Comparative analysis of methods and means of automation of measurements in special studies of technical means of electronic computing equipment on PEMIN..
Comparative analysis of methods and means of automation of measurements in special studies of technical means of electronic computing
techniques on PEMIN.
As is known, side electromagnetic radiation and interference (SEMI) from computing equipment is one of the possible channels of information leakage. Searching for and measuring SEMI of computing equipment manually is a labor-intensive and lengthy process. At the same time, a significant part of the operations performed by a research engineer are monotonous routine procedures: frequency tuning, registration of detected information SEMI, recording the frequencies and levels of these emissions in a table. Therefore, automation of SEMI measurements is an important and urgent task, both in practical terms and in theoretical terms. From the theoretical point of view, it is necessary to reasonably select a measurement automation method, and in practice it is necessary to form a measuring complex that implements this measurement automation method. Unfortunately, at present, significant discrepancies are observed in the measurement results obtained by known automated complexes, as well as between the results obtained using automated systems and by a qualified research engineer manually. This article examines the currently existing methods for automating the measurement of PEMIN and their specific implementations, and defines the advantages, disadvantages, and application limits of each method.
The task of measuring PEMIN
According to the current regulatory and methodological documents (RMD), when conducting special studies, it is necessary to measure informative PEMIN, i.e. such emissions and interference created by the technical means under study that contain information processed by this technical means. Such emissions constitute only a small share of the entire spectrum of emissions of the technical means. All other emissions should not be recorded. In order to isolate informational PEMIN, special test modes of operation are provided for on the technical means under study. The requirements for tests are defined in the relevant GOSTs and methods. Informational PEMIN from the technical means in the test mode must have the highest possible level and be easily recognized by ear. When searching for PEMIN, the researcher listens to the signals at the demodulator output of the measuring device through headphones, simultaneously observing the oscillograms of these signals. If a signal similar to the desired test signal is detected, the researcher, by switching off and on the test mode of the technical device under study, makes sure that the signal is actually generated by this device and is information side radiation (interference). Thus, the first criterion for the researcher is the information coloring of the desired signal. The second, no less important criterion is the change in the level at the frequency of the detected signal when the test is switched on and off on the technical device under study. The engineer may encounter difficulties in registering changes in the level if the PEMIN level in the test mode differs slightly from the level in the normal mode, and in this case, it is often necessary to make a decision on assigning this signal to the PEMIN spectrum based only on the presence of information coloring.
As is known, most information PEMINs are a sequence of rectangular pulses (packets of rectangular pulses). The spectrum of such a signal is described by the function (sin x)/x and has the following form:
Since the level of the side components of signals with an audio envelope is lower than the level of the central peak, at a signal level slightly exceeding the noise level, the side components may be lower than the noise level, and the audio envelope will not be heard. In order to hear the audio envelope, it is necessary to bring the antenna close to the technical means, thereby increasing the signal-to-noise ratio. To measure the signal level, the antenna, in accordance with the methodology, should be moved to a distance of 1 m or more. Since this operation must be repeated at a large number of frequencies, it takes a lot of time in total.
The search for information PEMIN requires the researcher to be constantly focused and attentive. But a person can work in this mode only for a limited time: one, maximum two hours, after which he needs rest, the required duration of which is determined by his individual characteristics. With longer work, an effect is observed, colloquially called «blurring», when the researcher stops recognizing signals among noise, misses harmonic components, makes mistakes in measurements. The constantly growing fleet of electronic computing equipment, including those processing secret and confidential information, requires an increase in the volume of special research. It is not always possible to proportionally increase the staff of research engineers, for obvious reasons. Thus, automation of the PEMIN measurement process is a natural solution to the problem. But it would be desirable that the quality of work does not suffer.
Automation Methods
Until recently, the domestic market for automated PEMIN measurement tools was represented by only two systems: «Navigator» manufactured by ZAO «Nelk» and «Zarnitsa» manufactured by GUP «SNPO Eleron». Currently, two more families of systems for conducting special studies have appeared on the market: «Legenda» from FSUE «NPP «Gamma» and «Sigurd» manufactured by ZAO «Maskom». All systems, in one way or another, solve the same problem, therefore, undoubtedly, have quite a lot in common, but there are also differences, and quite significant ones. To clarify these differences, let us consider the automation methods used in the complexes.
- Automation of detection of harmonic components of the test signal.As noted above, a research engineer searches for harmonic components «by ear», recognizing the sought-after components by the sound and shape of the demodulated signal oscillogram. Instrumental implementation of such a mode usually leads to the fact that the automatic system recognizing signals by their shape works only slightly faster than a qualified research engineer. Therefore, this mode was not implemented in the first complexes, and recognition is performed by the criterion of changing signal levels when the test mode is turned on on the technical device under study (the so-called «energy criterion»). This method gives good results: all detection work is reduced to two scans of the special research range: during the first pass, the noise pattern is remembered with the test mode turned off, during the second pass, the technical device under study is transferred to the test mode, and the levels of all signals exceeding the remembered noise by a specified threshold value are measured.
The acceleration of work is very significant: instead of several hours (or even working days), a special study is performed in a matter of minutes. As a result, the research engineer receives a table of frequencies and signal levels (the typical number of detected components is several hundred) and can calculate the reconnaissance availability zones. Unfortunately, the calculation results may be incorrect. The fact is that the electromagnetic environment tends to change over time. Thousands of radio stations and sources of radio interference operate in the range from 9 kHz to 1000 MHz. Some of them turn on and off from time to time, and if some source of radio emission was not working during the noise spectrum scanning, but turned on during the second pass, its frequency will be in the list of detected components. Naturally, this can randomly change the calculated sizes of reconnaissance availability zones. Thus, the operator has to manually check all the detected components, which, with sufficiently weak signals, will again take time. This method works truly effectively in anechoic shielded chambers, which, due to their high cost, are available to very few enterprises.
The more advanced last two complexes use automatic recognition of information signals. According to the methodology, the research engineer is asked to search for any harmonic component manually or in a special «semi-automatic» mode, or to create a reference image of the desired signal using an editor (generator), or to select a previously created image from the library, after which the complex automatically detects signals similar to the specified signal on the air. To identify signals, both complexes use a cross-correlation function, in «Legend» — in combination with the Bayesian criterion of minimum risk. As noted above, this is a more time-consuming method, but also significantly more accurate. According to the results of certification tests of the «Legend» complex, the data obtained with its help do not require manual verification (certificate of the State Technical Commission under the President of the Russian Federation No. 603). The Sigurd complex has not yet passed certification tests, but, apparently, their results will be no worse. To overcome the lag in operating speed from complexes using the «energy» criterion, developers of complexes operating on the «information» criterion use various techniques, such as signal analysis in the vicinity of frequencies that are multiples of the test clock frequency, and measurements of batches of similar technical equipment using frequency templates. This leads to a noticeable acceleration of operation without reducing accuracy, but, unfortunately, not in all operating modes of the complexes: for example, the search mode for parasitic high-frequency generation, declared in the «Legend» complex, can only be implemented with «non-gap», and therefore relatively slow, control. It should be understood that although the Legend and Sigurd systems do not require the «greenhouse» conditions of a shielded chamber to operate, the average time savings compared to manual measurements performed by highly qualified research engineers may be lower than one might expect from reading the advertisement.
- Automation of signal level measurement.
The systems considered in this article are based on foreign-made radio receivers. The Zarnitsa system uses modified scanning radio receivers from AOR (Japan). The modification consists in the fact that an intermediate frequency digital signal output is added to the receiver, and the system is calibrated to ensure measurement of absolute values of signal levels. The accuracy and stability of the measured values of levels in such devices raises serious doubts. The fact is that AOR receivers are not intended for radio measurements and AOR does not guarantee the stability of their parameters. Therefore, no one can guarantee the accuracy of measurements with changes in temperature, humidity, atmospheric pressure, as well as some time after calibration and certification. The company «Nelk», which had experience in developing measuring systems based on similarly modernized devices, was forced to abandon this approach, since, according to specialists from the company «Nelk», scanning receivers, a year after calibration, gave discrepancies with the actual values of the signal level by 9-10 dB, that is, several times.The Navigator and Legenda complexes use spectrum analyzers from Agilent Technologies of the 85xx and ESA (E4411B, etc.) series. The Legenda can also be equipped with measuring receivers from Rohde&Schwarz or domestic devices (manufactured by SKB RIAP). However, these configuration options can hardly be recommended for mass use due to the high cost of the first of the named devices and the inaccessibility of the latter (the delivery time of a domestic device may even be longer than the delivery time of a foreign-made device, measured in months). The Sigurd complexes work with spectrum analyzers from IFR (Marconi) and Agilent Technologies (ESA series). These are metrological devices that have a detector and allow measuring peak values of electromagnetic radiation levels. Of course, the device must be calibrated and verified, which must be confirmed by a metrological certificate or, if the device is included in the register of measuring equipment of the State Standard, a verification certificate. However, only those devices that are equipped with a quasi-peak detector can directly obtain peak (quasi-peak) values. If the device does not have such a detector, then a special technique must be used to correctly measure peak level values. It is necessary to strictly adhere to the measurement parameters, such as scanning time, bandwidth, detector selection, otherwise, for some signal forms, the level values obtained using a spectrum analyzer may differ from the actual peak values. Consequently, automated systems built on the basis of spectrum analyzers must implement a technique for correctly measuring peak values, and its operation must be verified during certification tests of the system. Currently, such a technique, agreed upon with Rostest, and confirmed by the certificate of the State Technical Commission under the President of the Russian Federation, is used only in the Legenda complexes.
- Measuring interference in the power supply network, lines and communications.
According to the current NMD, the measurement of interference in the power supply network should be carried out using a network equivalent or voltage probes. A network equivalent is known to be a rather complex and relatively expensive device, but the measurements made with it are usually more accurate than those made with a voltage probe. A «clean» network, simulated by a network equivalent, allows measuring the interference created by the technical device under study in the power supply network, the level of which is 4-6 dB higher than the intrinsic noise of the network equivalent, while the accuracy of measurements made with a voltage probe depends on the noise levels of the power supply network. The EMCO 3810/2 network equivalent is included in the basic delivery set of the «Legend» systems (at the customer's request, the manufacturer can replace the network equivalent with voltage probes manufactured by Agilent Technologies or domestically produced by SKB «RIAP», etc.). The voltage probe manufactured by IFR is supplied as an additional option to the «Sigurd» complexes. The «Navigator» and «Zarnitsa» complexes are not supplied with voltage probes or network equivalents, the measuring current collectors included in their composition can only be used to detect interference, but not to measure their levels, but this contradicts the current methodology, which requires measuring voltage levels, not current strength. It would seem that this point has no direct relation to automation, however, this is not entirely true. In our opinion, the ability to use various receiving devices in their composition is very important for automated measuring systems: antennas, voltage probes, network equivalents. Accordingly, the software of the complex should provide a mechanism for supporting additional receiving devices, namely, the ability to enter such parameters as the operating range, antenna coefficients (attenuation or gain coefficients) and their automatic consideration during the measurement process. Today, the Legend complexes have such a mechanism.
Conclusions
The complexes for special research presented on the domestic market allow to solve a number of problems of PEMIN measurements in automatic mode and are capable to some extent to facilitate the work of the research engineer, to increase his productivity. Complexes based on scanning receivers («Zarnitsa») are suitable for rapid analysis of the PEMIN spectrum emitted by the technical means, but do not provide high measurement accuracy. If it is necessary to issue an order for the operation of the technical means, the measurements made by means of this complex are subject to mandatory manual verification using metrological measuring equipment (measuring receivers or spectrum analyzers). The «Navigator» complexes are suitable for conducting sufficiently accurate PEMIN measurements in shielded premises (anechoic shielded chambers), however, the measurement results can be correct only with their careful manual verification using the «Navigator» itself. The Legenda complexes allow measurements to be taken that do not require manual verification, which is confirmed by the State Technical Commission certificate. Apparently, the same can be said about the Sigurd complexes after the relevant certification tests have been completed. However, the use of the Legenda and Sigurd complexes requires the operator to be highly qualified and have a clear knowledge of the methodology for conducting special studies, since the actual research and creative part of the methodology — identifying the structure of the test signal, creating a reference image, and formulating a task for conducting measurements — remains with the person. However, whether this circumstance should be considered a disadvantage is up to you to decide. We are convinced that high qualifications and knowledge have always been and remain the basis of a professional approach.