Online market driven spectrum scheduling and auction

In this paper we study the online spectrum scheduling using a market driven approach. We assume that each new arrival spectrum request, when it arrives, requests for the exclusive usage of some channels for a certain time interval. The spectrum owner has to decide immediately whether to grant its exclusive usage or not. If it is granted, the secondary user will be charged a payment. We assume that existing running requests can be preempted with a penalty depending on its bid value, its requested time duration and the remaining unserved time. For various possible known information, we present efficient scheduling algorithms to schedule requests and prove that the competitive ratios of our methods are asymptotically optimum, i.e., within small constant factors of the optimum online methods. Our extensive simulations show that our algorithms perform almost optimum: most of our methods can get a total profit that is more than 80% of the optimum.