A variable neighbourhood search algorithm for the constrained task allocation problem

Abstract

A variable neighbourhood search algorithm that employs new neighbourhoods is proposed for solving a task allocation problem whose main characteristics are: (i) each task requires a certain amount of resources and each processor has a capacity constraint which limits the total resource of the tasks that are assigned to it; (ii) the cost of solution includes fixed costs when using processors, task assignment costs, and communication costs between tasks assigned to different processors. A computational study shows that the algorithm performs well in terms of time and solution quality relative to other local search procedures that have been proposed.

Keywords:

• task allocation problem
• variable neighbourhood search
• local search

Acknowledgements

The authors are grateful to James Bean and Atidel Hadj-Alouane for providing test data. The research by the second author was partly supported by INTAS, Project 03-51-5501.