@article{article_1225077, title={TVWS Geolocation Database for Secondary-User TVWS Devices for Spectrum Forecasting}, journal={The Eurasia Proceedings of Science Technology Engineering and Mathematics}, volume={21}, pages={188–195}, year={2022}, DOI={10.55549/epstem.1225077}, author={Pakzad, Armie and Manuel, Raine Mattheus and Materum, Lawrence}, keywords={Reinforcement learning, Television white space database, TVWS spectrum forecasting, Artificial intelligence}, abstract={This paper suggests a MATLAB-based television white space (TVWS) database with forecasting
capabilities. The availability of multiple TV frequencies for a secondary user, depending on the day and time of
the inquiry, location, and device, was forecasted using reinforcement learning (RL) software. A MATLAB live
script RL application was created and tested. After passing numerous tests, the algorithm was integrated into a
MATLAB App Designer application so that an SU could request projections for spectrum availability. Three
categories might be used to categorize the forecast—medium-term (MT), long-term (LT), and short-term (LT).
The forecast for that day is referred to as ST, the following day as MT, and the days after that as LT. Based on
the SU’s device, the program was designed to predict the spectrum availability in relation to the SU’s query time,
location, and transmission details. The SU device is a fixed white space device with an EIRP power cap of 4
watts. The database and AI are both necessary for accurate predictions. The forecast’s results indicated 100%
accuracy, assuming the database is frequently updated. Both the forecasting and reinforcement learning
programs were created using MATLAB’s App Designer.}, publisher={ISRES Publishing}