Broadband Long-Term Spectrum Prediction Based on Trend Based SAX

With the development of communication technology and the growth of equipment, spectrum prediction technology has received more and more attention because of its wide application in spectrum resource management. However, due to the high burst of spectrum usage, there are still some difficulties in spectrum prediction. This paper proposes a new algorithmic framework (TrSAX-seq2seq) for the difficult problem of broadband and long-term prediction in the spectrum prediction problem. In this paper, a trend based symbolic aggregate approximation (TrSAX) method is used to reduce the dimension and represent the historical spectrum observation data, and then perform hierarchical clustering on the symbol sequence after dimension reduction to achieve the purpose of dividing the wide frequency band into multiple narrow frequency bands. Then, we use the LSTM network of seq2seq architecture to predict the spectrum occupation. We validate our method on a real spectrum monitoring dataset. The experimental results show that the method proposed in this paper can effectively improve the prediction accuracy compared with other methods.