Market-Based Control of Complex Computational Systems

Overview

Market-Based Control of Complex Computational Systems was an EPSRC funded project that investigated the use of market-based methods for design and management of complex distributed computational systems. Examples of applications for such methods are: very large-scale demand-driven resource allocation such as Grid computing, decentralised content provision over peer-to-peer networks, and real-time co-ordination systems for commercial and military systems.

Mechanism Design

Although research into strategies of agents operating in such systems is an important part of the project, the ultimate aim is to investigate methods of automating the mechanism design process, which hitherto has relied on the hard work of human experts. Automated mechanism design's importance is also demonstrated by the fact that perfectly sensible human-designed auction mechanisms, such as the ''Continuous Double Auction'' turn out not to be optimal (D. Cliff, 2003).

Collaboration

The project was a collaboration of three leading British universities, and three high-technology companies, each bringing a diverse range of skills and expertise to the project. The University of Birmingham's principal strength lies in the area of evolutionary computation, which finds significant application in the area of market-based control.

Continuation

The EPSRC MBC project has now ended, but CERCIA's work on market-based control continues, for example in the EU funded EPiCS Project.