Preventing Collusion in Cloud Computing Auctions

Shunit Agmon, M.Sc. Thesis Seminar
Thursday, 13.6.2019, 14:30
Taub 601
Prof. A. Schuster

In recent years, cloud providers have been moving towards offering their clients separate cloud resources for short periods of time instead of offering bundles of resources for longer periods. In parallel, the providers are moving towards using economic mechanisms, such as auctions, to allocate these resources. Vickrey-Clarke-Groves (VCG) auctions are likely to be used for that purpose. These auctions are incentive compatible: by allocating the resource to the highest bidders first, they maximize social welfare---the participants' aggregate valuation of the resources. However, VCG auctions are prone to various types of collusion, where users try to increase their profits at the expense of auction efficiency. We propose a coalition formation mechanism for cloud users that helps providers prevent user collusion. Our mechanism allows the auction participants to collaborate profitably while also maintaining the auction's resource allocation efficiency. By increasing users' profits, we reduce their incentive to collude in ways that harm the social welfare and allocation efficiency. We also propose a negotiation protocol for guests who wish to use the mechanism. Our experiments show that when using our mechanism, participants' mean profit increases up to the maximal possible collusion profit, without harming the provider's allocation efficiency. We also examine types of collusion mechanisms that involve changing the guest bids, and prove that when interactions are limited to two guests, all bid-altering interactions harm the social welfare.

Back to the index of events