Title: Using Shannon entropy as a systems' measure

Authors: Gerry Frizelle

Addresses: Institute for Manufacturing, University of Cambridge, 17 Charles Babbage Road, Cambridge, CB3 0FS, England, UK

Abstract: The paper pulls together research that the author and other researchers have carried out on using Shannon entropy as a measure for goal seeking input/output systems. Its aim is to show how the approach developed here can help to identify where and when systems start to behave unpredictably. Entropic measures provide some useful features, a key one being the ability to assign effects to causes. However, it is not possible simply to transfer Shannon|s results to a measurement system. Instead, one must first build a model of the measurement process. This led to a recognition of the importance queues play in revealing a system|s health; the more turbulent the queue, the bigger the problem. Moreover, sources of turbulence can be compared to pinpoint trouble spots. Shannon entropy seems to provide sensible and meaningful results when used to measure how far goal seeking input/output systems stray from their goals.

Keywords: Shannon entropy; system measurement; complexity; input output systems; goal seeking IO systems; turbulence; information.

DOI: 10.1504/IJIDS.2011.043024

International Journal of Information and Decision Sciences, 2011 Vol.3 No.4, pp.298 - 312

Published online: 30 Oct 2014 *

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