Publication - Distributed Problem Solving Using Iterative Refinement

Authors: Brooks, R.S. and Lesser, V.R
Title: Distributed Problem Solving Using Iterative Refinement
Abstract: The research presented in this paper concerns the development of a distributed iterative refinement algorithms for network traffic light control. Many distributed problem solving applications, including distributed network traffic light control, exhibit a natural spatial distribution of sensors an/or effectors. In order for these types of applications to be distributed effectively, processing should reflect the natural spatial distribution. Often this will require processors to operate on local data-bases that are incomplete and possibly inconsistent because the cost of maintaining complete and consistent data-bases can be prohibitive. The iterative refinement technique appears well suited for this type of distributed problem solving because it supports the desired problem decomposition and can operate on incomplete, local data bases. Two general classes of iterative refinement algorithms are examined: single-label and multi-label. Members of the first class of algorithms resemble classical "hill-climbing" algorithms; member of the second class of algorithms resemble "relaxation" algorithms which are often used in image processing applications. The various algorithms developed are described, results of experiments with the algorithms are presented, and error and uncertainty in the algorithms is discussed.
Publication: University of Massachusetts/Amherst Computer and Information Science Department Technical Report 79-14
Date: 1979
Sources: Other: http://www.cs.umass.edu/publication/docs/1979/UM-CS-1979-013.pdf
Reference: Brooks, R.S. and Lesser, V.R. Distributed Problem Solving Using Iterative Refinement. University of Massachusetts/Amherst Computer and Information Science Department Technical Report 79-14. 1979.
bibtex:
@techreport{Brooks-320,
  author    = "R.S. Brooks and V.R Lesser",
  title     = "{Distributed Problem Solving Using Iterative
               Refinement}",
  year      = "1979",
  url       = "http://mas.cs.umass.edu/paper/320",
}