A Framework for Dynamic Spectrum Sharing Between Cognitive Radios

We consider a cognitive radio system like the future 802.22 networks where license-exempt service providers (SPs) will share a fixed spectrum in a non-interference basis to each other and also to the licensed users in that spectrum. The percentage of spectrum utilized by one SP depends on how many users it is serving and how much spectrum each user application demands. We assume that an user can obtain service from all the SPs. The quality of service depends on system parameters such as number of users and SPs, the channel conditions between the users and SPs and the total power available at each user. We adopt an user utility maximization framework to analyze this system. Given the user utility functions, and the above mentioned system parameters we derive optimal values of spectrum that the users should obtain from the SPs. We also introduce the notion of spectrum price and use it to demonstrate several key results about spectrum allocation. The spectrum price proves to be the regulatory mechanism that brings about coordination amongst the SPs with minimal control messaging. Our approach thus strikes a balance between a centralized network and a fully uncoordinated open access network.