Learning from House-Hunting Ants: Collective Decision-Making in Organic Computing Systems
This paper proposes ant-inspired strategies for self-organized and decentralized collective decision-making in computing systems which employ reconfigurable units. The particular principles used for the design of these strategies are inspired by the house-hunting of the ant Temnothorax albipennis. The considered computing system consists of two types of units: so-called worker units that are able to execute jobs that come into the system, and scout units that are additionally responsible for the reconfiguration process of all units. The ant-inspired strategies are analyzed experimentally and are compared to a non-adaptive reference strategy. It is shown that the ant-inspired strategies lead to a collective decentralized decision process through which the units are able to find good configurations that lead to a high system throughput even in complex configuration spaces.
KeywordsNest Site Task Allocation Work Unit Supplementary Online Material Adaptive Model
Unable to display preview. Download preview PDF.
- 1.IBM Research: Autonomic computing: IBM’s perspective on the state of information technology. Technical report, IBM Research (2001)Google Scholar
- 2.Branke, J., Mnif, M., Müller-Schloer, C., Prothmann, H., Richter, U., Rochner, F., Schmeck, H.: Organic computing – addressing complexity by controlled self-organization. In: ISoLA 2006 (2006)Google Scholar
- 3.Würtz, R.P. (ed.): Organic Computing. Springer, Heidelberg (2008)Google Scholar
- 4.Compton, K., Hauck, S.: Configurable computing: A survey of systems and software. Technical report, Northwestern University, Department of ECE (1999)Google Scholar
- 10.Parker, C.A.C., Zhang, H.: Collective decision making: A biologically inspired approach to making up all of your minds. In: IEEE Int. Conf. on Robotics and Biomimetics, pp. 250–255 (2004)Google Scholar
- 16.Merkle, D., Middendorf, M., Scheidler, A.: Self-organized task allocation for computing systems with reconfigurable components. In: IPDPS 2006, pp. 25–29 (2006)Google Scholar