A new wideband HF simulation system for testing HF radios

Construction of interference and noise model for high-frequency channels based on alpha-stable distribution

“…In 1997, John F. Mastrangelo et al proposed that the noise and interference model consists of four components: bandlimited Gaussian noise, narrowband interference (sinusoidal), man-made noise, and atmospheric noise [9].…”

“…The generalized t-distribution was used by Hall in 1966 to describe atmospheric noise [9] and defined a probability density function for the interval between burst noise pulses. This approach was adopted in a variety of improved models of ITS wide-band HF channel models [2].…”

“…On the other hand, there should be a lot of narrowband noise. The frequency and phase of narrowband noise are uniformly distributed, and the amplitude of narrowband noise can be represented by the Hall model [9], a noise model specifically designed to represent impulsive noise. However, to represent narrowband noise, it is not only necessary to know that the narrowband process in the frequency domain is impulsive noise but also the amplitude distribution of the noise in the frequency domain, so it is not enough to use the Hall model to represent narrowband noise, which was discussed in detail by J. J. Lemmon [10].…”

“…Therefore, in addition to Gaussian and narrowband noise, there must also be a third component in the HF channel noise model: impulsive noise. Of the 42 sets of noise records taken for the experiment [10], only one set was detected with impulsive noise, and the detected pulses obeyed a Hall distribution with different parameters in the time domain [9] for the amplitude distribution and a uniform distribution for the phase. The pulses appear randomly in time, and the test results show that the time interval between pulses is approximately 500 us, which indicates that the impulse noise is man-made noise, rather than atmospheric.…”