Publication - Simplifying solar harvesting model-development in situated agents using pre-deployment learning and information sharing

Authors: Zafar, Huzaifa and Corkill, Daniel
Title: Simplifying solar harvesting model-development in situated agents using pre-deployment learning and information sharing
Abstract: Agents deployed in real-world applications typically need to determine various aspects of their local environments in order to make appropriate decisions. This must be done quickly, as performance can suffer until each agent develops a model of its environment. We use a strategy of factoring this model development (that would typically be done using multi-agent learning) into two phases: pre-deployment (site independent, individual agent) learning and post-deployment (site dependent, multi-agent) model completion. By performing as much learning as possible prior to deployment, we simplify what needs to be determined on-site by each agent. Furthermore, we use collective sharing of local-observation information, in conjunction with temporal and spatial constraints in relating information, to reduce the number of observations needed to perform each agents model-completion activities. In this paper, we apply this two-phase strategy in developing prediction models for solar energy to be harvested by each agent in a power-aware wireless sensor network. In all but the most unlikely of environmental conditions, this strategy allows individual-agent harvesting models to be completed using only the first and second days observations.
Keywords: Communication, Learning, Multi-Agent Systems
Publication: Proceedings of Second International Workshop on Agent Technology for Sensor Networks (ATSN-08), pp. 41 - 48
Location: Estoril, Portugal
Date: May 2008
Sources: PDF: http://www.cs.umass.edu/~hzafar/ATSN08.pdf
Notes: (A version of this paper is also available as a UMass Tech Report #08-07)
Reference: Zafar, Huzaifa and Corkill, Daniel. Simplifying solar harvesting model-development in situated agents using pre-deployment learning and information sharing. Proceedings of Second International Workshop on Agent Technology for Sensor Networks (ATSN-08), pp. 41-48. May 2008. (A version of this paper is also available as a UMass Tech Report #08-07)
bibtex:
@inproceedings{Zafar-447,
  author    = "Huzaifa Zafar and Daniel Corkill",
  title     = "{Simplifying solar harvesting model-development in
               situated agents using pre-deployment learning and
               information sharing}",
  booktitle = "Proceedings of Second International Workshop on
               Agent Technology for Sensor Networks (ATSN-08)",
  pages     = "41-48",
  month     = "May",
  year      = "2008",
  address   = "Estoril, Portugal",
  url       = "http://mas.cs.umass.edu/paper/447",
}