rus
Issue #6/2025
V. V. Grishachev
The Speech Information Leakage Through the Fiber-Optic Technical Channels of the Office: Noise Control (Noise Reduction)
The Speech Information Leakage Through the Fiber-Optic Technical Channels of the Office: Noise Control (Noise Reduction)
DOI: 10.22184/1993-7296.FRos.2025.19.6.486.499
The efficiency of functioning and neutralization of speech information leakage channels through the fiber-optic utilities is largely determined by the noise pollution of the optical channel in the structured cabling systems of the facility. Adaptation of the well-known noise control methods in the fiber-optic speech information leakage channels allows to significantly increase the signal-to-noise ratio and, consequently, the interception efficiency. In particular, the fiber-optic subsystem of structured cabling systems at the facility with the distributed measuring properties can be represented as a phased acoustic fiber-optic array. While using the integration method for multichannel measurement, an informative signal can be isolated against the noise background. Another noise control method represents the comparison of two optical channels with asymmetric acoustic sensitivity (the differential method) that is possible due to the placement of optic fiber duplex cables in the cable ducts. The differential noise reduction method was tested on an experimental unit for investigating the speech information leakage channel that demonstrated a significant increase in the signal-to-noise ratio even at the short cable lengths. The differential method implementation is possible due to the correlation of low-frequency noise in the optical channel over a long period of time (0.1 ms) and a significant light path in the optical fiber (20 km). The given theoretical and practical studies show a high level of danger of acoustic fiber-optic speech information leakage channels and possible countermeasures.
The efficiency of functioning and neutralization of speech information leakage channels through the fiber-optic utilities is largely determined by the noise pollution of the optical channel in the structured cabling systems of the facility. Adaptation of the well-known noise control methods in the fiber-optic speech information leakage channels allows to significantly increase the signal-to-noise ratio and, consequently, the interception efficiency. In particular, the fiber-optic subsystem of structured cabling systems at the facility with the distributed measuring properties can be represented as a phased acoustic fiber-optic array. While using the integration method for multichannel measurement, an informative signal can be isolated against the noise background. Another noise control method represents the comparison of two optical channels with asymmetric acoustic sensitivity (the differential method) that is possible due to the placement of optic fiber duplex cables in the cable ducts. The differential noise reduction method was tested on an experimental unit for investigating the speech information leakage channel that demonstrated a significant increase in the signal-to-noise ratio even at the short cable lengths. The differential method implementation is possible due to the correlation of low-frequency noise in the optical channel over a long period of time (0.1 ms) and a significant light path in the optical fiber (20 km). The given theoretical and practical studies show a high level of danger of acoustic fiber-optic speech information leakage channels and possible countermeasures.
Теги: acoustic fiber-optic arrays in the leakage channel fiber-optic channel noise pollution low-frequency noise suppression in a fiber-optic channel noise reduction by differential measurements phased array noise reduction speech information interception through the fiber-optic communi акустические волоконно-оптические решетки в канале утечки зашумление волоконно-оптического канала информационная безопасность волоконно-оптических технологий перехват речевой информации через волоконно-оптические коммуник подавление низкочастотного шума в волоконно-оптическом канале шумоподавление дифференциальными измерениями.keywords: informati шумоподавление фазированными распределенными измерениями
The Speech Information Leakage Through the Fiber-Optic Technical Channels of the Office: Noise Control (Noise Reduction)
V. V. Grishachev
Institute for Information Sciences and Security Technologies
Russian State University of the Humanities, Moscow, Russia
The efficiency of functioning and neutralization of speech information leakage channels through the fiber-optic utilities is largely determined by the noise pollution of the optical channel in the structured cabling systems of the facility. Adaptation of the well-known noise control methods in the fiber-optic speech information leakage channels allows to significantly increase the signal-to-noise ratio and, consequently, the interception efficiency. In particular, the fiber-optic subsystem of structured cabling systems at the facility with the distributed measuring properties can be represented as a phased acoustic fiber-optic array. While using the integration method for multichannel measurement, an informative signal can be isolated against the noise background. Another noise control method represents the comparison of two optical channels with asymmetric acoustic sensitivity (the differential method) that is possible due to the placement of optic fiber duplex cables in the cable ducts. The differential noise reduction method was tested on an experimental unit for investigating the speech information leakage channel that demonstrated a significant increase in the signal-to-noise ratio even at the short cable lengths.
The differential method implementation is possible due to the correlation of low-frequency noise in the optical channel over a long period of time (0.1 ms) and a significant light path in the optical fiber (20 km). The given theoretical and practical studies show a high level of danger of acoustic fiber-optic speech information leakage channels and possible countermeasures.
Keywords: information security of fiber-optic technologies, speech information interception through the fiber-optic communications, fiber-optic channel noise pollution, low-frequency noise suppression in a fiber-optic channel, phased array noise reduction, acoustic fiber-optic arrays in the leakage channel, noise reduction by differential measurements
Article received: August 14, 2025
Article accepted: September 2, 2025
Introduction
Methods for enhancing/degrading the speech signal quality in a technical information leakage channel (TILC) via the fiber-optic communications (FOC) play a significant role in both information interception and protection. Preparation of the technical information protection (TIP) system is impossible without any knowledge of the methods used by an intruder to enhance the leakage channel efficiency [1–5]. The interception structure separately identifies the processing methods for the received speech signal by noise suppression or sound cleaning using any technical and computer methods, but their capabilities are determined by the quality of speech signal intercepted by the technical intelligence (TI) means. If the task is limited to data disclosure in the form of a speech report of the received message, the interception quality may be quite low. If it is necessary to obtain information in a form sufficient for publishing an audio message in the mass media or presenting in court, with the ability to determine not only the message meaning but also the identification and emotional coloring of the negotiators, then a higher quality is required.
To obtain the high-quality speech information, the efficient speech TILCs are used, where an increased quality of the received audio signal is achieved by the technical means by reducing the noise contribution into the overall signal relative to the useful signal (SNR). A fairly large number of such methods is available, among which the integration methods over time or over multiple multichannel measurements, autocorrelation methods, noise filtering, and other noise control methods are most popular [6, 7]. The application of a specific method is related to the specifications of information signal formation, propagation, and recording.
The TI structure and means for intercepting speech information using the acoustic fiber-optic TILCs are largely the same as the structure and technology of fiber-optic perimeter security systems (FOPS), but they have significant differences [8, 9], among which the following key distinguishing features of the acoustic fiber-optic TILCs and FOPS can be determined:
the source of speech signal in the leakage channel is distributed in the form of diffuse airborne sound throughout the entire designated area, in contrast to the security systems, where the source is single-point vibroacoustic vibrations of the ground or fences;
the primary propagation medium in the leakage channel is the ambient air of the premises and building walls with the natural waveguide properties, in contrast to the security systems, where the propagation medium is the free ambient space (soil, fences, air) without any waveguide properties;
the leakage channel structure is generated by the standard transparent fiber-optic structured cabling systems (SCS), covering the entire building and extending beyond its boundaries, in contrast to the security systems, where the fiber-optic cable is placed along the facility perimeter and intersects with the vibroacoustic wave from the source, as a rule, only in one point.
The given differences between the acoustic fiber-optic TILCs and FOPS are fundamental and prevent direct transfer of security system technologies to the TI systems, particularly in the field of noise reduction. The primary method for reducing the noise contribution into the received signal in security systems based on the optical reflectometry is the noise from the laser probing sources and local vibrational transformations [8]. In the speech information leakage channel, the noise reduction possibilities are significantly greater, as will be demonstrated in this article.
1. Noise reduction (noise control) by the phased distributed measurement methods [10]
One of the most efficient and widely used noise control methods is the useful signal integration, implemented in various ways [6]. In the case of acoustic measurements, the integration methods include the phased acoustic array method, when the signal is measured by an orderly arranged acoustic sensors and combined with the phase weight coefficients. Adaptation of this method to the speech TILCs via the FOC is determined by the stationary placement of a standard fiber-optic SCS relative to the acoustic signal source. The SCS elements have high acoustic sensitivity, and the application of passive fiber-optic inserts in the form of optical SCS inhomogeneities significantly increases the useful signal value against the noise background [11]. The structured cabling system of the protected facility contains a large number of passive fiber-optic elements that can be used for real-time recording of the speech signal by a reflectometer. Although the arrangement of acoustic transducers in the fiber is random, their arrangement stationarity in the building allows for determination of the phase weight coefficient of each transducer. The method efficiency is also affected by the number of optical fibers used in the probing process, as well as the number of embedded/detected fiber-optic inserts that are highly sensitive to the airborne and structure-borne sound. However, the risk of method implementation is largely individual and is determined by the acoustic specifications of the facility and the fiber-optic SCS [10].
In fact, the standard fiber-optic SCS of the protected facility is a phased antenna array with a random stationary spatial arrangement of acoustic transducers based on the distributed fiber-optic measuring systems (fig. 1). The acoustic fiber-optic transducers in the form of localized inhomogeneities of cable fibers are stationary and randomly placed at the facility that is a single source of an informative signal in the form of airborne and structure-borne sound that allows for adjustment of the entire measuring system by determining the weighting factor of each transducer based on the recorded signal level and the phase relations of the sound delay from the source (premises). The number N of acoustic transducers in the measuring system depends on the specifications of the fiber-optic SCS, namely the number of optical inhomogeneities sensitive to the acoustic impact, as well as on the sensitivity of the applied TI means. The TILC implementation will increase the signal-to-noise ratio by N times relative to a certain average value [6], and under the certain conditions it can be increased up to N3/2 times [10].
2. Noise reduction (noise control) by the differential measurement methods
Another method for reducing the information signal noise level involves comparing the signal that has passed through the leakage channel with the control channel that differ in the conditions of interaction with the protected speech signal while their interactions with the background ambient fields coincide. The generated modulations of optical radiation remain correlated with each other after traveling a certain distance during the period of autocorrelation (FIG. 2). The main feature of the analog speech information signal is a low upper spectrum limit of approximately 10 kHz that allows for an autocorrelation time of 10–4 seconds when the optical radiation travels a distance of 20,000 m.
When the optical radiation (2) propagates from a background source of low-frequency oscillations (3) affecting an optical inhomogeneity (4) in an optical fiber (1), the modulation distortions (noise) of the optical radiation (4) are generated and remain coherent during the autocorrelation time τ (5). The optical flow passes through a fiber with a refractive index n at a distance L. The long-term autocorrelation of low-frequency modulations of optical radiation makes it possible to reduce noise at the leakage channel output by comparing the measured parameters of the probing optical radiation from the sections located closely along the optical fiber. The problem of efficient separation of the informative signal from the noise signal is reduced to isolating the test fiber section in one signal from the control section in the other. Such implementation can be performed in various ways, but the general physical principles are related to a significant autocorrelation time [7].
Let us assume that in the probing scheme (fig. 3) along an optical fiber (1) with an optical pulse (2) of the fiber sections (4, 5), the acoustic signal from the sound source (3) causes generation of an informative optical signal in the form of parasitic modulation of the optical radiation parameters. If one section (4) has increased sensitivity relative to the adjacent sections (5), then separate registration of the optical radiation parameters forms the informative signals with correlated noise n(t) from the section (5) and with a time delay τ of noise n(t + τ) from the section (4), where the useful signal is generated s(t + τ). The signal comparison by subtraction allows us to isolate the useful signal
s(t + τ) + n(t + τ) − n(t).
The root-mean-square power of the useful signal S and noise N in the total power is equal to
S + 2N(1 − k),
where, k( τ) is the noise autocorrelation coefficient, dependent on the delay time τ between the primary and secondary measurements. If the delay between measurements is less than the correlation time of 10−4 sec, then k > 0 and tends to 1 as the control measurement approaches the test measurement that allows for complete suppression of channel noise at the short distances in the limit (in our case, less than 20 km).
Implementation of the differential measurement mechanism by subtracting noise from the informative signal can be performed in many ways, the efficiency of which mainly depends on the range and depth of the optical radiation acoustic modulation in the fiber-optic insert during generation of the useful component in the informative signal relative to the background parasitic modulations along the length of the fiber-optic leakage channel, as well as the possible isolation of the noise component in the signal.
2.1. Differential probing by the optical reflectometry methods
The requirements for generating two noise-correlated test and control signals are best met by the optical reflectometry methods (fig. 4) that allow separation of the informative signal and leakage channel noise in time [9]. The reflectometer (A) sends probing pulses into the optical fiber (B) and receives the backscattered radiation in real time. This activity results in the reflectogram (C) formation. The reflectometer compares the signals in real time with the integrated acoustic signal output, using the time separation of response (D) from the adjacent points on the optical fiber (1) and (2).
Differential probing in the reflectometric scheme can be implemented not only by real-time processing of reflectograms using the software methods, but also by directly comparing the optical responses from selected fiber sections with the signal separation using a Mach- Zehnder interferometer at the recording system input. In this case, a device is implemented for analog calculating the correlation coefficient with a time equal to the delay time between the interferometer arms. Under certain conditions, this optical circuit allows for continuous-wave probing. Another method is two-pulse reflectometry that also allows for separation of the two signals being compared.
2.2. Differential Probing in the Transmitted Light of a Duplex Cable with Asymmetric Optical Inhomogeneity
Although the reflectometric probing scheme is preferable for probing over long distances, the efficient probing implementation in direct (transmitted) radiation is possible at the short distances with a dedicated fiber-optic insert having increased acoustic sensitivity relative to other parts of the optical fiber. One of the possible circuits is related to the installation of fiber-optic SCS from a cable with several optical fibers. For example (fig. 5), while passing through the same cable channels, the fibers of a duplex optical cable (2) between the transceivers (1) are subject to the same effects (4, 5) with a uniformity size exceeding the distance between the fibers in the duplex cable, therefore the noise components of information signal correlate with each other. In this system, any asymmetric conditions for the optical radiation propagation are achieved by forming the fiber-optic inserts (3) on one optical fiber without changing the condition of the other. By conducting forward flow probing simultaneously on two fibers, regardless of the propagation direction of optical radiation, and comparing them, it is possible to increase the signal-to-noise ratio for a speech signal by comparing two signals.
The discussed differential probing methods used to improve the speech signal quality being monitored are not the only ones. Their implementation depends on the specific SCS of the building: its placement, the cable used, fragmentation, application of security systems, and other factors.
3. Experimental studies of noise suppression by the differential measurements
In the operation of a speech information leakage channel through the fiber-optic communications, the optical inhomogeneities with high acoustic sensitivity (fiber-optic inserts), where the information signal is generated, are of great importance. One of such elements is a detachable optical connection that is almost typical in design to a fiber-optic microphone. By manipulating these elements, the modulation depth of forward and reverse optical radiation can be increased to 100% that may have a minor impact on transmitted traffic due to the low modulation frequency. This is one of the most dangerous elements in the threat model, as confirmed by the practical studies [12, 13]. Therefore, practical testing of efficiency of the differential noise suppression methods was conducted with a detachable connection.
The experimental studies of differential noise suppression methods were conducted on a modified laboratory setup previously used to study the sound impact on detachable connections [12, 13]. The setup (fig. 6) includes a control multimedia computer (1) playing back an audio file with a speech, a noise generator, and a tone signal generator. The optical probing was performed using a highly stable helium-neon laser (2) with a fiber-optic output to an FC connector with a power of several mW at a wavelength of 632 nm, to which a 50/50 fiber-optic splitter (5) was connected. One part of the equally divided radiation was directed into the test channel with the SC-SC detachable connection under study (4) to the photodiode (6), and the other part was directed to the control channel to the photodiode (7). The investigated SC-SC fiber-optic connector was located in a small-sized chamber (3) with a sound insulation from the environment of no more than 20 dB, where a homogeneous diffuse acoustic field was developed using an acoustic system and monitored by a microphone connected to a control computer (1). The photodiodes (6) and (7) were operated in the photovoltaic mode and connected through a 10 dB differential amplifier (8) to the monitoring devices: a selective nanovoltmeter of the UNIPAN type (9) with an output to the on-ear headphones (10), a digital oscilloscope (11) and a computer with a software audio spectrum analyzer (12). The monitoring devices (11) and (12) were connected directly to the differential amplifier (8) or through a selective nanovoltmeter (9) instead of the headphones (10) and were used for qualitative assessments and adjustment of the measuring system. The fiber-optic system had a total length of approximately 5 m after the splitter (5). All fiber-optic elements (laser, cable, splitter, detachable connections, soundproofing chamber) and measuring instruments were compactly arranged on a single table in three tiers, thus developing conditions for uniform exposure to the background sounds on all elements except the detachable connection being tested in the soundproofing chamber.
The research was conducted by two methods as follows:
the subjective articulation measurements in the form of listening to the informative speech signals through a measuring system and qualitative speech evaluation by ear, for which a speech signal was applied to a detachable connection through an acoustic system in a small-sized chamber; an amplifier with a selectivity of 0 dB/octave was used to achieve a sound pressure level sufficient for listening to the intercepted text through the headphones; the next step was to optically break the control channel and repeat the break and restoration of the optical channel to perform a comparison of the speech quality at a fixed sound pressure level in the small-sized chamber;
in the objective measurements of the signal-to-noise ratio, the effect of tonal sounds was used, for which a fixed-power tone signal at a frequency of 1016 Hz was applied to a detachable connection through an acoustic system in a small-sized chamber; an amplifier with a maximum selectivity of 54 dB/octave was used to measure the sound pressure level of the signal in μV; while maintaining the settings, the tone signal was turned off and the noise sound pressure level in μV was measured; the next step was an optical break in the control channel with subsequent repetition of the break and restoration of the optical channel, the sound pressure level of the signal and noise in the small-sized chamber was measured.
During the system adjustment, the sound field in a small-sized chamber, as well as at the output of a nanovoltmeter was monitored for the spectrum of background, speech and tone signals.
The experimental measurements by comparing signals during the direct measurement in one channel and during differential measurement of two channels showed the following:
articulation measurements (comparison by ear in the headphones) of the differential and direct connection demonstrate the advantage of differential measurement scheme represented by the fact that the direct scheme produces noise that approximately coincides with the speech, while in the case of differential scheme the noise level is lower than the speech level;
a study of the noise characteristics of different switching modes when exposed to a tone signal at a frequency of 1016 Hz and a selectivity of 54 dB/octave provides the differential mode: tone 3.3 μV, background 0.17 μV, SNR = 19.4; direct activation (break of the control channel): tone 2.5 μV; background 0.20 μV, SNR = 12.5;
As can be seen from the results of experimental studies, the differential measurement mode allows to improve the quality of intercepted speech signal.
4. Application of differential probing technologies
The low-frequency noise reduction capabilities of an optical channel can be used to improve the efficiency of an acoustic fiber-optic TILC. The basic structure of a leakage channel with a differential noise reduction can be constructed on the basis of the standard optical network of the facility (fig. 7). A duplex optical cable (4) with the counter-directional information flows (5) can be used to connect two workstations (3). The cable is placed in the cable ducts of the fiber-optic SCS subsystems, one channel of which has an optical heterogeneity (2) with the acoustic contact through the air or vibroacoustic contact through the walls with the speech information source (1). While monitoring the average optical power of the standard counter-directional information flows with a time resolution of 10 kHz and a magnitude in a dynamic range of 8 bits, the intruder (6) receives two signals with the same noise in each, but with an asymmetric value relative to the information signal. The control data is transmitted to the intruder from the workstations in real time and processed using differential noise reduction methods.
The leakage channel structure depends on the optical network topology. If the topology is of a star, tree, or bus structure, with the cable ducts running through the controlled area, an external intruder can intercept the network through its own computers. An optical network within the controlled area requires an internal intruder with unauthorized access to the facility’s workstations. The confidential information interception is significantly simplified with a ring network topology, as it is possible to combine all three computers in a single TI device that can be located either within the controlled area (internal intruder) or outside the controlled area (external intruder).
In addition to the FOC-based speech TILCs, the differential method of low-frequency noise reduction in an optical channel can be used in other fiber-optic technologies, for example, to increase the protected perimeter length with a single reflectometer in the FOPS without intermediate optical amplifiers [8, 9]. In this case, while using the differential probing scheme with a reflectometer (FIG. 4), it is possible to increase the controlled perimeter length by an order of magnitude, since the main controlled parameter is the vibroacoustic sources with a frequency of up to 1 kHz. It corresponds to a correlation time of 1 ms or a fiber length of 200 km. Moreover, the technology can be used to increase the fiber-optic cable length in the fiber-optic microphones [4] by using a duplex cable with one converter fiber and another control fiber without a converter (FIG. 5). In this case, the microphone cable length can be increased by any required value up to 20 km.
Conclusion
The available transparent fiber-optic SCS subsystems with the ability to converge data transmission and measurement functions pose a significant threat to the speech information circulating within a facility due to the propagation of informative acoustic fields throughout the entire facility (building) that is completely covered by a fixed fiber-optic SCS with high acoustic sensitivity. The intruder can apply the noise reduction methods to intercept information from many positions. However, the knowledge of leakage channels can significantly improve technical information security by installing asymmetrical induced noise generation systems in the optical cables.
REFERENCES
Buzov G. A. Zashchita informacii ogranichennogo dostupa ot utechki po tekhnicheskim kanalam. – M.: Goryachaya liniya – Telekom, 2024. 586 s. ISBN 978‑5‑9912‑0424‑8.
Katorin YU.F., Razumovskij A. V., Spivak A. I. Zashchita informacii tekhnicheskimi sredstvami. – SPb: NIU ITMO, 2012. 416 s.
Zajcev A.P., Shelupanov A. A., Meshcheryakov R. V. et al. Tekhnicheskie sredstva i metody zashchity informacii. – M.: OOO «Izdatel’stvo MashinostroeniE», 2009. 508 s. ISBN 978‑5‑94275‑454‑9.
Khorev A. A. Tekhnicheskaya zashchita informacii. Tom 1. Tekhnicheskie kanaly utechki informacii. – M.: NPC «AnalitikA», 2008. 436 s.
Khalyapin D. B. Zashchita informacii. Vas podslushivayut? Zashchishchajtes’! – M.: NOU SHO «BayarD», 2004. 432 s.
Kharkevich A. A. Bor’ba s pomekhami. – M.: Librokom. Lenand.2018–280 s. – ISBN 978‑5‑9710‑5339‑2, 978-5-397-03458-6
Bendat Dh., Pirsol A. Prikladnoj analiz sluchajnykh dannykh. – M.: Mir, 1989. 540 s.
Kul’chin YU.N. Raspredelennye volokonno-opticheskie izmeritel’nye sistemy. – M.: FIZMATLIT. 2001. 272 s. – ISBN 978‑5‑9221‑0072‑4.
Listvin A.V., Listvin V. N. Reflektometriya opticheskikh volokon. – LESARart, 2005. 208 s. ISBN 5‑902367‑03‑4.
Grishachev V.V., Kazarin O. V., Kalinina YU.D. Fizicheskaya model’ ugrozy utechki akusticheskoj (rechevoj) informacii cherez volokonno-opticheskie kommunikacii. Voprosy zashchity informacii. 2018; 3: 35–51.
Grishachev V.V., Zabolockaya A. D. Problema informacionnoj bezopasnosti volokonno-opticheskikh tekhnologij. Photonics Russia. 2022; 16(6): 484–500. DOI: 10.22184/1993‑7296.FRos.2022.16.6.484.500.
Grishachev V.V., Kalinina YU.D., Tarasov A. A. Ocenka glubiny parazitnoj modulyacii sveta v opticheskoj kabel’noj sisteme s neodnorodnostyami. Voprosy zashchity informacii. 2016; 1: 62–73.
Grishachev V.V., Kazarin O. V., Kalinina YU.D. Parazitnye akusticheskie modulyacii svetovykh potokov v raz’emnykh soedineniyakh opticheskoj seti ob’ekta informatizacii. Voprosy zashchity informacii. 2018; 4: 47–56.
AUTHOR
Grishachev Vladimir V., Cand. of Sc.(Phys.&Math.), associate professor, Institute for Information Sciences and Security Technologies (IISST), Russian State University of the Humanities (RSUH), e-mail: grishachev@mail.ru, Moscow, Russia
ORCID: 0000-0002-7585-7282
V. V. Grishachev
Institute for Information Sciences and Security Technologies
Russian State University of the Humanities, Moscow, Russia
The efficiency of functioning and neutralization of speech information leakage channels through the fiber-optic utilities is largely determined by the noise pollution of the optical channel in the structured cabling systems of the facility. Adaptation of the well-known noise control methods in the fiber-optic speech information leakage channels allows to significantly increase the signal-to-noise ratio and, consequently, the interception efficiency. In particular, the fiber-optic subsystem of structured cabling systems at the facility with the distributed measuring properties can be represented as a phased acoustic fiber-optic array. While using the integration method for multichannel measurement, an informative signal can be isolated against the noise background. Another noise control method represents the comparison of two optical channels with asymmetric acoustic sensitivity (the differential method) that is possible due to the placement of optic fiber duplex cables in the cable ducts. The differential noise reduction method was tested on an experimental unit for investigating the speech information leakage channel that demonstrated a significant increase in the signal-to-noise ratio even at the short cable lengths.
The differential method implementation is possible due to the correlation of low-frequency noise in the optical channel over a long period of time (0.1 ms) and a significant light path in the optical fiber (20 km). The given theoretical and practical studies show a high level of danger of acoustic fiber-optic speech information leakage channels and possible countermeasures.
Keywords: information security of fiber-optic technologies, speech information interception through the fiber-optic communications, fiber-optic channel noise pollution, low-frequency noise suppression in a fiber-optic channel, phased array noise reduction, acoustic fiber-optic arrays in the leakage channel, noise reduction by differential measurements
Article received: August 14, 2025
Article accepted: September 2, 2025
Introduction
Methods for enhancing/degrading the speech signal quality in a technical information leakage channel (TILC) via the fiber-optic communications (FOC) play a significant role in both information interception and protection. Preparation of the technical information protection (TIP) system is impossible without any knowledge of the methods used by an intruder to enhance the leakage channel efficiency [1–5]. The interception structure separately identifies the processing methods for the received speech signal by noise suppression or sound cleaning using any technical and computer methods, but their capabilities are determined by the quality of speech signal intercepted by the technical intelligence (TI) means. If the task is limited to data disclosure in the form of a speech report of the received message, the interception quality may be quite low. If it is necessary to obtain information in a form sufficient for publishing an audio message in the mass media or presenting in court, with the ability to determine not only the message meaning but also the identification and emotional coloring of the negotiators, then a higher quality is required.
To obtain the high-quality speech information, the efficient speech TILCs are used, where an increased quality of the received audio signal is achieved by the technical means by reducing the noise contribution into the overall signal relative to the useful signal (SNR). A fairly large number of such methods is available, among which the integration methods over time or over multiple multichannel measurements, autocorrelation methods, noise filtering, and other noise control methods are most popular [6, 7]. The application of a specific method is related to the specifications of information signal formation, propagation, and recording.
The TI structure and means for intercepting speech information using the acoustic fiber-optic TILCs are largely the same as the structure and technology of fiber-optic perimeter security systems (FOPS), but they have significant differences [8, 9], among which the following key distinguishing features of the acoustic fiber-optic TILCs and FOPS can be determined:
the source of speech signal in the leakage channel is distributed in the form of diffuse airborne sound throughout the entire designated area, in contrast to the security systems, where the source is single-point vibroacoustic vibrations of the ground or fences;
the primary propagation medium in the leakage channel is the ambient air of the premises and building walls with the natural waveguide properties, in contrast to the security systems, where the propagation medium is the free ambient space (soil, fences, air) without any waveguide properties;
the leakage channel structure is generated by the standard transparent fiber-optic structured cabling systems (SCS), covering the entire building and extending beyond its boundaries, in contrast to the security systems, where the fiber-optic cable is placed along the facility perimeter and intersects with the vibroacoustic wave from the source, as a rule, only in one point.
The given differences between the acoustic fiber-optic TILCs and FOPS are fundamental and prevent direct transfer of security system technologies to the TI systems, particularly in the field of noise reduction. The primary method for reducing the noise contribution into the received signal in security systems based on the optical reflectometry is the noise from the laser probing sources and local vibrational transformations [8]. In the speech information leakage channel, the noise reduction possibilities are significantly greater, as will be demonstrated in this article.
1. Noise reduction (noise control) by the phased distributed measurement methods [10]
One of the most efficient and widely used noise control methods is the useful signal integration, implemented in various ways [6]. In the case of acoustic measurements, the integration methods include the phased acoustic array method, when the signal is measured by an orderly arranged acoustic sensors and combined with the phase weight coefficients. Adaptation of this method to the speech TILCs via the FOC is determined by the stationary placement of a standard fiber-optic SCS relative to the acoustic signal source. The SCS elements have high acoustic sensitivity, and the application of passive fiber-optic inserts in the form of optical SCS inhomogeneities significantly increases the useful signal value against the noise background [11]. The structured cabling system of the protected facility contains a large number of passive fiber-optic elements that can be used for real-time recording of the speech signal by a reflectometer. Although the arrangement of acoustic transducers in the fiber is random, their arrangement stationarity in the building allows for determination of the phase weight coefficient of each transducer. The method efficiency is also affected by the number of optical fibers used in the probing process, as well as the number of embedded/detected fiber-optic inserts that are highly sensitive to the airborne and structure-borne sound. However, the risk of method implementation is largely individual and is determined by the acoustic specifications of the facility and the fiber-optic SCS [10].
In fact, the standard fiber-optic SCS of the protected facility is a phased antenna array with a random stationary spatial arrangement of acoustic transducers based on the distributed fiber-optic measuring systems (fig. 1). The acoustic fiber-optic transducers in the form of localized inhomogeneities of cable fibers are stationary and randomly placed at the facility that is a single source of an informative signal in the form of airborne and structure-borne sound that allows for adjustment of the entire measuring system by determining the weighting factor of each transducer based on the recorded signal level and the phase relations of the sound delay from the source (premises). The number N of acoustic transducers in the measuring system depends on the specifications of the fiber-optic SCS, namely the number of optical inhomogeneities sensitive to the acoustic impact, as well as on the sensitivity of the applied TI means. The TILC implementation will increase the signal-to-noise ratio by N times relative to a certain average value [6], and under the certain conditions it can be increased up to N3/2 times [10].
2. Noise reduction (noise control) by the differential measurement methods
Another method for reducing the information signal noise level involves comparing the signal that has passed through the leakage channel with the control channel that differ in the conditions of interaction with the protected speech signal while their interactions with the background ambient fields coincide. The generated modulations of optical radiation remain correlated with each other after traveling a certain distance during the period of autocorrelation (FIG. 2). The main feature of the analog speech information signal is a low upper spectrum limit of approximately 10 kHz that allows for an autocorrelation time of 10–4 seconds when the optical radiation travels a distance of 20,000 m.
When the optical radiation (2) propagates from a background source of low-frequency oscillations (3) affecting an optical inhomogeneity (4) in an optical fiber (1), the modulation distortions (noise) of the optical radiation (4) are generated and remain coherent during the autocorrelation time τ (5). The optical flow passes through a fiber with a refractive index n at a distance L. The long-term autocorrelation of low-frequency modulations of optical radiation makes it possible to reduce noise at the leakage channel output by comparing the measured parameters of the probing optical radiation from the sections located closely along the optical fiber. The problem of efficient separation of the informative signal from the noise signal is reduced to isolating the test fiber section in one signal from the control section in the other. Such implementation can be performed in various ways, but the general physical principles are related to a significant autocorrelation time [7].
Let us assume that in the probing scheme (fig. 3) along an optical fiber (1) with an optical pulse (2) of the fiber sections (4, 5), the acoustic signal from the sound source (3) causes generation of an informative optical signal in the form of parasitic modulation of the optical radiation parameters. If one section (4) has increased sensitivity relative to the adjacent sections (5), then separate registration of the optical radiation parameters forms the informative signals with correlated noise n(t) from the section (5) and with a time delay τ of noise n(t + τ) from the section (4), where the useful signal is generated s(t + τ). The signal comparison by subtraction allows us to isolate the useful signal
s(t + τ) + n(t + τ) − n(t).
The root-mean-square power of the useful signal S and noise N in the total power is equal to
S + 2N(1 − k),
where, k( τ) is the noise autocorrelation coefficient, dependent on the delay time τ between the primary and secondary measurements. If the delay between measurements is less than the correlation time of 10−4 sec, then k > 0 and tends to 1 as the control measurement approaches the test measurement that allows for complete suppression of channel noise at the short distances in the limit (in our case, less than 20 km).
Implementation of the differential measurement mechanism by subtracting noise from the informative signal can be performed in many ways, the efficiency of which mainly depends on the range and depth of the optical radiation acoustic modulation in the fiber-optic insert during generation of the useful component in the informative signal relative to the background parasitic modulations along the length of the fiber-optic leakage channel, as well as the possible isolation of the noise component in the signal.
2.1. Differential probing by the optical reflectometry methods
The requirements for generating two noise-correlated test and control signals are best met by the optical reflectometry methods (fig. 4) that allow separation of the informative signal and leakage channel noise in time [9]. The reflectometer (A) sends probing pulses into the optical fiber (B) and receives the backscattered radiation in real time. This activity results in the reflectogram (C) formation. The reflectometer compares the signals in real time with the integrated acoustic signal output, using the time separation of response (D) from the adjacent points on the optical fiber (1) and (2).
Differential probing in the reflectometric scheme can be implemented not only by real-time processing of reflectograms using the software methods, but also by directly comparing the optical responses from selected fiber sections with the signal separation using a Mach- Zehnder interferometer at the recording system input. In this case, a device is implemented for analog calculating the correlation coefficient with a time equal to the delay time between the interferometer arms. Under certain conditions, this optical circuit allows for continuous-wave probing. Another method is two-pulse reflectometry that also allows for separation of the two signals being compared.
2.2. Differential Probing in the Transmitted Light of a Duplex Cable with Asymmetric Optical Inhomogeneity
Although the reflectometric probing scheme is preferable for probing over long distances, the efficient probing implementation in direct (transmitted) radiation is possible at the short distances with a dedicated fiber-optic insert having increased acoustic sensitivity relative to other parts of the optical fiber. One of the possible circuits is related to the installation of fiber-optic SCS from a cable with several optical fibers. For example (fig. 5), while passing through the same cable channels, the fibers of a duplex optical cable (2) between the transceivers (1) are subject to the same effects (4, 5) with a uniformity size exceeding the distance between the fibers in the duplex cable, therefore the noise components of information signal correlate with each other. In this system, any asymmetric conditions for the optical radiation propagation are achieved by forming the fiber-optic inserts (3) on one optical fiber without changing the condition of the other. By conducting forward flow probing simultaneously on two fibers, regardless of the propagation direction of optical radiation, and comparing them, it is possible to increase the signal-to-noise ratio for a speech signal by comparing two signals.
The discussed differential probing methods used to improve the speech signal quality being monitored are not the only ones. Their implementation depends on the specific SCS of the building: its placement, the cable used, fragmentation, application of security systems, and other factors.
3. Experimental studies of noise suppression by the differential measurements
In the operation of a speech information leakage channel through the fiber-optic communications, the optical inhomogeneities with high acoustic sensitivity (fiber-optic inserts), where the information signal is generated, are of great importance. One of such elements is a detachable optical connection that is almost typical in design to a fiber-optic microphone. By manipulating these elements, the modulation depth of forward and reverse optical radiation can be increased to 100% that may have a minor impact on transmitted traffic due to the low modulation frequency. This is one of the most dangerous elements in the threat model, as confirmed by the practical studies [12, 13]. Therefore, practical testing of efficiency of the differential noise suppression methods was conducted with a detachable connection.
The experimental studies of differential noise suppression methods were conducted on a modified laboratory setup previously used to study the sound impact on detachable connections [12, 13]. The setup (fig. 6) includes a control multimedia computer (1) playing back an audio file with a speech, a noise generator, and a tone signal generator. The optical probing was performed using a highly stable helium-neon laser (2) with a fiber-optic output to an FC connector with a power of several mW at a wavelength of 632 nm, to which a 50/50 fiber-optic splitter (5) was connected. One part of the equally divided radiation was directed into the test channel with the SC-SC detachable connection under study (4) to the photodiode (6), and the other part was directed to the control channel to the photodiode (7). The investigated SC-SC fiber-optic connector was located in a small-sized chamber (3) with a sound insulation from the environment of no more than 20 dB, where a homogeneous diffuse acoustic field was developed using an acoustic system and monitored by a microphone connected to a control computer (1). The photodiodes (6) and (7) were operated in the photovoltaic mode and connected through a 10 dB differential amplifier (8) to the monitoring devices: a selective nanovoltmeter of the UNIPAN type (9) with an output to the on-ear headphones (10), a digital oscilloscope (11) and a computer with a software audio spectrum analyzer (12). The monitoring devices (11) and (12) were connected directly to the differential amplifier (8) or through a selective nanovoltmeter (9) instead of the headphones (10) and were used for qualitative assessments and adjustment of the measuring system. The fiber-optic system had a total length of approximately 5 m after the splitter (5). All fiber-optic elements (laser, cable, splitter, detachable connections, soundproofing chamber) and measuring instruments were compactly arranged on a single table in three tiers, thus developing conditions for uniform exposure to the background sounds on all elements except the detachable connection being tested in the soundproofing chamber.
The research was conducted by two methods as follows:
the subjective articulation measurements in the form of listening to the informative speech signals through a measuring system and qualitative speech evaluation by ear, for which a speech signal was applied to a detachable connection through an acoustic system in a small-sized chamber; an amplifier with a selectivity of 0 dB/octave was used to achieve a sound pressure level sufficient for listening to the intercepted text through the headphones; the next step was to optically break the control channel and repeat the break and restoration of the optical channel to perform a comparison of the speech quality at a fixed sound pressure level in the small-sized chamber;
in the objective measurements of the signal-to-noise ratio, the effect of tonal sounds was used, for which a fixed-power tone signal at a frequency of 1016 Hz was applied to a detachable connection through an acoustic system in a small-sized chamber; an amplifier with a maximum selectivity of 54 dB/octave was used to measure the sound pressure level of the signal in μV; while maintaining the settings, the tone signal was turned off and the noise sound pressure level in μV was measured; the next step was an optical break in the control channel with subsequent repetition of the break and restoration of the optical channel, the sound pressure level of the signal and noise in the small-sized chamber was measured.
During the system adjustment, the sound field in a small-sized chamber, as well as at the output of a nanovoltmeter was monitored for the spectrum of background, speech and tone signals.
The experimental measurements by comparing signals during the direct measurement in one channel and during differential measurement of two channels showed the following:
articulation measurements (comparison by ear in the headphones) of the differential and direct connection demonstrate the advantage of differential measurement scheme represented by the fact that the direct scheme produces noise that approximately coincides with the speech, while in the case of differential scheme the noise level is lower than the speech level;
a study of the noise characteristics of different switching modes when exposed to a tone signal at a frequency of 1016 Hz and a selectivity of 54 dB/octave provides the differential mode: tone 3.3 μV, background 0.17 μV, SNR = 19.4; direct activation (break of the control channel): tone 2.5 μV; background 0.20 μV, SNR = 12.5;
As can be seen from the results of experimental studies, the differential measurement mode allows to improve the quality of intercepted speech signal.
4. Application of differential probing technologies
The low-frequency noise reduction capabilities of an optical channel can be used to improve the efficiency of an acoustic fiber-optic TILC. The basic structure of a leakage channel with a differential noise reduction can be constructed on the basis of the standard optical network of the facility (fig. 7). A duplex optical cable (4) with the counter-directional information flows (5) can be used to connect two workstations (3). The cable is placed in the cable ducts of the fiber-optic SCS subsystems, one channel of which has an optical heterogeneity (2) with the acoustic contact through the air or vibroacoustic contact through the walls with the speech information source (1). While monitoring the average optical power of the standard counter-directional information flows with a time resolution of 10 kHz and a magnitude in a dynamic range of 8 bits, the intruder (6) receives two signals with the same noise in each, but with an asymmetric value relative to the information signal. The control data is transmitted to the intruder from the workstations in real time and processed using differential noise reduction methods.
The leakage channel structure depends on the optical network topology. If the topology is of a star, tree, or bus structure, with the cable ducts running through the controlled area, an external intruder can intercept the network through its own computers. An optical network within the controlled area requires an internal intruder with unauthorized access to the facility’s workstations. The confidential information interception is significantly simplified with a ring network topology, as it is possible to combine all three computers in a single TI device that can be located either within the controlled area (internal intruder) or outside the controlled area (external intruder).
In addition to the FOC-based speech TILCs, the differential method of low-frequency noise reduction in an optical channel can be used in other fiber-optic technologies, for example, to increase the protected perimeter length with a single reflectometer in the FOPS without intermediate optical amplifiers [8, 9]. In this case, while using the differential probing scheme with a reflectometer (FIG. 4), it is possible to increase the controlled perimeter length by an order of magnitude, since the main controlled parameter is the vibroacoustic sources with a frequency of up to 1 kHz. It corresponds to a correlation time of 1 ms or a fiber length of 200 km. Moreover, the technology can be used to increase the fiber-optic cable length in the fiber-optic microphones [4] by using a duplex cable with one converter fiber and another control fiber without a converter (FIG. 5). In this case, the microphone cable length can be increased by any required value up to 20 km.
Conclusion
The available transparent fiber-optic SCS subsystems with the ability to converge data transmission and measurement functions pose a significant threat to the speech information circulating within a facility due to the propagation of informative acoustic fields throughout the entire facility (building) that is completely covered by a fixed fiber-optic SCS with high acoustic sensitivity. The intruder can apply the noise reduction methods to intercept information from many positions. However, the knowledge of leakage channels can significantly improve technical information security by installing asymmetrical induced noise generation systems in the optical cables.
REFERENCES
Buzov G. A. Zashchita informacii ogranichennogo dostupa ot utechki po tekhnicheskim kanalam. – M.: Goryachaya liniya – Telekom, 2024. 586 s. ISBN 978‑5‑9912‑0424‑8.
Katorin YU.F., Razumovskij A. V., Spivak A. I. Zashchita informacii tekhnicheskimi sredstvami. – SPb: NIU ITMO, 2012. 416 s.
Zajcev A.P., Shelupanov A. A., Meshcheryakov R. V. et al. Tekhnicheskie sredstva i metody zashchity informacii. – M.: OOO «Izdatel’stvo MashinostroeniE», 2009. 508 s. ISBN 978‑5‑94275‑454‑9.
Khorev A. A. Tekhnicheskaya zashchita informacii. Tom 1. Tekhnicheskie kanaly utechki informacii. – M.: NPC «AnalitikA», 2008. 436 s.
Khalyapin D. B. Zashchita informacii. Vas podslushivayut? Zashchishchajtes’! – M.: NOU SHO «BayarD», 2004. 432 s.
Kharkevich A. A. Bor’ba s pomekhami. – M.: Librokom. Lenand.2018–280 s. – ISBN 978‑5‑9710‑5339‑2, 978-5-397-03458-6
Bendat Dh., Pirsol A. Prikladnoj analiz sluchajnykh dannykh. – M.: Mir, 1989. 540 s.
Kul’chin YU.N. Raspredelennye volokonno-opticheskie izmeritel’nye sistemy. – M.: FIZMATLIT. 2001. 272 s. – ISBN 978‑5‑9221‑0072‑4.
Listvin A.V., Listvin V. N. Reflektometriya opticheskikh volokon. – LESARart, 2005. 208 s. ISBN 5‑902367‑03‑4.
Grishachev V.V., Kazarin O. V., Kalinina YU.D. Fizicheskaya model’ ugrozy utechki akusticheskoj (rechevoj) informacii cherez volokonno-opticheskie kommunikacii. Voprosy zashchity informacii. 2018; 3: 35–51.
Grishachev V.V., Zabolockaya A. D. Problema informacionnoj bezopasnosti volokonno-opticheskikh tekhnologij. Photonics Russia. 2022; 16(6): 484–500. DOI: 10.22184/1993‑7296.FRos.2022.16.6.484.500.
Grishachev V.V., Kalinina YU.D., Tarasov A. A. Ocenka glubiny parazitnoj modulyacii sveta v opticheskoj kabel’noj sisteme s neodnorodnostyami. Voprosy zashchity informacii. 2016; 1: 62–73.
Grishachev V.V., Kazarin O. V., Kalinina YU.D. Parazitnye akusticheskie modulyacii svetovykh potokov v raz’emnykh soedineniyakh opticheskoj seti ob’ekta informatizacii. Voprosy zashchity informacii. 2018; 4: 47–56.
AUTHOR
Grishachev Vladimir V., Cand. of Sc.(Phys.&Math.), associate professor, Institute for Information Sciences and Security Technologies (IISST), Russian State University of the Humanities (RSUH), e-mail: grishachev@mail.ru, Moscow, Russia
ORCID: 0000-0002-7585-7282
Readers feedback
