A privacy-preserving combinatorial auction mechanism for spectrum redistribution

Abstract

The problem of dynamic spectrum redistribution has been extensively studied in recent years. Combinatorial auctions are believed to be among the most effective tools to allocate channels when bidders have diverse preferences on different bundles of channels. A great number of strategy-proof combinatorial auction mechanisms have been proposed to improve spectrum allocation efficiency by stimulating bidders to truthfully reveal their requested bundles and bidding values, which are bidders' private information. However, none of them has taken the bidders' privacy on both bidding values and bundles into account. In this paper, we consider the problem of privacy preservation on both bidders' bidding values and bundles, and present PICASSO, a Privacy-preservIng Combinatorial Auction mechaniSm for Spectrum redistributiOn. Specifically, PICASSO is designed based on BCP homomorphic cryptosystem by employing two non-colluding but semi-honest parties (an auctioneer and an auction issuer). Detailed analysis are given to demonstrate that PICASSO can guarantee strong protection for bidders' privacy on both bidding values and bundles without changing the traditional paradigm of communication in combinatorial spectrum auction. Furthermore, the results of comprehensive evaluations show that PICASSO achieves good performance in terms of spectrum redistribution with light overheads.