FAucS: An FCC Spectrum Auction Simulator for Autonomous Bidding Agents

摘要

We introduce FAucS, a software testbed for studying automated agent bidding strategies in simulated auctions, specifically the United States FCC wireless frequency spectrum auctions. In addition to the complexity of these auctions, which provides ample opportunities for intelligent approaches to bidding, this type of auction has huge commercial importance, each bringing in billions of dollars to governments around the world. We implement straightforward sample agents in FAucS and use them to replicate known beneficial bidding strategies in this type of auction. We then discuss potential in-depth studies of autonomous bidding agent behaviors using FAucS. The main contribution of this work is the implementation, description, and empirical validation of the FAucS testbed. We present it as a challenging and promising AI research domain.