Parallel demand side auction mechanism for dynamic and efficient resource allocation

摘要

Effective allocation mechanisms are required for infinitesimally divisible demand side resources capturing self-interest and non-cooperation among the users who consume the resources, in order to achieve efficient resource allocation maximising the social welfare with decentralisation of information sharing and decision making. The authors study an auction based mechanism to allocate the demand side resource, where a set of users who consume resources game with each other by bidding. The proposed auction mechanism fits into the progressive second price (PSP) mechanism, which follows a Vickrey-Clarke-Groves type payment. These kind of mechanisms admit incentive compatibility and existence of the efficient Nash equilibrium (NE). Furthermore, considering the fairness and privacy of the bidders as well as limitation of sequential algorithms in practical applications, the authors present a parallel updating algorithm in which all the bidders synchronously update their bids to achieve the efficient NE. By rigorous theoretical analyses, the proposed algorithm is guaranteed to converge to a bid profile which is shown to be the efficient NE. Numerical simulations demonstrate the performance of the proposed auction mechanism and parallel algorithm.