1. Introduction

The rapid developments in optical communications, wavelength division multiplexing lightwave systems, optical fiber sensing and component testing have sped-up the research progress on cascaded multiwavelength generation [1]. In this field, the main focuses of interest are to achieve uniform Stokes lines with wider spacing [2,3] over a broader multiwavelength bandwidth (MBW) [4–6] with excellent optical-to-signal ratio (OSNR) [7]. One of the efficient methods of realizing this type of lasing is by combining stimulated Brillouin scattering (SBS) with stimulated Raman scattering in a nonlinear gain medium. The selection of Brillouin-Raman transition relates to its favorable property of inhomogeneous broadening. This justifies the controllable gain competition between the Brillouin lines with the self-lasing modes. Amongst the advantageous nature of multiwavelength Brillouin-Raman fiber laser (MBRFL) includes simple designs, high stability performances, ease of use and enhanced flatness with wide gain bandwidth. The key point to realize the latter criteria begins from better exploitation of multiple Rayleigh scattering that initiates random distributed feedback [8]. By including this, the hybrid structure of MBRFL has been very well established in 10 GHz [4–6,9,10] and 20 GHz [10,11] spacing.

In a single-wavelength spacing MBRFL, 57 to 62 nm MBWs are attained which correspond to remarkably 715 to 798 Stokes lines [4,5]. These necessitate the implementation of three pumping structures to provide nearly 5 to 10 W pump power levels. Although outstanding results are obtained, one of the drawbacks is the lower OSNR around 10 dB-order. Beside this, further advancements by introducing more device simplification are needed to compete with the complicated pumping architecture of these predecessors. By utilizing only a single Raman pump unit at 1.36 W power, the widest MBW of 40 nm is generated [6]. Up to 500 Stokes waves are produced where the OSNR varies from 12.5 to 16.5 dB approximation at different wavelength segment. Interestingly, all of these characterizations are performed by counting all lasing lines within 3-dB peak power discrepancy [4–6]. Despite of these, another attempt demonstrates that almost 460 cascaded Stokes waves contained in the 37 nm MBW are measured with 2.3 dB peak power difference [9]. A lower pumping level around 1 W is utilized when the passive gain medium is served by a strand of dispersion compensating fiber (DCF). In contrast to [6], the average OSNR is maintained at 16.8 dB estimation over the entire bandwidth regime. This is done with the suppression of turbulent wave broadening that is realized through the incorporation of a large effective area fiber.

When looking towards double-wavelength spacing MBRFL built in a ring cavity, 12 to 16 Stokes lines are produced [12,13]. These pertain to non-flat spectra of 10 dB and more than 15 dB respectively. Further reduction in spectral flatness to 3-dB span yields only 20 higher-order Stokes lines [14] which implies the physical limitation in this ordinary layout. Nevertheless by utilizing a fully open cavity [8], the problem can be addressed. The growth of over 200 Stokes numbers are accomplished with an average OSNR of 24 [10] and 27.5 dB [11]. The first represents 32.5 nm MBW that is estimated at 5 dB peak power variation. This is further truncated to 2.5 dB variation that covers 29 nm MBW when the Raman pumping is divided symmetrically at forward and backward directions. For a brief review, the extremely improved flatness to 1.8 dB is realized together with an unprecedented ultrahigh OSNR of nearly 47 dB [7]. This extraordinary attainment in a multiwavelength Brillouin random fiber laser is reversed by the generation of only six-orders Stokes emission. As a result, this accelerates the continuing exploration to search for a better technology to counter-balance the spectral flatness achievement with wider MBW and reasonable OSNR.

One of the alternative solutions is by introducing a micro-air gap that has many important benefits. In a pioneering report [15], the structure induces additional Fresnel reflectivities that assist in decreasing multiple higher-order Stokes wave thresholds. This favors more cascaded narrow line formation where the multiwavelength amplitude is extremely flattened. Another role of modifying the micrometer air-gap length is to introduce a simple way for switchable frequency spacing [16]. Around 5 to 6 Stokes emissions are produced during the conversion from double- to single-wavelength spacing, respectively. Furthermore in 10 GHz, multiwavelength Brillioun-Erbium fiber lasers, the numbers of Stokes channels are reduced from 200 [17] to 180 [18] at separate optimized pumping parameters. These are recompensated by the smoothing of the spectral envelope feature. The former attempt that has no micro-air gap demonstrates a channel flatness of 4.65 dB is improved to 3-dB span in the latter assessment that includes this scheme. The results obtained are followed also by the increase in average OSNR from 15 to 18 dB, correspondingly. Therefore to further upgrade the flatness in MBW, the air-gap (AG) is embedded externally from the cavity of our proposed double-wavelength spacing MBRFL. In contrast to the micrometer length utilization in the previous reports [15,16,18], the sub-millimeter length separation, $L_{mm}$ from 0.05 to 1.00 mm series is used. The concept is similar to implementing an artificial optical mirror with alterable reflectivities. The discussion is initiated by explaining the light behaviors as a function of this distance. Then, the principles of Brillouin and Stokes photon propagation in the forward pumping laser arrangement are detailed. These understanding are crucial to explain the formation of 20 GHz spacing that manifests the absence of four-wave mixing. The investigations to determine the optimized value of $L_{mm}$ are also carried-out precisely. To justify the contribution by the AG, another setup that excludes this structure is also constructed under exactly the same pumping conditions.

2. Experimental setup

The forward-pumping MBRFL that includes an AG is arranged in the configuration that is demonstrated in Fig. 1. The oscillator consists of a 7 km piece of DCF, obtained from Corning that behaves as a nonlinear Brillouin-Raman gain medium. The DCF has an insertion loss of 7.5 dB, a nonlinear coefficient of 7.3 (Wkm)⁻¹ and an effective area of 20 μm². This tight-confinement geometry justifies its elemental properties of strong nonlinearities. The Brillouin pump (BP) source is contributed by an external cavity tunable laser source (TLS) that accommodates a flexible 31 nm tunability from 1535 to 1566 nm. The device has maximum power of 13 dBm and 200 kHz linewidth. It is injected into the MBRFL through an isolator (ISO1) that allows a unidirectional light propagation scheme, thus preventing any returning beam that might harm the input signal. In addition, the Raman pump unit (RPU) with maximum 1 W power is utilized as a pump scheme at a wavelength of 1455 nm. Both RPU and BP seed are coupled by the 1480/1550 nm wavelength selective coupler (WSC) into the system. In contrast to previous assessments [15,16,18], the AG is incorporated externally outside the half-open cavity as manifested in Fig. 1. The two fiber-end surfaces labeled as I and II mounting on a Fujikura splicer model FSM-100P are aligned coaxially. The machine introduces a series of precise control with an adjustable $L_{mm}$ from 0.05 to 1.00 mm. Both fiber end facets are smooth and cleaved carefully to attain flat-angles within the acceptable tolerance of less than 2°. This results in the cumulative Fresnel reflectance, $R_F$ back-coupling to the cavity owing to the alteration in refractive indices between fiber cores, $n_1=1.47$ and air space, $n_0=1.0$ as specified in [15]. Before further analysis on the AG fundamentals, the value of 3.62% is estimated from the equation, $R_F=\frac{n_1-n_0}{n_1+n_0}$ [19] by taking only one side of fiber I-to-air interface. This is evaluated by assuming that the incident light beam radiating perfectly on the normal path of the fiber core where ${\theta }_i=0^0$.

 

Fig. 1 Experimental layout of MBRFL with an AG arranged between fiber interfaces I and II.

Download Full Size | PDF

It is very critical to underline two aspects of studies that relate to the incorporation of AG where firstly, it behaves as a set of individually changeable output couplers. Secondly, the design introduces the way of tailoring the percentages of optical feedback that completes the laser cavity. To understand these, the fiber-coupling diagram that is demonstrated in Fig. 2(a) is built. The light beam (1550 nm) from the TLS unit at 15 mW power propagates from port 1 to 2 of the circulator before passing through the AG. To satisfy the first aspect aforementioned above, the first optical power meter (OPM1) assesses the remaining transmitted power after having a variable insertion loss induced by the adjustable$L_{mm}$. For the second aspect, any reflected power from the fiber to air interfaces directing towards port 2 to 3 of the circulator is evaluated by the second optical power meter (OPM2). The propagation loss is carefully characterized and all the optical power is back-calculated to both ends of AG (labeled as point A and B). In this case, point A serves as the incidence and reflectance nodes where the transmitted power is referred to the point B. From the plot in Fig. 2(b), the insertion of $L_{mm}$ that is increased from 0.05 to 0.90 mm distance results in the reduction of $R_F$ from 10.4% to 4.13% respectively. Similarly, this corresponds to 75% down to 2.3% power transmission. Further length extension up to 1 mm induces maximum insertion loss that results in merely 1.9% light transmission as detected by OPM1 and only 4% $R_F$ [19] is initiated. This denotes the insignificance of AG because the function of fiber II is trivial owing to the residual 0.38% contribution. With regards to this assessment, we concentrate essentially on the roles of this structure to improve the lasing performances compared to those when the $L_{mm}$ is set at 0 mm. In the latter case, the absence of fiber separation indicates 0% $R_F$ that corresponds to nearly 100% transmittivity. This implies a fully open cavity that ascertains the full contribution by Rayleigh scattering to the lasing behaviors (see Fig. 1 in [8]). The main prerequisite to remove any unnecessary light waves coming from the end tip of the fiber right-arm is by employing the second isolator (ISO2). During experiment, an optical spectrum analyzer (OSA) that has a resolution bandwidth of 0.02 nm is located at the output port to carry out data analysis.

 

Fig. 2 (a) Arrangement for the AG characterization and (b) the optical properties as a function of L_mm.

Download Full Size | PDF

3. Results and discussions

This report is started by discussing the operation principles of the proposed multiwavelength laser before presenting the experimental results obtained. In the physical pumping that employs full-forward direction, the Raman gain distribution is given as [20],

 

Fig. 3 (Above) Forward-pumped multiwavelength fiber laser. (Bottom) Microscopic propagation of Brillouin photons in the fiber waveguide where the Raman gain satisfies Eq. (1). BP: Brillouin pump wave, S1 and S2: Brillouin Stokes waves 1 and 2.

Download Full Size | PDF

From Fig. 3 when the BP photon is injected to the fiber input (brown arrow), the first Brillouin Stokes line (S1) is produced in the opposite way (green arrow) once the threshold level is satisfied. The adjustable AG initiates a pump recycling mechanism [21] by reflecting back some fractions of the remaining BP photons to the DCF. More energy that assists in the additional amplification of SBS is provided which supports the reduction in the threshold of S1. Equivalent to the task accomplished by BP, the S1 also behaves as a new pump seed. This results in the generation of the second Brillouin Stokes line (S2) (red arrow) in the same BP photon propagation direction. At $z=0$ the $RG(z)$ is maximum, the residual S1 photons is backscattered in the same direction of S2. Most of these light beams are consumed to sustain the additional amplification of S2 and favor its decrease in threshold. In a full analogy to that occurs to BP waves, the process repeats its cycle. Some percentages of S2 waves are reflected back to the DCF, thus facilitating in the amplification of S3 waves. In practice, the generation for the next successive cascaded levels continues to carry on as long as the preceding Brillouin Stokes (BS) thresholds are fulfilled. The pump recycling behavior for BP and BS even-orders strengthened by the AG is responsible for the lower threshold acquirement of the consequent higher odd-orders. As a result, odd-order channels are realized at the left fiber arm together with the Fresnel reflected components of even-orders, thus producing 10 GHz. wavelength operation. In contrast to this, the growth for those of even-orders initiated by SBS is attained at the output port. The main factor that influences the decline in threshold for this wave type is determined by those of the backscattered odd-orders. Most of these latter components experience attenuation as represented by the “depleted S1” for green arrow in Fig. 3 that happens in proportion to the amplification growth of the higher even-orders. This fulfills the exponential decay of $RG(z)$ along the fiber length that obeys Eq. (1) as simulated before (see Fig. 1 in [8]). As minimum $RG(z)$ is achieved at $z=L$ together with the depletion process, the Rayleigh scattering elements of odd-orders BS lines at the output tap are not significant. This justifies the formation of 20 GHz (0.158 nm) cascaded SBS shifts [10,11] as illustrated clearly in Fig. 4(a).

 

Fig. 4 Representation of output spectra for (a) the enlarged view at 20 GHz. wavelength spacing and at different AG lengths of (b) 0.05 mm, (c) 0.2 mm and (d) 1 mm (BP power = 2 dBm, BP wavelength = 1529 nm, RPU power = 950 mW).

Download Full Size | PDF

At the outset of this assessment, several investigations to determine the best selection of pumping criteria that satisfy the research objectives are implemented. These are met when the BP power is fixed at 2 dBm and BP wavelength at 1529 nm. In addition, the RPU power is maintained at 950 mW in the entire experiment. By changing the $L_{mm}$ gradually in the desired range [Fig. 2(b)], the spectral properties are quantified. A few depictions are demonstrated in Figs. 4(b)–4(d) where three contrasts at various AG lengths are specifically chosen to show its promising role for any improvement. The figure denotes the technique of assessing the channel flatness within the MBW definition that is set constantly at 45 nm. All the peak power differences between the neighboring lasing lines that start 1 nm away from the fixed BP signal of 1529 nm and end at the standard band-edge of 1575 nm are sum up. Other higher or lower peak power growths outside this bandwidth domain are not counted to maintain consistency and standardize the data measurements that are further summarized in Fig. 5. This figure also comprises the plot of average output power $(P_{out})$ with respect to various AG lengths. The physical configuration at $L_{mm}=0$ results in the channel flatness of 7.8 dB. The power scales-up to 10.4 dBm which is accredited to Rayleigh scattering activity that serves as virtual reflectors to complete the laser cavity. The power attainment at this level is maximum owing to no insertion loss, thus convenient for highly efficient laser operation.

 

Fig. 5 Channels flatness and average output power measured by OSA as a function of L_mm (MBW = 45 nm, BP power = 2 dBm, BP wavelength = 1529 nm, RPU power = 950 mW).

Download Full Size | PDF

Next with the insertion of AG that introduces a lumped feedback, the lower $P_{out}$ trend is observed. It is found to be inversely proportional to $L_{mm}$ as a result of the reduction in the accumulative $R_F$ and output coupling transmission as presented in Fig. 2(b). The AG that behaves alternatively as a low-cost solution to highly changeable reflective mirror is not intended for enhancing the optical-to-optical efficiency. It is instead a rather beneficial approach to control the peak power discrepancy between the Stokes channels. Although this aggravates to 10.2 dB once the first AG distance of 0.05 mm is embedded [see Figs. 4(b) and 5], a favorable flow can be seen from the asymmetrical blue line plot in Fig. 5. By ignoring the output coupling transmission, the phenomenon can be described as follows. The increase in $L_{mm}$ leads to lower back-reflection lasing as manifested in Fig. 2(b) which backs up the generation of smaller $P_{out}$. As the BP and RPU power are fixed at the original values, the consequent improvement in the self-flattening up to a certain level is in commensurate to the decrease in the mode competition between self-lasing modes and Brillouin lasing lines. The early example at 0.2 mm gap as depicted in Fig. 4(c) shows the relatively smoother feature. This represents that the right proportion of energy transfer between the self-lasing modes and BS waves is required to saturate the peak lasing lines in order for initiating more efficient self-flattening. A minimum flatness of 1.1 dB is accomplished at 0.4 mm distance which corresponds to −1.3 dBm $P_{out}$. This identifies the best counter-balance attainment that necessitates the right tuning of $R_F$ up to 5.1%. As a result, we stick at this $L_{mm}$ for the next evaluations on various lasing properties. Beyond this value, the quality of managing the peak power discrepancy is deteriorated. The continuous reduction in back-reflection lasing that yields lower mode competition remains influential [Fig. 2(b)]. Despite of this, the consequent decrease in $P_{out}$ implies that the proportion of energy transfer between the self-lasing modes and BS waves is not sufficient to saturate the peak lasing line intensities. Accordingly at 1 mm distance that corresponds to 4% $R_F$ [19], up to 5.2 dB flatness is achieved. The drawback is clearly noticeable in the entire range especially at the band-edge of the spectral envelope as depicted in Fig. 4(d). In this case, the AG is irrelevant and only provides the method for very low output coupling transmission [Fig. 2(b)] where the $P_{out}$ is lowered down to −8.3 dBm. It is very critical to point out that the four-wave mixing [22,23] neither assists in the development of higher-order cascaded Stokes waves nor the self-flattening effect. This is because no anti-Stokes lines are observed at all spectral features illustrated in Fig. 4(b and c) which ascertain this claim.

In all the following assessments we pay attention at particularly two setups, the first is without AG that is categorized as Setup 1 and the second is Setup 2 that signifies the insertion of 0.4 mm AG. This is done in the evaluation for output spectrum as illustrated in Fig. 6. From this figure, the study is begun when the BP is turned off and the RPU power remains at 950 mW. In Setup 1, the physical feature is almost identical to some of those analyzed earlier (see Fig. 2 in [11]). The Raman peak gain is estimated at 1555 nm which matches the 13.3 THz. Raman shift with regards to the pump waves at 1455 nm wavelength. Additionally, the spectrum depicts no any signs of spikes generation. In contrast, more pronounced stochastic modes from the build-up of dynamics lasing spikes are the traits for Setup 2. In comparison to that of −42.0 dBm attained in the first case, a slightly lower noise floor in the vicinity of −53.2 dBm is realized. Both are measured at 1552 nm wavelength. Together with more positive evidence shown by the distinguishable increase in the peak power for the black profile, the potential improvement on the OSNR is predicted. In fact, its strong nonlinear properties that appreciate the overall cavity gain is promising for the attainment of more BS lines that implies broader bandwidth [4–6]. Interestingly, the repetitive FSR of 2.9 nm is clearly seen from the lasing output domain that covers 1550 to 1575 nm wavelength which pertains to the etalon effect introduced by the $L_{mm}$. By taking any value within this emission wavelength band as $\lambda$, the formula for free spectral range (FSR) is given as $\Delta {\lambda }_{FSR}={\lambda }^2/2.n_0.L_{mm}$. When substituting the refractive index of the air $n_0=1$, the $\Delta {\lambda }_{FSR}$ is approximated from the calculation as 3 nm which approaches closely the measured value. In addition to this by examining Fig. 6, higher intensity fluctuations are anticipated at shorter AG length. This is owing to the stronger reflection property as characterized in Fig. 2(b) that results in stronger gain competition between self-lasing modes and BS waves. As a consequence, the flatness of the entire spectral envelope is deteriorated. This justifies the worse peak power discrepancy for multiwavelength lasing when the $L_{mm}=$ 0.05 mm is implemented rather than that obtained at $L_{mm}=$ 0 mm as demonstrated in Fig. 5. From this figure, the particular deficiency inspires more experiments that lead to the determination of optimized AG length. Once this is fulfilled, further elaborations are presented in the next figures.

 

Fig. 6 Output spectrum at RPU power of 950 mW and BP is turned off. The noise floors in red texts are measured at 1552 nm wavelength.

Download Full Size | PDF

In Fig. 7, the roles of BP wavelength are studied at the fixed 0.4 mm AG to confirm about the right selection responsible for the optimization of channel flatness (red and blue lines). From this figure, the peak power discrepancies between the maximum to minimum channels at both setups are measured within the same MBW scope as shown by the green line. Similar to the previously reported assessment (see Fig. 6 in [10]), the increase in BP wavelengths leads to the decrease of the remaining accessible bandwidth of the Raman gain. As a result, the MBW reduces from 45.0 nm to 5.5 nm for the tuning in BP wavelengths from 1529 to 1565 nm. When this latter parameter is set at 1529 nm, the best self-flattening of 1.1 dB is realized at Setup 2 in full agreement to that presented in Fig. 5. Below 1550 nm, the flatness is better for Setup 2, but the implementation of AG in the cavity is pointless at a longer BP wavelength as signified by the overlap of blue to red plots in Fig. 7. By comparing this result to that presented in Fig. 5 expose several aspects of operation. The same plots but not the same values for blue and green lines in Fig. 5 are attained at different BP wavelength tunability. The latter variable has minimum effect on the $R_F$ and power transmission that is directly dependent on the changes of $L_{mm}$. This also implies that the channel flatness enhancement influenced by the optical and lasing characteristics described before mostly relies upon the optimization of AG length. Nevertheless at certain $L_{mm}$, the BP wavelength still plays the role on flatness trends as clarified in Fig. 7. This pumping factor mainly affects the number of Stokes lines and the corresponding MBW which verifies the selection of 1529 nm wavelength. From this figure also, changing the $L_{mm}$ still maintains the equivalent behaviors for the red and blue lines although their exact values are not necessarily identical. Once the right pumping criteria needed to achieve the objectives of this assessment have been determined, laser thresholds at both setups are inspected. By maintaining the BP power at 2 dBm and BP wavelength at 1529 nm, the RPU power is scaled-up slowly. In this work, the threshold is defined as the first appearance of 20 GHz shift with respect to the BP signal, having a minimum peak power of −25 dBm. The threshold is found to be 780 mW for Setup 1 and this is further reduced by 11.5% to 690 mW for Setup 2. This relates to the benefit of the AG structure rather than that contributed completely by Rayleigh back-scattering in the first setup. The underlying physics behind the lower threshold attainment in the second setup has been described earlier in Fig. 3. A report by X. Li and associates also presents almost the same behavior particularly at higher-orders Stokes waves (see Fig. 5 in [15]). Other studies on this topic have also been explained in [24,25].

 

Fig. 7 Channels flatness as a function of BP wavelength at the specified multiwavelength bandwidth coverage (RPU power = 950 mW, BP power = 2 dBm).

Download Full Size | PDF

A lower threshold yields more generation of Brillouin lines when the same RPU power is provided. This is proven in Fig. 8 which indicates the superior benefits of incorporating the AG into the cavity. Before further elaboration, we need to highlight again that the same spectra are measured in Figs. 5 and 8 at the same AG distance. The exact 45 nm MBW description that results in 7.8 dB and 1.1 dB flatness for Setup 1 and 2 as applied in Fig. 5 is used for clarity to standardize the measurements at all $L_{mm}$. The image in Fig. 8(a) that demonstrates 15.9 dB flatness at $L_{mm}$ = 0 is determined within the 46.28 nm bandwidth. The total count of Stokes waves is 294 lines which is considered by including all lasing lines from 1529.16 nm to 1575.44 nm wavelength. More gain competition between the BS waves to the self-lasing modes can be seen which justifies the degradation in the flatness quality. As a consequence of stronger spectral broadening [26–28], the feature shows a higher noise floor. Close to this level, the OSNR is 17.2 dB and it is very difficult to control the uniformity in the signal amplitude as illustrated by the corresponding magnified view in Fig. 8(c). On the other hand, the graphic in Fig. 8(b) represents a smoother spectrum with 1.8 dB flatness at $L_{mm}$ = 0.4 mm. A broader multiwavelength bandwidth of 46.60 nm is achieved, this comprises of 296 Stokes lines that extends from 1529.16 to 1575.76 nm wavelength. This is to date the finest results obtained in multiwavelength fiber lasers arranged in a simple configuration that employs only a single RPU. To confirm this, other results based on Brillouin-Raman and Brillouin-Erbium transitions are listed in Table 1. From this table, the closest competition is offered by our preliminary findings in the same area that only show 29 to 32.5 nm MBW. These comprise of more than 212 Stokes lines, evaluated within the peak lasing line intensities that vary from 2.5 [11] to 5 dB [10] approximation. The noise floor is almost maintained at all wavelength slices which leads to the higher average OSNR of 20 dB as manifested in Fig. 8(d). This is already predictable from Setup 2 by referring to the noise floor properties explained in Fig. 6.

 

Fig. 8 Output spectra of MBRFL for (a) Setup 1 (no AG) and (b) Setup 2 (with 0.4 mm AG) and their corresponding enlarged visuals in (c) and (d), respectively. (BP wavelength = 1529 nm, BP power = 2 dBm, RPU power = 950 mW).

Download Full Size | PDF

Table 1. Achievements on 20 GHz. multiwavelength fiber lasers.

View Table | View all tables in this article

From the stability-wise outlook, the output signals at three separate Stokes wavelengths of 1535, 1550 and 1560 nm are scanned every single minute for half an hour. This is monitored by using an OSA because the possibility of background error from the remaining Raman pump wavelength at 1455 nm does not permit the implementation of OPM. The behavior is depicted in Fig. 9 where the corresponding numbers are listed in Table 2 for easy understanding. From this observation, the bright prospect of the second setup is validated by its performance around 1.2 to 1.5 dB fluctuation at the aforementioned wavelengths. This is closely comparable to those assessments that have below 1 dB stability elucidated previously [15,18,22]. The several times drastic improvement is confirmed by referring to that of 9 to 14 dB fluctuation achieved in the first setup as expressed in Table 2. This can be attributed to the self-stabilization by gain saturation [29]. The intense BP power saturates the peak power level of the higher-order Stokes waves, thus reducing their power fluctuations. By further improving this value in the future attempts promises the reliability and robustness of the device for various applications in optical communication, fiber optics sensing and etc.

 

Fig. 9 Peak power fluctuation at (a) Setup 1 and (b) Setup 2 where all pumping properties are similar to those mentioned in Fig. 8.

Download Full Size | PDF

Table 2. Numerical summary for the results obtained in Fig. 9.

View Table | View all tables in this article

4. Conclusion

We set our aim mainly on investigating the $L_{mm}$ of fiber-to-air interfaces employed outside the laser cavity to enhance the spectral flatness in double-spacing MBRFL. This is accomplished at an acceptable OSNR without compromising the MBW and number of Stokes lines. The main factor is determined by the AG structure that initiates multiple Fresnel reflectivities re-injecting back to the cavity that dominate over relatively weaker Rayleigh scattering effects. Together with the original seed source, this auxiliary pump source results in low thresholds of the next Stokes waves which continuously justify more generation of successive higher-order components. However, this is not only our main concern since most of the energy is consumed to initiate self-flattening of the overall signal envelope. At the optimized $L_{mm}$ of 0.4 mm, we believe that the right counterbalance between the 5.1% FR and the 2 dBm Brillouin pump waves results in the right suppression of gain competition. In this case, the smooth energy transfer between the self-lasing modes to BS waves is initiated. The almost same peak lasing amplitude of the 296 Stokes lines is maintained at 1.8 dB-order through gain saturation. Other pumping levels that assist this operation include 950 mW Raman power and 1529 nm BP wavelength. Amongst other excellent results are the clean-cut output feature with an average 20 dB OSNR that covers over the entire bandwidth of 46.60 nm. In conclusion, we have successfully accomplished the main objective as the contemporary results are better to other findings that incorporate a micrometer air gap inside the laser oscillator. This opens-up a new standpoint towards exploring the advantages of this structure for improving the performances of multiwavelength lasing generation.