MERCURY: Trading Platforms for Dynamic Spectrum Distribution

We propose to use short-term dynamic spectrum auctions to distribute spectrum driven by dynamic user demand and willingness-to-pay. Our goal is to develop efficient, real-time (eBay like) and economic-robust spectrum auction and trading systems. The challenges include (1) how to clear auctions in real-time while addressing interference constraints, (2) how to achieve economic robustness and prevent market manipulations while addressing interference constraints, (3) how to maintain spectrum efficiency, all at the same time.

  • To Preempt or Not: Tackling Bid and Time-based Cheating in Online Spectrum Auctions: We extend our auction designs into online spectrum auctions where bidders can request and obtain spectrum on-the-fly. But these online operations allow selfish bidders to game the system by both rigging their bids and falsely reporting their arrival time. These time-based cheating is easy to perform and yet produces severe damage to auction outcomes. We propose Topaz, a truthful online spectrum auction that distributes spectrum on-the-fly while discouraging bidders from misreporting their bids or time report. More details can be found in our Infocom 2011 paper.

  • Breaking Bidder Collusion in Large-Scale Spectrum Auctions: As the number of auction participants grows, bidder collusion becomes a serious threat. The nature of complex interference constraints among bidders provides a fertile breeding ground for colluders, leading to significant damage in auction efficiency and revenue. We propose Athena, a new collusion-resistant auction design for large-scale spectrum auctions. Athena implements a soft collusion resistance, allowing the auctioneer to configure the tradeoff between the level of collusion resistance and the cost of achieving such level of resistance. Unlike existing solutions, Athena enables spectrum reuse and maintains provable revenue bound, and does so with polynomial-time complexity. More details can be found in our Mobihoc 2010 paper.

  • Strategy-Proof Spectrum Trading using Double Auctions: We recently developed TRUST, a spectrum trading algorithm where spectrum owners and users can trade spectrum dynamically on-the-fly. TRUST implements the trading using truthful double spectrum auctions while enabling spatial-reuse to improve utilization. Additional details can be found in our INFOCOM 2009 paper. 

  • eBay in the Sky -- Efficient and Truthful Spectrum Auctions:  Unlike goods (e.g. paintings and bonds) in conventional auctions, spectrum is reusable among bidders and must be allocated under complex combinatorial interference constraints. We show that existing truthful designs, when applied to spectrum auctions, either fail or become computationally intractable. Our MobiCom 08 paper proposed VERITAS, a generate framework to design efficient and truthful spectrum auctions. It is highly flexible and competitive, allowing the auctioneer to optimize its desired utility and providing bidderswith diverse bidding and demand formats. Our paper was among the 4 finalists of the Best Paper Award at MobiCom. 

  • Real-time Auction Clearing: Our work in DySPAN07 and Com-Networks presented a fast auction clearing mechanism that can process 4000 requests in 90 seconds using a standard desktop. Our work can address both uniform and non-uniform pricing, and different bidding mechanisms. We show that non-uniform (often referred to as descriminatory) pricing mechanism is the key to revenue-maximizing and efficient spectrum auctions. Our paper received the Best Student Paper Award in DySPAN07.

  • Traffic-aware Spectrum Auctions: We examine the performance of on-demand spectrum auctions as well as the impact of bidding behaviors  using measured AP traffic traces. This preliminary work verifies the benefits of short-term spectrum auctions (which match spectrum allocation to dynamic traffic demand). It also shows that bids should depend heavily on traffic behaviors. See our WSDR08 paper for details.