A Study on Forecasting of Available Spectrum Resources for Sharing using Envelope Extraction

In recent years, the demand for spectrum resources has been increasing, which has raised expectations for dynamic spectrum access (DSA). DSA is a technique for allowing secondary systems to use the same spectrum band at locations and times when the primary system is not used. In conventional DSA systems, the protection area has been set conservatively wide in anticipation of the worst case scenarios. Meanwhile, especially in the millimeter-wave band, which is expected to be used more in the future, it is greatly affected by the surrounding environment, such as the number of people and vehicles. As a result, the coverage of radio waves varies depending on environmental changes. Hence, by considering the temporal changes caused by the variations described above, it is possible to detect additional spectrum resources that will be newly available to the secondary system, allowing for more dynamic sharing. In this paper, we propose a method for predicting the received power of the primary system, which varies with environmental changes, using machine learning, and detecting future available spectrum resources. In order to reduce the False Detection Rate, which causes a risk of intersystem interference, while maintaining the Detection Rate of available spectrum resources, the proposed method performs time series prediction after extracting the envelope from the time series data of received power. The evaluation by simulation results showed that the proposed method can reduce the false detection rate of available resources by about 9%.