CORTEN: A Real-Time Accurate Indoor White Space Prediction Mechanism

Exploring and utilizing indoor white spaces (vacant VHF and UHF TV channels) have been recognized as an effective way to satisfy the rapid growth of the radio frequency (RF) demand. Although a few methods of exploring indoor white spaces have been proposed in recent years, they only focus on the exploration of the current indoor white spaces. However, due to the dynamic nature of the spectrum and the time delay in the process of exploration, users often cannot get accurate white space information in time, resulting in issues, such as spectrum utilization conflicts or inadequate white space utilization. To solve the problem, in this paper, we first perform an indoor TV spectrum measurement to study how the spectrum state changes over time and the spatio-temporal-spectral correlation of spectrum. Then, we propose a real-time aCcurate indoOR whiTe spacE predictioN mechanism, called CORTEN. CORTEN can predict the white space distribution for various time spans with high accuracy. Furthermore, we build a prototype of CORTEN and evaluate its performance based on the real-world measured data. The evaluation results show that CORTEN can predict accurately 38.7% more indoor white spaces with 51.3% less false alarms compared with the baseline approach when predicting the white spaces one hour ahead.