Wireless Channel Availability Forecasting with a Sparse Geolocation Spectrum Database by Penalty-Regularization Logistic Models

Abstract

Television uses electromagnetic waves that carry audio and video. The unused frequencies or channels in broadcasting services are referred to as television white spaces. The unused spectrum can be managed to provide internet access in coordination with surrounding TV channels to avoid interference. Different ways of dynamically managing spectrum management have been conceived, and geolocation databases are considered the better option. Geolocation databases, when updated and complete, are helpful when frequencies are dynamically shared. In real life, the spectrum availability for a secondary user lacks numerous information; hence, it is sparse. This paper forecasts wireless channel availability given a sparse geolocation spectrum database. A dynamic sparse forecasting model is proposed through logistic penalized regression. Results show that forecasting accuracy is mostly above 90% on average when sparsity penalty terms are incorporated into the model. Forecasting accuracy is improved when penalty terms are integrated into the logistic regression models to account for sparsity.

Keywords

• Channel availability
• Forecasting
• Geolocation database
• Penalized logistic regression
• TVWS