Strategyproof Wireless Spectrum Auctions with Interference

Abstract

Abstract-Wireless spectrum is a regulated resource, whose control and usage is regulated by government agencies. The allocation of spectrum to interested parties is usually conducted through auctions, and are an important source of income for these regulatory agencies. However, previous spectrum auction design fail to take into consideration the effect of interference, which can adversely affect the truthfulness of an auction. In this paper, we explicitly consider interference effects, and design truthful auctions for maximizing social welfare. Since the spectrum allocation problem is NP-Hard, we first show how to compute an approximate spectrum allocation scheme that is within a constant factor of the optimal solution, under certain simplifying assumptions on the interference graph. We then proceed to make this scheme strategyproof by tailoring a payment scheme based on the idea of minimum bids. A naive method to compute such a payment scheme requires O(n) iterations of the spectrum allocation algorithm, where n is the number of bidders. We show how to reduce the complexity to O(1) iterations instead. We conclude by discussing possible directions for future research in this area.

Designing Two-Dimensional Spectrum Auctions for Mobile Secondary Users

“…Our design of GR 2D has been inspired by two previous works: the classic Vickrey auction [7] and a recent greedy spectrum auction due to Gopinathan et al [4]. More specifically, we are inspired by and adapt ideas in the former for computing charges in outcome S, and the latter for outcome M. The Vickrey auction is a simple yet truthful mechanism suitable for allocating a single item to competing bidders.…”

“…Therefore, when the efficient outcome is S, the social welfare in its good equilibria would be v max > n i=1 o ivi , which is at least 1 5 optimal social welfare. Efficient outcome is M: Then suppose the efficient outcome is M. If the actual outcome is M as well, the agents in equilibria would bid truthfully as in Gopinathan's [4], and the social welfare is at least 1 5 of the optimal. If the outcome is S, then the winner would be max in equilibria; otherwise, max would be a loser and it would bid at least its true Svalue, in which others are bidding at least their true values as well, leading to max a negative utility.…”

“…The auction of Gopinathan et al achieves 5-approximate social welfare [4]. We apply the same technique of greedy channel allocation coupled with charging critical bids, to ensure truthfulness for outcome M in our 2D spectrum auction.…”

“…We say i and j interfere if both are assigned the same channel, and d(i, j) ≤ 2r. These interference constraints can be captured by a conflict graph G [4]. A channel allocation is feasible if no two agents interfere.…”

“…When there is only one channel to be sold by the PU in the region, we propose GR 2D , a 2D channel auction that refers to recent techniques due to Gopinathan et al [4] for channel allocation and payment calculation in outcome M, with the classic Vickrey auction mechanism [7] to determine the pricing rules. We design the rule for deciding whether the outcome should be S or M, and tailor the pricing rule to elicit good social welfare.…”