Binary Time Series Approach to Spectrum Prediction for Cognitive Radio

Abstract

One of the major goals of cognitive radio (CR) is to alleviate the inefficient use of the spectrum. CR can sense the spectrum steadily and gather information about the evolution of the spectrum in time. Spectrum occupancy information can be used for both learning the usage of the spectrum and predicting the future occupancy status. In this study, the use of binary time series for spectrum occupancy characterization and prediction is proposed. Both deterministic and non-deterministic occupancy schemes are examined. Numerical results and discussions are presented.