Analysis on the Performance of Spectrum Prediction in HF Communications

Abstract

Automatic Linking Established (ALE) is a very important process in HF communications. An important indicator to measure the performance of ALE is the link establishing time. As a newer technology, spectrum prediction can be applied to optimize channel scheduling thus reducing the link establishing time. In this article, we will discuss to what degree the link establishing time of HF ALE can be reduced by introducing spectrum prediction. Firstly, based on the existing traditional thresholds, a more reasonable GMM-HMM threshold is proposed, which can more accurately classify the idle or occupied spectrum state of channel. Secondly, we use the entropy theory to analyze the spectrum state sequences, proving that the spectrum state of the HF channel is predictable and calculating the upper and lower bounds of spectrum prediction accuracy. In addition, a simple prediction algorithm, linear prediction algorithm, is introduced as a comparison and supplement to the prediction upper and lower bounds. Finally, we discuss the upper and lower bounds of the link establishing time reduced when spectrum prediction algorithms are applied to HF ALE.