Distributed energy efficiency in future home environments |
| |
Authors: | Helmut Hlavacs Roman Weidlich Karin A. Hummel Amine M. Houyou Andreas Berl Hermann de Meer |
| |
Affiliation: | 1. Department of Distributed and Multimedia Systems, University of Vienna, Lenaug. 2/8, 1080, Vienna, Austria 2. Faculty of Computer Science and Mathematics, University of Passau, Innstr. 43, 94032, Passau, Germany
|
| |
Abstract: | In this paper, a new architecture for sharing resources among home environments is proposed. Our approach goes far beyond traditional systems for distributed virtualization, like PlanetLab or grid computing, as it relies on complete decentralization in a peer-to-peer (P2P) like manner and, above all, aims at energy efficiency. Energy metrics are defined, which have to be optimized by the system. The system itself uses virtualization to transparently move tasks from one home to another to optimally utilize the existing computing power. We present an overview of our proposed architecture, consisting of a middleware interconnecting computers and routers in possibly millions of homes using P2P techniques. For demonstrating the potential energy saving of distributed applications, we present an analytical model for sharing downloads, which is verified by discrete event simulation. The model represents an optimistic case without P2P overhead and fairness. The model allows to assess the upper limit of the saving potential. An enhanced version of the simulation model also shows the effect of fairness. The fairer the system gets, the less efficient it is. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|