Despite the extraordinary development of mobile communications, HF radio links have several advantages that make them unique in some application areas. Knowledge of HF noise is critical for the setup of a point to point HF communication. The paper presents HF noise measurements performed using a broadband antenna and Software Defined Radio equipment for signal acquisition, followed by offline signal processing. The recording has been split into several sub-bands of small bandwidth and for each sub-band the mean power has been computed, to find the sub-band with the lowest power. It might be assumed that the frequency band having the lowest power contains only noise. An analysis of the influence of sub-band size on the noise level hasalso been made. Measurements cover a large part of HF band and a large time interval to observe the noise behaviour.The paper analyses the noise level over short and long time periods, and over a large frequency band, and reports also simultaneous noise measurements in two locations in the western part of Romania, a dense urban one and a small town one, to highlight the differences caused by man-made noise of industrial type.
Even with the present-day development of mobile communications in UHF and SHF spectral bands and recent trends to move to mmWave, HF radio communications still offer advantages that make them exclusive in some application areas. Cognitive Radio principles can be applied to HF communications, to make use of the spectrum more efficiently. A significant improvement of Cognitive Radio technique can be achieved by spectrum occupancy prediction, which enables proactive efficient spectrum utilization. Hidden Markov Models (HMM) can be a useful tool for statistical spectrum occupancy prediction. In some of our previous works, we investigated a simple HMM in time domain, as well as a bidimensional model, that take into account both the time domain and the frequency domain. In this paper, we propose to extend the coverage of the model, by considering in addition to time and frequency the geographical position of the two points involved in the radio connection. The work is motivated by the fact that the two points of the radio link in HF can be far apart, and the channel availability at the two points could be different. Therefore, finding and exploiting correlations between the spectrum occupancy at the two points might improve the connectivity. The proposed model is validated using real spectrum occupancy data collected simultaneously at two locations situated in two cities, namely Timișoara and Sibiu, Romania. The measurements have been made using two identical USRP equipment, GPS coordinated, and identical HF broadband antennas.