocessors and the evolution of computing networks have enabled the construction of sophisticated distributed systems. The computing capacity of these systems motivated the adoption of clusters to build high performance solutions. The improvement of the process scheduling over clusters originated several proposals of scheduling and load balancing algorithms. These proposals have motivated this work, which defines, evaluates and implements a new load balancing algorithm for heterogeneous capacity clusters. This algorithm, named Ant Scheduler, uses concepts of ant colonies for the development of optimization solutions. Experimental results obtained in the comparison of Ant Scheduler with other approaches investigated in the literature show its ability to minimize process mean response times, improving the performance. (AU)