Noise Level Estimation in the Shortwave Frequency Range

The presence of noise is the fundamental principle of wireless communication which must be taken into account always when setting the parameters of radio system such as sensitivity, modulation type and coding, but also choosing location of the receiving site etc. Generally, electromagnetic noise is classified according to its source – atmospheric or man-made noise. It is common to apply the recommendation ITU-R P.372-9 for estimating the environmental character of background noise. The recommendation defines five typical environments in the context of RF (Radio Frequency) reception. At lower frequencies of short-wave band, the atmospheric noise predominates. The main causes of atmospheric noise are thunderstorms, occurring mostly in the tropical regions of the Earth. Electromagnetic noise, developed by these storms, uses much the same propagation mechanisms as the skywave. Temporal grouping of noises depends on daily changes in the ionosphere, time of the year, and solar activity. Total atmospheric noise level at a receiving site is in tight correlation with local weather conditions. For example the local thunderstorm may increase the noise levels by about 10 dB, compared to a silent period [1]. Level of a man-made noise is less dependent on the number of people, living in a certain area, than the technical sophistication of local infrastructure and lifestyle. Power supplies of some lighting equipment, starting systems for electrical motors, generators, different impulse power supplies and big computer farms contribute to the level of local noise environment. Because of this, the background noise levels in peak hours can raise substantially over the top level set by the standards [2]. As the infrastructure in Estonia has changed greatly within the last decade and the level of technological sophistication has increased significantly also in rural areas, it is appropriate to explore how those 5 different noise environments described in the recommendation ITUR P.372-9 can be empirically identified and which particular sources of noise they mainly depend on. The monitoring was carried out to evaluate short-wave radio communication sites with respect to the quality of reception and to get an overview of spectral occupancy within the frequency range from 1.5 to 15 MHz.


Introduction
The presence of noise is the fundamental principle of wireless communication which must be taken into account always when setting the parameters of radio system such as sensitivity, modulation type and coding, but also choosing location of the receiving site etc.Generally, electromagnetic noise is classified according to its sourceatmospheric or man-made noise.It is common to apply the recommendation ITU-R P.372-9 for estimating the environmental character of background noise.The recommendation defines five typical environments in the context of RF (Radio Frequency) reception.
At lower frequencies of short-wave band, the atmospheric noise predominates.The main causes of atmospheric noise are thunderstorms, occurring mostly in the tropical regions of the Earth.Electromagnetic noise, developed by these storms, uses much the same propagation mechanisms as the skywave.Temporal grouping of noises depends on daily changes in the ionosphere, time of the year, and solar activity.Total atmospheric noise level at a receiving site is in tight correlation with local weather conditions.For example the local thunderstorm may increase the noise levels by about 10 dB, compared to a silent period [1].
Level of a man-made noise is less dependent on the number of people, living in a certain area, than the technical sophistication of local infrastructure and lifestyle.Power supplies of some lighting equipment, starting systems for electrical motors, generators, different impulse power supplies and big computer farms contribute to the level of local noise environment.Because of this, the background noise levels in peak hours can raise substantially over the top level set by the standards [2].
As the infrastructure in Estonia has changed greatly within the last decade and the level of technological sophistication has increased significantly also in rural areas, it is appropriate to explore how those 5 different noise environments described in the recommendation ITU-R P.372-9 can be empirically identified and which particular sources of noise they mainly depend on.The monitoring was carried out to evaluate short-wave radio communication sites with respect to the quality of reception and to get an overview of spectral occupancy within the frequency range from 1.5 to 15 MHz.

Measurements
Since short-wave range is not in the focus of commercial interest, there is a lack of comprehensive data about spectrum usage and electromagnetic noise level in Estonia.In order to determine the character of background noise in the prospective shortwave reception sites and to gather data about spectral occupancy in general the current survey was initiated by the Estonian Defense Forces.
We present several samples of the measured data and give an outline of monitoring equipment in this paper, but the more detailed description of the research procedure has been proposed by the authors in [3].
Right choice of the antenna is critical to the measuring of interferences, because the directivity, polarization and gain characteristics affect significantly the results of the measurement [4].Also the stability of power supply has crucial importance.We conducted comparative measurements, using power from the battery to check the noise level induced by the local mains power.The monitoring receiver in use was Rohde & Schwarz ESMB with the active monopole (rod) antenna and the laptop computer equipped with the monitoring application ARGUS (Fig. 1).

Fig. 1. Layout of the monitoring system
The RF band was scanned using 5 kHz frequency step and with receiver bandwidth 4 kHz (Fig. 2).Measurement results were analyzed using software package MATLAB.External noise is among the most important factors determining the noise floor when estimating the signal-tonoise ratio in short-wave range.Certainly it is most advantageous to operate on short-wave channel with no interference at the distant end receiver in the optimum operating frequency, yet this is not commonly feasible.The useful radio signal has to compete with the disturbances in the radio channel at every given moment, considering theoretically unlimited number of noise and interference sources.In addition to the local sources of background noise and interference one also needs to take into account noises, originating thousands of kilometers away, as ionosphere is equally well providing propagation of signal, noise and diverse interferences.
ITU-R P.372-9 gives the common methodology for specification of noise electromagnetic pattern in four environmental categories plus galactic noise.While predicting the expected noise levels, the characteristic trends with frequency, time of day, season, and the geographical location are taken into account explicitly.There are other variations that could be considered only statistically.The Recommendation gives the prediction methodology for approximated calculations of background noise level on the assumption that interferences due to surplus co-channel transmissions and other sources of impulse noise in close range are not present [5,6].
The external noise figure F a which is defined by ITU-R P.372-9 in logarithmic notation for the frequency f applies to a short vertical antenna over a perfectly conducting ground plane.This parameter is related to rms noise field strength E n along the antenna by where E n is in dB (above 1 µV/m), frequency f is expressed in MHz, and the receiver bandwidth B -in dB (where B is in dB-Hz).
In a real communication environment the character of external noise power is highly impulsive and non-Gaussian, hence fitting of probabilistic distribution of the received random noise waveform is required.Nevertheless, for the long-term predictions it is more convenient to use the median level of man-made noise.For estimation of median values of man-made noise power for different environments and frequencies the following expression is given by: where frequency f is expressed in MHz and environmental constants c and d are listed in the Table 1.The curves illustrating the expected levels of background noise are shown on Fig. 3.These curves are idealized.The recommendation also gives table for decile values of average man-made noise power expressed in dB above or below the median.These values were measured in the 1970s and may change considerably with day-to-day, in order to the activities which may generate man-made noise.It shows that shortterm behaviour of the noise level can vary rather largely.

Comparison of the empirical data to the Recommendation
In real communication environment the value of F a changes stochastically, as both the development of thunderstorms and propagation conditions are changing randomly.Usually domestic appliances and their power supplies can cause the noise on low frequencies.The atmospheric noise predominates at frequencies below 10 MHz, but simultaneously we can find the man-made noise and interference pattern as well.This is the case illustrated on Fig. 4. Measurements were carried out on 28 th October 2009 during daytime between 12 and 13 UTC.
Fig. 5 and Fig. 6 are examples of background noise pattern at two rural sites and the associated contours plot were recorded on the 21 st and 25 th of October 2009.We refer to it as the quiet period of solar activity.These plots are shown to illustrate strong interferences and man-made noise regarding the location, the time-of-day and propagation conditions.The comparisons between theoretical level of predicted data and those measured in practice reveal considerable disparities [6].Although the noise dissemination may use either sky wave or ground wave methods the primary sources of noise are local ones.
Fig. 9 shows measurement results performed in an office at the University of Technology equipped with large PC farms.Yet the ITU-R P.372-9 does not cover the indoor noise levels.This example is present to view as a reference to get the idea of the EMC scenario.The PC emission dominates over the spectrum in question as it was expected.Similar measurements were made by Weinmann and Dostert [7].

Conclusions
The lower end of shortwave band is a very complicated communications environment with respect to noise and interference.There exist significant deviations from the expected background noise level, especially at lower frequencies from 1.5 to 4 MHz since the amplitude of man-made noise decreases with increasing frequency.The noise originates mainly from electric motors and ignition systems located in the close range of receiving antenna.The good grounding is also very important for shortwave communication.However, business sites with a number of interference sources such as computer farms and various communication systems produce equally very high levels of background noise within the whole shortwave spectrum.
Altogether the noise level on short waves could be characterized relatively well by using the data provided in ITU-R P.372-9.
There is a need for more detailed knowledge about the variations of interference levels.It is of crucial importance to monitor shortwave band over a longer period of time with averaging over multiple scans, to observe seasonal changes, solar activity etc for long-term channel assignment.

Fig. 3 .
Fig. 3. Curves of the expected background noise level for the frequency range from 1.5 to 15 MHz

Fig. 6
reveals broadband disturbances below 5 MHz originated from the combination of an engine-generator and bad grounding of the communication equipment.Fig. 7 and Fig. 8 present plots of the noise pattern in two residential areas.Spectrum was scanned on 21 st of October and 10 th of November 2009 between 1400 and 1500 UTC.The daily sunspot number from 11 to 23 during the monitoring.

Table 1 .
Values of environmental constants c and d