Browsing by Author "Safianowska, Maria Barbara"

Now showing 1 - 2 of 2

Combinatorial Recurrent Multi-Unit Auctions for Fog Services
(2016-01-01)
Safianowska, Maria Barbara
;
Gdowski, Robert
;
Huang, ChingYao
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
An emerging Fog Computing paradigm aims at reducing overall latency and end-to-end data transfer costs between IoT devices and analytic entity as in comparison to purely Cloud based solutions. In the near future, Fog services will be offered on virtual marketplaces by Fog Node owners. To facilitate the exchange in such volatile markets, we devise a recurrent, multi-unit, combinatorial auction with bidder drop and resource waste control and a two-step winner determination algorithm. Proposed solution extends on Participation Incentive Optimal Recurring Auction (PI-ORA) algorithm, which separately solves bidder drop problem for single-unit, non combinatorial auction, to include the case of multi-unit combinatorial auction which is necessary for Fog service markets. The results show stability and efficiency of the proposed algorithm.
Revenue Maximizing Auction for Perishable IoT Resources
(2016-01-01)
Safianowska, Maria Barbara
;
Gdowski, Robert
;
Huang, ChingYao
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
In near future autonomous things will trade services on IoT marketplaces. However, in the resulting recurrent combinatorial auctions, the bidder drop causes the market collapse in a low competition scenario. Adding fairness can prevent this, however the resulting revenue is not optimal. We show that revenue may be improved above fairness solution by alternating winners only within minimum set of strongest bidders. We introduce and compare two algorithms: Proportional Fair Auction and Revenue Maximizing Auction. The second algorithm performs the best in both high and low competition scenario, making it best suited to IoT when revenue maximization is a goal.