Application of Dual-Frequency Self-Injection Locked DFB Laser for Brillouin Optical Time Domain Analysis

Self-injection locking to an external fiber cavity is an efficient technique enabling drastic linewidth narrowing of semiconductor lasers. Recently, we constructed a simple dual-frequency laser source that employs self-injection locking of a DFB laser in the external ring fiber cavity and Brillouin lasing in the same cavity. The laser performance characteristics are on the level of the laser modules commonly used with BOTDA. The use of a laser source operating two frequencies strongly locked through the Brillouin resonance simplifies the BOTDA system, avoiding the use of a broadband electrooptical modulator (EOM) and high-frequency electronics. Here, in a direct comparison with the commercial BOTDA, we explore the capacity of our low-cost solution for BOTDA sensing, demonstrating distributed measurements of the Brillouin frequency shift in a 10 km sensing fiber with a 1.5 m spatial resolution.


Introduction
Distributed optical fiber sensors show superior advantages over their electronic counterparts due to their high sensitivity to external disturbances and low loss transmission, which are important for remote sensing [1]. Unlike single-point optical fiber sensors [2,3], distributed optical fiber sensors can interrogate and spatially resolve measurands along an optical fiber due to their specific sensing mechanisms [4][5][6][7]. Among them, Brillouin-based distributed sensors [8][9][10] have attracted immense interest in recent years in fields such as the health monitoring of large structures in oil and gas pipelines [11], railways and highvoltage transmission lines [12], high-temperature distributed measurement in industrial applications [13], distributed strain measurement for cracks detection [14], and structural health monitoring [15]. The conventional distributed Brillouin optical fiber sensing is based on the backward-stimulated Brillouin scattering, where the strain or temperature is a linear function of the Brillouin frequency shift (BFS) and so can be recovered from the distribution of Brillouin gain spectra (BGS) along the sensing fiber [16,17]. Over the past two decades, many efforts have been devoted to improving its performance, including spatial resolution, measurement time, and sensing range [18][19][20]. Although the Brillouin sensing instruments have become commercially available, their relatively high cost remains the major critical factor limiting the range of their potential applications. A key and most expensive part of the traditional BOTDA system is a master-oscillator module employed for generation and tuning of the pump and Stokes signal frequencies. Commonly, such modules implement phase-locked loop (PLL) [21] or optical side-band (OSB) generation techniques [22]. In the PLL technique, two narrowband laser sources are used to generate pump and Stokes laser frequencies. The frequency of one laser is locked to the frequency of another laser through outputs; the spectra are centered at 1531. 13 and 1531.21 nm, respectively. The spectrum widths are determined by the resolution (~0.01 nm) of the spectrum analyzer. The difference of~0.08 nm between the peak wavelengths corresponds to a Brillouin frequency shift (BFS) of~11 GHz. A weak Rayleigh scattering signal at the pump frequency was also observed through the Stokes laser output. natural Lorenz linewidths of ~270 and ~110 Hz, and powers of ~9 mW and ~100μW for pump and Stokes outputs, respectively. The laser optical spectra shown in Figure 2 demonstrate the contrast of ~70 dB between the peak and background powers for both laser outputs; the spectra are centered at 1531. 13 and 1531.21 nm, respectively. The spectrum widths are determined by the resolution (~0.01 nm) of the spectrum analyzer. The difference of ~0.08 nm between the peak wavelengths corresponds to a Brillouin frequency shift (BFS) of ~11 GHz. A weak Rayleigh scattering signal at the pump frequency was also observed through the Stokes laser output. The beating between two laser outputs is characterized by a stable radio-frequency (RF) spectrum with 290 Hz -linewidth and fixed peak frequency at , as presented in Figure 3a. The peak RF spectrum corresponds to the Brillouin frequency shift in SMF-28 fiber (Corning, Inc., Corning, NY, USA) used in the laser cavity at 1535 nm. To reduce the effect of the environmental noise, a spliced laser configuration was placed into a thermostabilized (∼25 °C) foam box. Additional thermal control was applied to the laser box as a whole and used to keep the feedback circuit within its dynamic range ensuring long-term laser operation stability. The drift in the RF beat recorded each minute for 1 h is shown in Figure 3b. Variations in the RF spectrum peak frequency within a limited range ( RF δν < 5 kHz) reflects the effect of the thermal control. No mode hopping event was observed when the RF beat frequency was within this range. Details on the laser operation and its performance characteristics are provided in Appendix A. natural Lorenz linewidths of ~270 and ~110 Hz, and powers of ~9 mW and ~100μW for pump and Stokes outputs, respectively. The laser optical spectra shown in Figure 2 demonstrate the contrast of ~70 dB between the peak and background powers for both laser outputs; the spectra are centered at 1531.13 and 1531.21 nm, respectively. The spectrum widths are determined by the resolution (~0.01 nm) of the spectrum analyzer. The difference of ~0.08 nm between the peak wavelengths corresponds to a Brillouin frequency shift (BFS) of ~11 GHz. A weak Rayleigh scattering signal at the pump frequency was also observed through the Stokes laser output. The beating between two laser outputs is characterized by a stable radio-frequency (RF) spectrum with 290 Hz -linewidth and fixed peak frequency at , as presented in Figure 3a. The peak RF spectrum corresponds to the Brillouin frequency shift in SMF-28 fiber (Corning, Inc., Corning, NY, USA) used in the laser cavity at 1535 nm. To reduce the effect of the environmental noise, a spliced laser configuration was placed into a thermostabilized (∼25 °C) foam box. Additional thermal control was applied to the laser box as a whole and used to keep the feedback circuit within its dynamic range ensuring long-term laser operation stability. The drift in the RF beat recorded each minute for 1 h is shown in Figure 3b. Variations in the RF spectrum peak frequency within a limited range ( RF δν < 5 kHz) reflects the effect of the thermal control. No mode hopping event was observed when the RF beat frequency was within this range. Details on the laser operation and its performance characteristics are provided in Appendix A. The beating between two laser outputs is characterized by a stable radio-frequency (RF) spectrum with 290 Hz -linewidth and fixed peak frequency at ν L0 − ν S0 = ∆ν 0 ∼ 10.946 GHz, as presented in Figure 3a. The peak RF spectrum corresponds to the Brillouin frequency shift in SMF-28 fiber (Corning, Inc., Corning, NY, USA) used in the laser cavity at 1535 nm. To reduce the effect of the environmental noise, a spliced laser configuration was placed into a thermostabilized (∼25 • C) foam box. Additional thermal control was applied to the laser box as a whole and used to keep the feedback circuit within its dynamic range ensuring long-term laser operation stability. The drift in the RF beat recorded each minute for 1 h is shown in Figure 3b. Variations in the RF spectrum peak frequency within a limited range (δν RF < 5 kHz) reflects the effect of the thermal control. No mode hopping event was observed when the RF beat frequency was within this range. Details on the laser operation and its performance characteristics are provided in Appendix A. The BOTDA built in the laboratory is a simple modification of the traditional pumpprobe setup [35]. The laser radiation at the pump frequency 0 L ν passes through the electro-optical intensity modulator (EOM 1) and erbium-doped fiber amplifier (EDFA) to form a periodic train of rectangular pulses with the peak power of ∼300 mW, pulse duration of ∼15 ns, and repetition rate of ~10 kHz. The laser radiation emitted at the Stokes frequency introduced into the fiber under test from the opposite fiber ends. Their interaction through the Brillouin process in the fiber under test caused an energy transfer from the pump pulse to the CW Stokes signal, leading to its intensity modulation recorded at the fiber output by the fast photodetector and PC acquisition card. The modulation amplitude (typically <100 nW) of the probe signal is proportional to the local Brillouin gain at the fiber point where Brillouin resonance is achieved. The recorded probe signal traces were averaged over 4096 pump pulses and used to map the distribution of the Brillouin gain over the fiber length. The traces recorded at different S δν were used to build the Brillouin gain spectrum at each fiber point. Then, these data were mathematically processed to find the position of the Brillouin spectrum peak, i.e., the BFS, in each fiber point. The pump pulse duration of ~15 ns sets the system spatial resolution of ~1.5 m. Notably, the use of the laser source operating two frequencies strongly locked through the Brillouin resonance as a master-oscillator in our BOTDA setup allowed replacing the broadband EOM used in [35] for the frequency shift by ~11 GHz and all high-frequency electronics by more cost-efficient and straightforward counterparts operating in the sub-gigahertz radio-frequency range.
To evaluate the performance of the laser operation with the built BOTDA, we built a fiber testing line similar to that commonly used with BOTDAs for their calibration. The 10 km length testing line shown in Figure 4 comprises two lengths of SMF-28 Corning fiber (9.1 and 1 km) and a length of OFS (G.657) fiber (~50 m) placed between them. The latter includes eleven altered fiber coils of different lengths (0.75-6.0 m) kept at different temperatures. Odd coils were placed into a heat chamber thermostabilized at 60 °C, while even coils were rested at room temperature (~25 °C). The measured distribution of the BFS over the OFS fiber followed the temperature fiber profile with a factor of ~1 MHz/°C. The BOTDA built in the laboratory is a simple modification of the traditional pumpprobe setup [35]. The laser radiation at the pump frequency ν L0 passes through the electro-optical intensity modulator (EOM 1) and erbium-doped fiber amplifier (EDFA) to form a periodic train of rectangular pulses with the peak power of ∼300 mW, pulse duration of ∼15 ns, and repetition rate of~10 kHz. The laser radiation emitted at the Stokes frequency ν S0 passes through the dual-drive electro-optic modulator (EOM 2, SSB-CS EOM, Sumitomo), variable optical attenuator (VOA), and polarization mode scrambler (PMS) to form a~10 µW CW probe signal at the frequency ν S = ν S0 − δν S . The frequency tuning of the probe signal is provided by a tunable radio-frequency generator supplying EOM 2 at δν S < 1 GHz. The pump pulses at ν L0 and the CW probe signal at ν S were introduced into the fiber under test from the opposite fiber ends. Their interaction through the Brillouin process in the fiber under test caused an energy transfer from the pump pulse to the CW Stokes signal, leading to its intensity modulation recorded at the fiber output by the fast photodetector and PC acquisition card. The modulation amplitude (typically <100 nW) of the probe signal is proportional to the local Brillouin gain at the fiber point where Brillouin resonance is achieved. The recorded probe signal traces were averaged over 4096 pump pulses and used to map the distribution of the Brillouin gain over the fiber length. The traces recorded at different δν S were used to build the Brillouin gain spectrum at each fiber point. Then, these data were mathematically processed to find the position of the Brillouin spectrum peak, i.e., the BFS, in each fiber point. The pump pulse duration of~15 ns sets the system spatial resolution of~1.5 m. Notably, the use of the laser source operating two frequencies strongly locked through the Brillouin resonance as a master-oscillator in our BOTDA setup allowed replacing the broadband EOM used in [35] for the frequency shift by~11 GHz and all high-frequency electronics by more cost-efficient and straightforward counterparts operating in the sub-gigahertz radio-frequency range.
To evaluate the performance of the laser operation with the built BOTDA, we built a fiber testing line similar to that commonly used with BOTDAs for their calibration. The 10 km length testing line shown in Figure 4 comprises two lengths of SMF-28 Corning fiber (9.1 and 1 km) and a length of OFS (G.657) fiber (~50 m) placed between them. The latter includes eleven altered fiber coils of different lengths (0.75-6.0 m) kept at different temperatures. Odd coils were placed into a heat chamber thermostabilized at 60 • C, while even coils were rested at room temperature (~25 • C). The measured distribution of the BFS over the OFS fiber followed the temperature fiber profile with a factor of~1 MHz/ • C.

Results
The testing experiments were performed with our BOTDA setup and then repeated with the commercial BOTDA system (OZ-optics) set to operate with a pump pulse of 15 ns. Note that the operating wavelength of the commercial BOTDA is 1550 nm, so the BFS data measured with this device are presented here with a scaling factor of 1550/1535. The experimental results are shown in Figures 5 and 6. First, we adjusted the frequency difference between the pump and probe signal 0 S ν δν Δ + to maximize the signal recorded from the first length of the SMS-28 fiber. In this case, 0 S δν ≈ , since the frequency difference of the two signals emitted by the laser corresponds to the BFS in the SMS-28 fiber at room temperature (~10.946 GHz ). Figure 5a compares the traces recorded with our setup and the commercial BOTDA: both traces exhibit similar behavior, including the gain modulation pronounced in the fiber points with an alternating temperature. The drop in both traces after 9.2 km is explained by pump power loss after splicing. Figure 5b provides more details on the Brillouin gain distribution along the OFS fiber. Both systems wellrecognized the fiber segments where the fiber temperature was higher than the room temperature. Since the BFS increases with temperature, the BFS in hot OFS fiber segments is closer to the frequency difference set by the system, leading to an increase in the local Brillouin gain measured at these points.
(a) (b) In order to accurately estimate the BFS distribution along the sensing fiber, the Brillouin gain spectrum (BGS) was measured at each fiber location. At each fiber point, the radio frequency S δν applied to the dual-drive EOM 2 was swept within the range of 100 MHz with a step of ~2 MHz. Then, the BFS was reconstructed by fitting the recorded data with the Lorentzian profile [17]. Figure 6a compares the BGSs at two OFS fiber points located at 9151 and 9057 m measured with our setup and the commercial BOTDA. Our system well-reproduced the BGS shape recorded with the commercial BOTDA. Notably, the BGS corresponding to the fiber points exposed to the temperature ~25 and ~60 °C are

Results
The testing experiments were performed with our BOTDA setup and then repeated with the commercial BOTDA system (OZ-optics) set to operate with a pump pulse of 15 ns. Note that the operating wavelength of the commercial BOTDA is 1550 nm, so the BFS data measured with this device are presented here with a scaling factor of 1550/1535. The experimental results are shown in Figures 5 and 6. First, we adjusted the frequency difference between the pump and probe signal ∆ν 0 + δν S to maximize the signal recorded from the first length of the SMS-28 fiber. In this case, δν S ≈ 0, since the frequency difference of the two signals emitted by the laser corresponds to the BFS in the SMS-28 fiber at room temperature (∼ 10.946 GHz). Figure 5a compares the traces recorded with our setup and the commercial BOTDA: both traces exhibit similar behavior, including the gain modulation pronounced in the fiber points with an alternating temperature. The drop in both traces after 9.2 km is explained by pump power loss after splicing. Figure 5b provides more details on the Brillouin gain distribution along the OFS fiber. Both systems well-recognized the fiber segments where the fiber temperature was higher than the room temperature. Since the BFS increases with temperature, the BFS in hot OFS fiber segments is closer to the frequency difference set by the system, leading to an increase in the local Brillouin gain measured at these points.

Results
The testing experiments were performed with our BOTDA setup and then repeated with the commercial BOTDA system (OZ-optics) set to operate with a pump pulse of 15 ns. Note that the operating wavelength of the commercial BOTDA is 1550 nm, so the BFS data measured with this device are presented here with a scaling factor of 1550/1535. The experimental results are shown in Figures 5 and 6. First, we adjusted the frequency difference between the pump and probe signal  Figure 5b provides more details on the Brillouin gain distribution along the OFS fiber. Both systems wellrecognized the fiber segments where the fiber temperature was higher than the room temperature. Since the BFS increases with temperature, the BFS in hot OFS fiber segments is closer to the frequency difference set by the system, leading to an increase in the local Brillouin gain measured at these points.
(a) (b) In order to accurately estimate the BFS distribution along the sensing fiber, the Brillouin gain spectrum (BGS) was measured at each fiber location. At each fiber point, the radio frequency S δν applied to the dual-drive EOM 2 was swept within the range of 100 MHz with a step of ~2 MHz. Then, the BFS was reconstructed by fitting the recorded data with the Lorentzian profile [17]. Figure 6a compares the BGSs at two OFS fiber points located at 9151 and 9057 m measured with our setup and the commercial BOTDA. Our system well-reproduced the BGS shape recorded with the commercial BOTDA. Notably, the BGS corresponding to the fiber points exposed to the temperature ~25 and ~60 °C are In order to accurately estimate the BFS distribution along the sensing fiber, the Brillouin gain spectrum (BGS) was measured at each fiber location. At each fiber point, the radio frequency δν S applied to the dual-drive EOM 2 was swept within the range of 100 MHz with a step of~2 MHz. Then, the BFS was reconstructed by fitting the recorded data with the Lorentzian profile [17]. Figure 6a compares the BGSs at two OFS fiber points located at 9151 and 9057 m measured with our setup and the commercial BOTDA. Our system well-reproduced the BGS shape recorded with the commercial BOTDA. Notably, the BGS corresponding to the fiber points exposed to the temperature~25 and~60 • C are distant by 35 MHz. Figure 6b shows the distributions of the BFS over the OFS fiber: the quality of BSF restoration is rather high. For the fiber segments longer than 1.5 m, both devices produced almost the same distribution, providing a good agreement in absolute Sensors 2021, 21, 6859 6 of 12 peak frequencies varying from 10,880 GHz (at~25 • C) to 10,915 GHz (at~60 • C), resolved slopes, and segment positions. All specific fiber segments were well-recognized. For the fiber segment shorter than 1.5 m, our experimental device demonstrated even higher accuracy than the commercial BOTDA operating with the same pulse duration.  Figure 6b shows the distributions of the BFS over the OFS fiber: the quality of BSF restoration is rather high. For the fiber segments longer than 1.5 m, both devices produced almost the same distribution, providing a good agreement in absolute peak frequencies varying from 10,880 GHz (at ~25 °C) to 10,915 GHz (at ~60 °C), resolved slopes, and segment positions. All specific fiber segments were well-recognized. For the fiber segment shorter than 1.5 m, our experimental device demonstrated even higher accuracy than the commercial BOTDA operating with the same pulse duration.

Conclusions
In conclusion, we used a simple dual-frequency laser configuration operated with BOTDA. The use of the laser operating two fixed frequencies significantly simplified the BOTDA system and allows excluding the broadband EOM (and high-power high-frequency microwave generator supplying the EOM) from the traditional BOTDA configuration based on a single narrowband laser source. These relatively high-cost devices constitute a significant part of the BOTDA system value. Additionally, the laser operation mechanism enabling self-stabilized generation of the Brillouin wave does not employ rather an expensive wavelength locking circuit, being a part of the Brillouin lasers offered for BOTDA sensing earlier [27]. Instead, a dual-drive EOM driven by a standard 1 GHz radio-frequency generator is employed in our home BOTDA setup. Moreover, the tuning range of 100 MHz was enough to perform the measurements shown in Figure 6. Under control of the commercial BOTDA, we evaluated the capacity of a low-cost laser solution to operate with the BOTDA sensing, demonstrating the distributed measurements of the BFS in a 10 km sensing fiber with a 1.5 m spatial resolution. No deterioration of the system performance characteristics associated with the use of the dual-frequency laser was found during the measurements. Further research will be directed to design and testing of new laser sources [36] for advanced sensor applications.

Conclusions
In conclusion, we used a simple dual-frequency laser configuration operated with BOTDA. The use of the laser operating two fixed frequencies significantly simplified the BOTDA system and allows excluding the broadband EOM (and high-power high-frequency microwave generator supplying the EOM) from the traditional BOTDA configuration based on a single narrowband laser source. These relatively high-cost devices constitute a significant part of the BOTDA system value. Additionally, the laser operation mechanism enabling self-stabilized generation of the Brillouin wave does not employ rather an expensive wavelength locking circuit, being a part of the Brillouin lasers offered for BOTDA sensing earlier [27]. Instead, a dual-drive EOM driven by a standard 1 GHz radio-frequency generator is employed in our home BOTDA setup. Moreover, the tuning range of 100 MHz was enough to perform the measurements shown in Figure 6. Under control of the commercial BOTDA, we evaluated the capacity of a low-cost laser solution to operate with the BOTDA sensing, demonstrating the distributed measurements of the BFS in a 10 km sensing fiber with a 1.5 m spatial resolution. No deterioration of the system performance characteristics associated with the use of the dual-frequency laser was found during the measurements. Further research will be directed to design and testing of new laser sources [36] for advanced sensor applications.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A. Dual-Frequency Self-Injection-Locked DFB Laser
This section contains details and supplementary data on the low-cost dual-frequency master-oscillator described earlier in [33] and used in these studies with the BOTDA setup shown in Figure 1. A simple laser configuration was spliced from standard telecom components. It employed a single DFB laser coupled to an all-fiber ring cavity. In this configuration, the same ring fiber cavity is used both for self-injection locking of the DFB laser and for generation of Stokes light through the Brillouin lasing. A low-cost USB-DAQ was used to stabilize the system, preventing mode-hopping. Specifically, a narrowband laser operation was provided by the self-injection locking mechanism, whereas the active feedback circuit just helped the laser maintain the desired operation. Additionally, the selfinjection locking supported permanent coupling between the DFB laser and the external fiber ring cavity, enabling an efficient resonant pumping for low-noise Brillouin wave generation. The ability to emit light from a single DFB laser at two frequencies strongly locked to the Brillouin resonance significantly simplifies the traditional BOTDA setup, in particular, avoiding the use of broadband electrooptical modulators and related electronic circuits.
Funding: This research was funded By the Ministry of Science and Higher Education of the Russian Federation, grant number 075-15-2021-581 and the Russian Science Foundation, grant number 18-12-00457P.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A. Dual-Frequency Self-Injection-Locked DFB Laser
This section contains details and supplementary data on the low-cost dual-frequency master-oscillator described earlier in [33] and used in these studies with the BOTDA setup shown in Figure 1. A simple laser configuration was spliced from standard telecom components. It employed a single DFB laser coupled to an all-fiber ring cavity. In this configuration, the same ring fiber cavity is used both for self-injection locking of the DFB laser and for generation of Stokes light through the Brillouin lasing. A low-cost USB-DAQ was used to stabilize the system, preventing mode-hopping. Specifically, a narrowband laser operation was provided by the self-injection locking mechanism, whereas the active feedback circuit just helped the laser maintain the desired operation. Additionally, the selfinjection locking supported permanent coupling between the DFB laser and the external fiber ring cavity, enabling an efficient resonant pumping for low-noise Brillouin wave generation. The ability to emit light from a single DFB laser at two frequencies strongly locked to the Brillouin resonance significantly simplifies the traditional BOTDA setup, in particular, avoiding the use of broadband electrooptical modulators and related electronic circuits.

. Experimental Laser Configuration
The experimental configuration of the dual-frequency laser is shown in Figure A1. The semiconductor laser was a commercial distributed feedback (DFB) laser diode (MITSUBISHI FU-68PDF-V520M27B, Mitsubishi Electric Corporation) delivering ~15 mW at ~1535 nm. It is assembled within a standard 14-pin butterfly package. The DFB laser is coupled to the fiber ring cavity through a circulator. The high-Q external ring cavity is spliced from two (99/1) couplers. Totally, it comprises a 20 m length of SMS-28 fiber (Corning, Inc.) perfectly adjusted for lasing at the Stokes Brillouin frequency [37]. To implement the self-injection locking, the first coupler provides passive feedback to the laser operation, redirecting a part of the light circulating clockwise (CW) inside the cavity through the circulator back into the DFB laser cavity. Polarization controller 1 is used to adjust the polarization state of the light inside the fiber ring cavity, enabling an optimal coupling between the laser radiation and one of the ring cavity modes. Polarization controller 2 adjusts the polarization state of the light before its injection into the DFB laser emitting a

. Experimental Laser Configuration
The experimental configuration of the dual-frequency laser is shown in Figure A1. The semiconductor laser was a commercial distributed feedback (DFB) laser diode (MIT-SUBISHI FU-68PDF-V520M27B, Mitsubishi Electric Corporation) delivering~15 mW at 1535 nm. It is assembled within a standard 14-pin butterfly package. The DFB laser is coupled to the fiber ring cavity through a circulator. The high-Q external ring cavity is spliced from two (99/1) couplers. Totally, it comprises a 20 m length of SMS-28 fiber (Corning, Inc.) perfectly adjusted for lasing at the Stokes Brillouin frequency [37]. To implement the self-injection locking, the first coupler provides passive feedback to the laser operation, redirecting a part of the light circulating clockwise (CW) inside the cavity through the circulator back into the DFB laser cavity. Polarization controller 1 is used to adjust the polarization state of the light inside the fiber ring cavity, enabling an optimal coupling between the laser radiation and one of the ring cavity modes. Polarization controller 2 adjusts the polarization state of the light before its injection into the DFB laser emitting a linear polarization, thus controlling the optical feedback strength. A piezo fiber-stretcher (PFS, Evanescent Optics Inc., Model 915B) is attached to the feedback loop. It is used as an optical phase shitter driven by a low-cost USB Multifunction DAQ (National Instrument NI USB-6009). The DAQ is connected with a PC and helps the laser maintain the desired operation in self-injection locking regime. The fiber configuration is spliced from standard telecom components and placed into a foam box stabilized at~25 • C, thus protecting the laser system from the laboratory environment. A fast photodetector detects the optical signal at port A, which is used as an error signal for operation of the active feedback circuit. The active feedback circuit tends to keep it as low as possible, providing an appropriate voltage to the piezo fiber-stretcher. Additional thermal control is applied to the laser box as a whole and used to keep the voltage applied to the piezo-stretcher within its dynamic range.

Operation and Stabilization Mechanisms
The mechanism of the laser operation and stabilization is discussed in [33,34,36,38]. A simplified view on the laser operation is depicted in Figure A2. The light emitted by a DFB laser passes the fiber ring cavity and optical feedback loop and is injected back into the DFB laser cavity, forcing the DFB laser to emit at ν L = ν R , where ν R is one of the ring cavity resonant frequencies. For stable laser operation, the laser frequency ν L should also be resonant for the coupled cavity comprising the DFB laser cavity and feedback loop ν FB+LD , so ν L = ν R = ν LD+FB . The position of the frequency ν LD+FB with respect to ν R is controlled by the piezo-stretcher through the applied voltage. The laser operation becomes unstable when the environment noise affecting the laser configuration fibers violates the equality ν R = ν LD+FB . The task of the active feedback circuit is to maintain the equality ν R = ν LD+FB , minimizing the error signal detected at port A. In this case, the laser power at ν L = ν R is perfectly coupled to the fiber ring cavity and enables effective backward lasing at Stokes frequency ν S = ν L − ∆ SBS , where ∆ SBS is the Brillouin frequency shift. With the well-adjusted fiber ring cavity length, the Stokes frequency coincides with one of the resonant ring resonances. Therefore, the perfect stabilization of the lasing at ν L = ν R = ν LD+FB ensures stabilization of lasing at frequency ν S . linear polarization, thus controlling the optical feedback strength. A piezo fiber-stretcher (PFS, Evanescent Optics Inc., Model 915B) is attached to the feedback loop. It is used as an optical phase shitter driven by a low-cost USB Multifunction DAQ (National Instrument NI USB-6009). The DAQ is connected with a PC and helps the laser maintain the desired operation in self-injection locking regime. The fiber configuration is spliced from standard telecom components and placed into a foam box stabilized at ~25 °C, thus protecting the laser system from the laboratory environment. A fast photodetector detects the optical signal at port A, which is used as an error signal for operation of the active feedback circuit. The active feedback circuit tends to keep it as low as possible, providing an appropriate voltage to the piezo fiber-stretcher. Additional thermal control is applied to the laser box as a whole and used to keep the voltage applied to the piezo-stretcher within its dynamic range.

Operation and Stabilization Mechanisms
The mechanism of the laser operation and stabilization is discussed in [33,34,36,38]. A simplified view on the laser operation is depicted in Figure A2  In accordance with our observations, there are two cases when mode-hopping in the laser occurs. First, a mechanical vibration such as a pencil knock on the laser box that causes short-time perturbations of the laser power (see Figure A3). We used a foam box to protect the laser from mechanical vibrations. Second is an extended temperature drift. While the active feedback circuit operates to counteract temperature drift, the voltage applied to the piezo-stretcher could exhaust its dynamic range (~40 rad ). In this case, for In accordance with our observations, there are two cases when mode-hopping in the laser occurs. First, a mechanical vibration such as a pencil knock on the laser box that causes short-time perturbations of the laser power (see Figure A3). We used a foam box to protect the laser from mechanical vibrations. Second is an extended temperature drift. While the active feedback circuit operates to counteract temperature drift, the voltage applied to the piezo-stretcher could exhaust its dynamic range (∼ 40 rad). In this case, for further operation, the phase delay induced by the piezo-stretcher should be reset by an integer number of 2π. Such a jump destabilizes the laser in the same way as the pencil kick. To avoid destabilization, the working point of the piezo-stretcher should be always maintained within its dynamic range. In our setup, this was implemented through the additional thermal control applied to the laser box and using the voltage applied to the piezo-stretcher as an error signal. further operation, the phase delay induced by the piezo-stretcher should be reset by an integer number of 2π . Such a jump destabilizes the laser in the same way as the pencil kick. To avoid destabilization, the working point of the piezo-stretcher should be always maintained within its dynamic range. In our setup, this was implemented through the additional thermal control applied to the laser box and using the voltage applied to the piezo-stretcher as an error signal. for the pump and Stokes laser outputs shown in Figure A1, respectively. The experimental dependence of the laser power at the Stokes frequency (Stokes output) on the laser power at the pump frequency (pump output) is shown in Figure A3. The pump power threshold for the Stokes frequency generation is ~2.9 mW. The pump-to-Stokes power conversion efficiency above the threshold is ~3.3%.

Appendix A.2.2. Laser Oscilloscope Traces and Stability against Perturbations
The typical laser oscilloscope traces recorded at pump output, Stokes output, optical error signal (port A), and the electrical feedback signal are shown in Figure A4. The recorded traces provide insight into the system stability, demonstrating the system response to a pencil knock on the spliced fiber setup. Due to the active feedback circuit, the power at port A (used as the error signal) always tends to its minimal value, thus enabling stable laser operation at two frequencies.  The maximal laser output powers emitted by the laser are ∼ 9 mW and ∼ 100 µW for the pump and Stokes laser outputs shown in Figure A1, respectively. The experimental dependence of the laser power at the Stokes frequency (Stokes output) on the laser power at the pump frequency (pump output) is shown in Figure A3. The pump power threshold for the Stokes frequency generation is~2.9 mW. The pump-to-Stokes power conversion efficiency above the threshold is~3.3%.

Appendix A.2.2. Laser Oscilloscope Traces and Stability against Perturbations
The typical laser oscilloscope traces recorded at pump output, Stokes output, optical error signal (port A), and the electrical feedback signal are shown in Figure A4. The recorded traces provide insight into the system stability, demonstrating the system response to a pencil knock on the spliced fiber setup. Due to the active feedback circuit, the power at port A (used as the error signal) always tends to its minimal value, thus enabling stable laser operation at two frequencies. All unperturbed optical power traces are almost flat and exhibit no fluctuations. When the configuration was perturbed by the pencil knock, the traces exhibited dynamic behavior. The optical error signal makes a few stochastic fluctuations and then stabilizes to the original level. The time constants of the feedback mechanism are τ L ∼ 0.2 s and τ S ∼ 0.05 s for the pump and Stokes outputs, respectively. The self-heterodyne spectra [39,40] of the pump and Stokes laser radiation averaged Figure A4. Typical oscilloscope traces of the optical error signal (port A), pump, and Stokes output powers and electrical feedback signal, highlighting the response to a pencil knock on the laser box.
The self-heterodyne spectra [39,40] of the pump and Stokes laser radiation averaged over 10 measurements each are shown in Figure A5. The measurements were performed with a fiber Mach-Zehnder interferometer (comprising the 55 km delay fiber and 25 MHz phase modulator), ~5 GHz photodetector, and RF spectrum analyzer (FSH8, Rohde & Schwarz). The recorded spectra were fitted by the Voigt profile and decomposed into the Gaussian and Lorentzian contributions [40,41]. The fitting parameter L w relates to the laser Lorentzian linewidth as 1 2 L w , resulting in the measured values of 270Hz  and 110Hz  for pump and Stokes laser outputs, respectively [40]. The power spectral density (PSD) of phase noise measured for the pump and Stokes radiations with an RF spectrum analyzer (Agilent N9320A) in the range of 10 100 kHz − is shown in Figure A6a. The measurements were performed by the self-heterodyne method [42][43][44] with the Mach-Zehnder interferometer comprising a ~1.3 km delay fiber (~5.76μs ) and 20 MHz frequency modulator. In comparison with the free-running Figure A5. The self-heterodyne spectra of the pump (a) and Stokes (b) laser radiation: the measured spectra (black) and their fitting by the Voigt functions (blue). The fitting parameters evaluating the Gaussian and Lorentzian contributions are w G = 2.2 kHz, w L = 540 Hz and w G = 1.7 kHz, w L = 220 Hz, for the pump and Stokes outputs, respectively. Appendix A.2.5. Noise Performance The power spectral density (PSD) of phase noise measured for the pump and Stokes radiations with an RF spectrum analyzer (Agilent N9320A) in the range of 10 − 100 kHz is shown in Figure A6a. The measurements were performed by the self-heterodyne method [42][43][44] with the Mach-Zehnder interferometer comprising a ∼ 1.3 km delay fiber (∼ 5.76 µs) and 20 MHz frequency modulator. In comparison with the free-running laser, the measured PSD is lower by ∼ (35-40) and ∼ (50-60) dBc/Hz for the pump and Stokes laser outputs, respectively. The relative intensity noise (RIN) measured with a lock-in amplifier SRS510 in the range of 1-100 kHz is shown in Figure A6b. In comparison with the free-running laser, the measured RIN is lower by ∼ (5-10) dB and higher by 30-40 dB for the pump and Stokes laser outputs, respectively.