Price-based Online Mechanisms for Settings with Uncertain Future Procurement Costs and Multi-unit Demand
Abstract
We examine the use of online mechanism design in settings where consumers have multi-unit demand, goods are procured and allocated over time, and future procurement costs are uncertain and only become known at the time of allocation. An important application with such characteristics is demand response, where electricity wholesale prices depend on overall demand and the availability of renewables. We formulate this as a mechanism design problem and focus specifically on the property that the mechanism does not revoke any allocated items. In this setting, we characterise a class of price-based mechanisms that guarantee dominant-strategy incentive compatibility, individual rationality, and no cancellation. We present three specific such mechanisms in this domain and evaluate them in an electric vehicle charging setting. By using extensive numerical simulations, we show that a mechanism based on the first-come first-served principle performs well in settings where future procurement costs can be estimated reliably or supply is very tight, while a responsive mechanism performs very well when the estimated procurement costs are highly uncertain and supply is not as tight. We moreover show that a well-defined pricebased mechanism can lead to high profits for the operator of the mechanism in many real-world situations.