Auction is process of selling goods or services through bids. Auction enables a seller to know the actual cost of the service or goods. In this talk, we consider the problem of auction design for simultaneous ascending auctions (SAA) and second price auctions (SPA). We begin with characterizing the average revenue as a function of price increment in the context of SAA. When all the bidders are independent identically distributed, we conclude that choosing the optimal price increment maximizes the revenue but only slightly. Next, we consider SPA under a more general collusive setting, where although each bidder has independent identically distributed private values, bidders collude to form rings. A regret minimization algorithm is proposed that, by taking collusion into account, yields higher revenue as compared to the existing framework where collusion is not modeled. Various strategies are proposed to estimate the parameters of the collusion and the performance of the proposed method is analyzed under all these strategies. This is a joint work with Arpit Chitransh and Ketan Rajawat
Adrish Banerjee received his Bachelors degree from Indian Institute of Technology, Kharagpur and Masters and Ph.D. degree from University of Notre Dame, Indiana. He is currently an Professor in the Department of Electrical Engineering at Indian Institute of Technology, Kanpur. He is a recipient of Microsoft Research India young faculty award, Institute of Engineers India young engineer award, and IETE-Prof. Sreenivasan Memorial Award - 2016. His research interests are in the physical layer aspects of wireless communications, particularly error control coding, cognitive radio and green communications.