Features of choosing a measuring instrument for automated hardware and software complexes for studying PEMIN..
KONDRATIEV Andrey Valerianovich,
NAGORNY Sergey Ivanovich,
DONTSOV Vadim Vladimirovich, Candidate of Technical Sciences, Senior Researcher
LOBASHEV Aleksey Konstantinovich, Candidate of Technical Sciences, Associate Professor
Features of Selecting a Measuring Instrument for Automated Hardware and Software Complexes for PEMIN Research
The issues of support and practical certification of information technology objects (IT) for information security requirements occupy an important place in matters of ensuring state secrets. This type of work is regulated by regulatory documents of the FSTEC (State Technical Commission) of Russia, which prescribe that all IT intended for processing information constituting a state secret, as well as conducting secret negotiations, must be subject to mandatory certification for information security requirements. The regulatory documents define the procedural and technical details of the certification of objects. At the same time, an analysis of these documents shows, on the one hand, the volume of the regulatory and procedural issues set out in them, and on the other hand, the complex technical implementation of specific studies and measurements that must be performed in the process of certification.
For technical implementation of certification, it is necessary to have perfect knowledge of theoretical and practical issues in various fields of science — acoustics and vibroacoustics, spectral analysis, antenna-feeder devices, laws of radio wave propagation, regulatory legal acts and guidelines on the protection of computer equipment (CEE), metrology, etc. No less complex issues are the acquisition and competent use of expensive and complex modern technical means for certification. One of the pressing problems that was identified in the process of using complex measuring systems is compliance (or non-compliance) with the quality criteria (compliance with the declared requirements) and, ultimately, the correctness of the special studies (SI). It should be noted that the use of modern automated hardware and software systems (AHSS) in the certification of OI is a priority. Particular care and thoroughness are required when choosing such unique, expensive and technically complex products as the APAC, intended for conducting special studies of the SVT. At the same time, the choice of a measuring instrument — a measuring device, which largely determines the cost and consumer properties of the APAC — is of great importance for the operation of the APAC.
The study of this problem, from the point of view of the authors, should be started with a study of the relevant standards. It should be noted that the following main standards are currently in effect in the Russian Federation, regulating activities in terms of PEMIN research:
- GOST 29339-92 «Protection of information from leakage due to PEMIN during its processing by means of computing equipment»;
- GOST R50752-95 «Protection of information from leakage due to PEMIN during its processing by means of computing equipment. Testing methods»;
- GOST R50543-93 «Base load-bearing structures of computing equipment. Requirements for ensuring information protection and electromagnetic compatibility by means of shielding»;
- GOST R51319-99 «Electromagnetic compatibility of technical equipment. Instruments for measuring industrial radio interference»;
- GOST R51320-99 «Electromagnetic compatibility of technical equipment. Industrial radio interference»;
- RMEK 60950-2002 «Safety of information technology equipment»;
- RD 50-715-92 «Protection of information from leakage due to PEMIN during its processing by computer equipment».
Taking into account that some GOSTs contain information related to state secrets, general approaches to substantiating the methodology for selecting a measuring receiver, from the authors' point of view, can be presented taking into account the analysis of international standards, for example ANSI C 63.2-1980 and similar ones (Karl-Otto Muller «Procedures for Granting Licenses for the Operation of RF Devices, Radio and TV Receivers in Western Germany», Rohde & Schwarz, Germany, 1987), which determine the procedure for measuring the PEMIN of computing equipment.
From the consideration of the presented materials it follows that the most critical characteristics of the measuring equipment used for the study of PEMIN SVT is the maximum sensitivity of the device within the specified measurement limits. The sensitivity of the device is most affected by the magnitude of the device's own noise and antenna-feeder devices. The magnitude of the device's own noise is of a standardized nature and is indicated in the device's passport with a mandatory listing of the settings and parameters at which the measurements were made.
Thermal noise, which characterizes both radio receivers and spectrum analyzers, is usually understood as intrinsic noise (Christoph Rausher «Fundamentals of Spectrum Analysis», Rohde & Schwarz GmbH&Co. KG, Germany, 2002). From a strictly scientific point of view, thermal noise «according to Nyquist», i.e. the noise of a 50 Ohm resistor, is orders of magnitude lower than the noise of the measuring path, reduced to the receiver input. These are shot noises, class «1/f» noises, flicker noises, conversion noises and a number of other sources specific to active semiconductor structures. The overall result of all these noises can be considered as conditionally «white». Note that the determining factor is the noise of the preselector (if any) or the noise of the first mixer cascade (if there is no preselector). Due to intrinsic noise, the signal/noise ratio at the input of the device decreases. Therefore, the intrinsic noise is a measure of the sensitivity of the spectrum analyzer. It allows us to draw conclusions about the required minimum level of the input signal that the analyzer can detect. The intrinsic noise of a radio receiver can be taken into account in various ways, and it is usually expressed in terms of the noise figure or noise factor.
The dimensionless noise figure F of a four-terminal element is the quotient of the signal-to-noise ratio at the input of the four-terminal element and the signal-to-noise ratio at its output.
F = (S1/N1)/(S2/N2), (1)
where S1/N1 is the signal-to-noise ratio at the input of the circuit,
S2/N2 is the signal-to-noise ratio at the circuit output.
The noise factor (noise figure in decibels) is determined by the formula:
NF = 10 log F (2)
The overall noise figure Ftotal of the cascaded circuits shown in Fig. 1 is determined as follows:
where Fi is the noise figure of an individual cascade,
Gi is the gain of an individual cascade.
Fig. 1. Cascade connection of noisy circuits
For passive circuits with losses, such as cables or adjustable attenuators, the following relationship is valid:
F = 10a/10
and NF = a, (4)
where F is the noise factor of the circuit,
NF is the noise factor of the circuit,
a is the attenuation introduced by the circuit, dB.
Equation (3) shows that the noise figure of the first stage is fully accounted for in the overall noise figure of the cascade chain. The attenuator is located at the input of the spectrum analyzer and is a passive stage whose noise figure can be calculated using expression (4).
Therefore, the overall noise figure of the analyzer depends on the attenuator setting. An increase in attenuation by 10 dB, for example, results in an increase in the overall noise figure by 10 dB. Therefore, maximum sensitivity is achieved with the attenuator set to 0 dB (Fig. 2).
The sensitivity of spectrum analyzers is usually characterized by the average indicated noise level (AIL), a parameter that can be directly determined from the image on the spectrum analyzer display.
The noise produced by the input circuits of a radio receiver is essentially «white» noise, meaning that it does not contain any discrete components. The probability that the noise voltage falls within a certain range of values can be determined from a Gaussian distribution, so the designation «Gaussian noise» is often used.
The indicated noise corresponds to the noise voltage generated by the envelope detector. The corresponding noise power can be calculated by integrating the noise density in the receiver noise bandwidth, which is the noise bandwidth of all stages before the detector. In the case of spectrum analyzers, this bandwidth is determined by the IF filter noise bandwidth. Accordingly, the indicated noise depends on the resolution bandwidth setting.
Fig. 2. Indicated average noise level of the spectrum analyzer as a function of radio frequency attenuation
Since the spectral power density of «white» noise is constant within the noise bandwidth, the indicated average noise level can be calculated (if the noise factor of the analyzer and the noise bandwidth of the IF filter are known) as follows:
LISH = 10 log (kTBsh.IF /10-3W) + NFAC — 2.5 dB, (5)
where LISH is the indicated average noise level, dBm,
k is the Boltzmann constant, k = 1.38×10-23 W/Hz,
T is the ambient temperature in degrees Kelvin,
Bsh.IF — IF filter noise bandwidth,
NFAC — spectrum analyzer noise factor, dB,
-2.5 dB is the error in determining the noise by the sample detector and in averaging the logarithmic values of the level.
For an ambient temperature of 290 K, the indicated average noise level is determined by the formula:
LISH = -174 dBm (1 Hz) + 10 log (Bn.IF/Hz) dB + NFAC — 2.5 dB. (6)
The value -174 dBm (1 Hz) corresponds to the thermal noise power acting on the ohmic resistance in the 1 Hz band at an average temperature of 290 K. This is the level of intrinsic noise or the absolute minimum noise level at a given temperature.
The sample detector, commonly used for noise measurements with spectrum analyzers, determines the arithmetic mean of the noise. In the case of Gaussian noise, this is 1.05 dB below the RMS value (effective noise power). Because the results are averaged on a logarithmic scale by averaging over several responses, the indicated noise is reduced by another 1.45 dB. When calculating the indicated average noise level according to equation (6), all this is taken into account by subtracting 2.5 dB. This correction is only valid for Gaussian noise, which is taken as a model of thermal noise.
From equation (5), the following relationship can be derived to estimate the variation in the indicated noise level when changing the IF bandwidth setting from BIF1 to BIF2:
DLISH = 10 log (BW.IF2/BW.IF1), (7)
where Bш.IF1, Bш.IF2 are the noise bandwidths of the IF filter before and after tuning, Hz,
DLИСН is the variation of the indicated noise as a function of the variation of the IF bandwidth, dB.
If both IF filters have the same ratios between the 3 dB bandwidth and the noise bandwidth, then the difference in the indicated noise can be calculated from the 3 dB bandwidth. The following relationship holds:
DLИСН = 10 log (ВПЧ2/ВПЧ1), (8)
where ВПЧ1, ВПЧ2 are the 3 dB bandwidths of the IF filter before and after tuning, Hz.
Figure 3 illustrates the effect of different IF bandwidth values on the displayed noise. Due to different practical implementations of IF filters in a spectrum analyzer, the noise figure of the analyzer may also depend on the selected resolution bandwidth value.
The analyzer sensitivity limit can also be determined from the indicated average noise level. It is understood as the minimum level of the input harmonic signal that ensures an excess of the noise level by 3 dB on the analyzer indicator, and is called the minimum detectable signal. Since the spectrum analyzer displays the sum of the input signal and noise (S + N), this condition is met at an input signal level that corresponds to the effective noise level of the intrinsic thermal noise (S = N). In this case, the signal-to-noise ratio is determined by the formulas:
(S + N)/N = 2
and 10 log ((S + N)/N) = 3 dB, (9)
where N = 8212; corresponds to the indicated noise level when using a root-mean-square detector.
Fig. 3. Indicated average noise level for different resolution bands
Fig. 4. Typical values of the indicated noise level
spectrum analyzer (excerpt from full specification table)
The list of characteristic data (specification table) for the indicated average noise level should always include the resolution bandwidth and attenuator setting. Typical settings: RF attenuator — 0 dB, resolution bandwidth — narrowest.
The maximum sensitivity of the spectrum analyzer is achieved when the attenuator is set to 0 dB. It is very important that the first mixer of the analyzer is not overloaded with a high-level signal, acting even outside the frequency range of measurements.
To further reduce the indicated noise level, it is necessary to reduce the resolution bandwidth. A compromise must be found between low indicated noise and high measurement speed. For indicating input signals with a very low signal-to-noise ratio, it is useful to reduce the video bandwidth as well as the resolution bandwidth or to increase the sweep time when using an RMS detector. The response is smoothed out, so the input signal is indicated more clearly. This stabilizes the measured levels, which is necessary for obtaining an accurate, reproducible result.
For level measurements it is important to know the frequency dependence of the preamplifier gain. This gain value in decibels must be subtracted from the measured levels. Many spectrum analyzers offer the possibility of taking into account the frequency dependence of the gain by means of special conversion tables. The measured spectrum can thus be determined with the correct levels.
High sensitivity of a spectrum analyzer is extremely important for many applications where the resolution bandwidth is limited by standards. In these cases, reducing the indicated noise by narrowing the resolution bandwidth is not allowed. Sensitivity is also important for high measurement speeds. At low sensitivity, narrow IF filters are required to achieve sufficiently low indicated noise, which in turn increases the sweep time. Spectrum analyzers with a low noise figure allow the use of wide resolution bandwidths and, therefore, shorter sweep times.
Practice shows that the above-considered requirements for sensitivity and frequency selectivity imposed on equipment for PEMIN studies are met by a rather narrow range of measuring instruments. At present, for conducting PEMIN studies, it is permissible to use only such a set of equipment, the basis of which is a measuring receiver or spectrum analyzer with a set of corresponding measuring antennas. Some generalized characteristics of these devices are presented in Table 1.
From the comparison of the characteristics of these devices it is clear that selective microvoltmeters are generally suitable for measuring the intensity of weak electric and magnetic fields. At the same time, due to the instability of the «drift» of the characteristics and the high measurement error (on average more than 5 dB), they do not stand up to comparison with modern measuring receivers and spectrum analyzers. In addition, they do not provide the ability to observe the panorama of the signals being studied.
Measuring receivers meet the requirements for equipment for PEMIN research to the greatest extent. They ensure high measurement accuracy with relatively low labor costs (which, incidentally, also applies to spectrum analyzers). A significant part of measuring receivers (and spectrum analyzers) allows you to see the panorama of the frequency range being studied, analyze signals while simultaneously observing the results of their detection by various types of detectors. However, the price of measuring receivers is very high.
Spectrum analyzers are quite comparable in their functionality to measuring receivers. At the stage of PEMIN detection, they are sometimes even more convenient than receivers. Most spectrum analyzers presented on the Russian market do not have a preselector. At the same time, the price of a modern spectrum analyzer is 2 — 3 times lower than the price of a similar measuring receiver in terms of frequency range.
Table 1. Comparative characteristics of measuring devices used for PEMIN studies
| Characteristic | Selective microvoltmeters |