Spectrum Entropy Prediction Assisted Channel Selection for Secondary Users

Abstract

Dealing with the spectrum bottleneck, the cognitive radio (CR) is envisioned to alleviate the unbalanced usage of spectrum and promote spectrum sharing. Essential to the CR node, channel usage information is required to ensure appropriate and efficient channel access. Spectrum prediction could help seek better access opportunities, thus improving the performance of the CR node. In this article, we first briefly review three types of spectrum prediction. Then we propose spectrum entropy to measure channel access stability which captures the temporal variation of channel utilization. We devise a hybrid model for the prediction of spectrum entropy. Our experiment with extensive real world spectrum data shows that the prediction of spectrum entropy can assist channel selection with improved channel available time for secondary users (SUs) in the urban area.