Testing BOI and BOB Algorithms for Solving the Winner Determination Problem in Radio Spectrum Auctions

Abstract

Combinatorial auctions are a promising auction format for allocating radio spectrum, as well as other goods. An important handicap of combinatorial auctions is determining the winner bids among many options, that is, solving the winner determination problem (WDP). This paper tackles this computational problem using two approaches in a combinatorial first-price sealed bid auction. The first one, is an A* based on items (BOI). The second one, is an A* based on bids (BOB). These two techniques are tested in several scenarios for allocating radio spectrum licenses. The results obtained reveal that the search algorithm A* with the BOB formulation outperforms the other and always finds the optimal solution very quickly.