Regulation of the buyers' distribution in management systems based on simultaneous auctions and intelligent agents

Abstract

Abstract In solutions based on simultaneous auctions, participants are enabled to conduct and clear their own auctions, so that many auctions can be running at the same time. This configuration is increasingly being proposed for the installation of distributed management systems based on economic criteria in contexts such as the Smart Grid, computational grids and the cloud. In particular, these solutions are commonly complemented by the presence of software agents which automate the users' participation in an intelligent manner. This work focuses on the effect that the distribution of the bidders' participation actually has on the effectiveness of parallel auctions as management system. Firstly, the problem is introduced, showing that, in practice, bidders have incentives to prefer some auctions over others, thus tending to concentrate their participation in a limited subset of auctions. Also, as part of the theoretical formulation, with the aim of preserving the essence of markets and keeping competition active, it establishes a set of players' essential rights that any solution to the problem should preserve. The text then continues with the design and implementation of a regulatory mechanism that uniformly distributes buyers' participation among the available auctions. Realistic experiments are provided in order to demonstrate both the negative effects that buyers' concentration actually causes, and the effectiveness of the new regulatory mechanism, which manages to strengthen parallel auctions as a distributed management system. Likewise, it is proven that the solution is scalable, reactive and suitable for large distributed environments.