Title: Task allocation in organic computing systems: networks with reconfigurable helper units

Authors: Daniel Merkle; Martin Middendorf; Alexander Scheidler

Addresses: Department of Mathematics and Computer Science, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark ' Parallel Computing and Complex Systems Group, Department of Computer Science, University of Leipzig, Johannisgasse 26, 04103 Leipzig, Germany ' Institut de Recherches Interdisciplinaires et de, Développements en Intelligence Artificielle (IRIDIA), Université Libre de Bruxelles, Av F. D. Roosevelt 50, CP 194/6, 1050 Bruxelles, Belgium

Abstract: In this paper, computing systems that have no central control and consist of many (partially) autonomous components are studied. Two types of components are distinguished which are called workers and helpers respectively. The components are connected via a network and the helper components perform service tasks for the worker components. It is assumed that the helper components exhibit reconfigurable hardware to be able to execute different service tasks efficiently. The problem addressed in this paper is how to organise such a decentralised system in a way that requests of the workers are executed by suitable helpers and the total reconfiguration costs of the helpers are small. Several decentralised task allocation methods are proposed. One of them uses a combination of a fully decentralised dynamic clustering algorithm and a self-organised task allocation system. The clustering algorithm is used to classify the service requests that are sent through the network in order to give the helpers hints which requests are suitable to be executed by them. Simulations are done for static and dynamic scenarios to investigate the reconfiguration costs and the number of dropped requests, i.e., requests that could not be satisfied. The results show that the clustering-based system has a strong adaptive behaviour and that the decentralised clustering is able to reduce the reconfiguration costs significantly.

Keywords: organic computing; self-organised task allocation; adaptivity; reconfigurable hardware; networks; decentralised systems; dynamic clustering algorithms; simulation; reconfiguration costs; dropped requests.

DOI: 10.1504/IJAACS.2015.067695

International Journal of Autonomous and Adaptive Communications Systems, 2015 Vol.8 No.1, pp.60 - 80

Accepted: 15 Nov 2011
Published online: 21 Mar 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article