Economic Allocation of Computation Time with Computation Markets
Nathaniel Rockwood Bogan
MIT Master's thesis, May 1994
Thesis Supervisor: Jon Doyle
Abstract
The use of economic principles in computer science is a comparatively recent development.
The application of economic concepts such as marginal utility, preferences, efficiency, and resource allocation have begun to be discussed at a theoretical level.
Several systems have been developed that attack problems of computer science using these principles.
The applications include use of idle resources in a network and approximating solutions of complex problems by transforming the problem into a general equilibrium framework.
This thesis uses market mechanisms more directly and addresses the problem of allocating processor time using a market in computation.
This allows for efficient use of processor time, and is especially useful in the case where many more useful tasks could be performed than time constraints may allow.
The unique problems associated with a market for a processor are discussed.
A protocol is proposed to rent the processor at a rate per-time-unit, determined by market forces.
A postscript version is available.
A related paper providing a motivation for this work is:
Jon Doyle, A reasoning economy for planning and replanning. Technical Papers of the ARPA Planning Initiative Workshop, February 1994.
Author: Michael Wellman (wellman@engin.umich.edu)
Last Updated: 5/31/94