Spectrum pricing for excess bandwidth in radio networks

Radio networks are designed to guarantee quality of service (QoS). This QoS guarantee is offered for customers with time varying load characteristics. Since the peak to average load of a network is usually very high the network utilizes the maximum capacity only for a fraction of time. Radio spectrum is a scarce resource and radio network providers pay huge amounts of money for licenses. For this reason applications with high QoS requirements are usually very expensive. For a better subscriber satisfaction the price for QoS should be as low as possible. From an economic perspective (X. R. et al., 2002), (U. Savagoankar et al.) the excess bandwidth could be disposed of in a secondary market. This would mean additional profit and by lowering the price for primary users the user satisfaction will increase. Surely there will be an optimum pricing strategy to maximize the profit. As a matter of fact, the QoS which can be offered in the secondary market is lower than the QoS which is offered in the primary market. In This work we will model a network which offers high QoS services to primary users and low QoS services for secondary users. The primary user applications are not queued whereas the secondary user applications are queued. Our focus is on a spectrum band at a dedicated frequency which is shared using a TDMA access scheme. We will model the traffic as Poisson arrival processes. Since the achievable QoS depends on the queue state of the network a state dependent pricing scheme for services in the secondary market will be applied. An acceptance function will characterize the behavior of the users in the secondary market.