Narrow Results

To many, high-frequency (HF) radio communications is obsolete in this age of long-distance satellite communications and undersea optical fiber. Yet despite this, the HF band is used by defense agencies for backup communications and spectrum surveillance, and is monitored by spectrum management organizations to enforce licensing. Such activity usually requires systems capable of locating distant transmitters, separating valid signals from interference and noise, and recognizing signal modulation. Our research targets the latter issue. The ultimate aim is to develop robust algorithms for automatic modulation recognition of real HF signals. By real, we mean signals propagating by multiple ionospheric modes with co-channel signals and non-Gaussian noise. However, many researchers adopt Gaussian noise for their modulation recognition algorithms for the sake of convenience at the cost of accuracy. Furthermore, literature describing the probability density function (PDF) of HF noise does not abound. So we describe a simple empirical technique, not found in the literature, that supports our work by showing that the probability density function (PDF) for HF noise is generally not Gaussian. In fact, the probability density function varies with the time of day, electro-magnetic environment, and state of the ionosphere.

The impetus for investigating the probability density function of high-frequency (HF) noise arises from the requirement for a better noise model for automatic modulation recognition techniques. Many current modulation recognition methods still assume Gaussian noise models for the transmission medium. For HF communications this can be an incorrect assumption. Whereas a previous investigation [1] focuses on the noise density function in an urban area of Adelaide Australia, this work studies the noise density function in a remote country location east of Adelaide near Swan Reach, South Australia. Here, the definition of HF noise is primarily of natural origins – it is therefore impulsive – and excludes man-made noise sources. A new method for measuring HF noise is introduced that is used over a 153 kHz bandwidth at various frequencies across the HF band. The method excises man-made signals and calculates the noise PDF from the residue. Indeed, the suitability of the Bi-Kappa distribution at modeling HF noise is found to be even more compelling than suggested by the results of the earlier investigation.