Modeling and Forecasting Secondary User Activity in Cognitive Radio Networks

Prior research efforts on cognitive radio networks have mainly focused on effective sensing of primary users to determine the availability of a spectrum band for opportunistic use. However, when multiple secondary users compete for a limited amount of spectrum, quality of service (QoS) for these users could degrade due to interference. In this paper, we present an integrated modeling and forecasting framework characterizing spectrum use by secondary users in a cognitive radio network (CRN). In our proposed method, each secondary user in a CRN, first enters a learning and modeling phase where it attempts to estimate the traffic parameters of other secondary users. Following this phase, the secondary user actively participates in opportunistic spectrum use. Assuming a continuous time Markov chain model for secondary user activity, we develop a Kalman filter based approach for estimating the number of secondary users. This estimate is in turn used to predict the number of secondary users in a future time instant. Experimental results show that our proposed forecasting technique provides a good upper bound prediction for the number of secondary users.