Content uploaded by Sébastien Le Floch
Author content
All content in this area was uploaded by Sébastien Le Floch on Oct 01, 2018
Content may be subject to copyright.
Unique Coded Sequence For Fast Brillouin
Distributed Sensors
S. Le Floch, M. Llera, O. Gloriod, Y. Meyer, S. Monnerat and F. Sauser
Haute Ecole ARC Ingénierie (University of Applied Sciences of Western Switzerland),
Rue de la Serre 7, 2610 Saint Imier, Switzerland.
sebastien.lefloch@he-arc.ch
Abstract: We present, for the first time, a colour coded sequence running in an endless loop
and dedicated to ultra-fast Brillouin distributed sensors. Measurements within the whole
elastic zone of single-mode optical fibers are presented.
OCIS codes: (060.2370) Fiber optics sensors, (290.5900) Scattering, stimulated Brillouin.
1. Introduction
The Brillouin Optical Time Domain Analysis (BOTDA) technique, based on stimulated Brillouin scattering, is
widely used for distributed temperature or strain sensing [1]. It consists in scanning the frequency difference
between two counter-propagating light waves (a pulsed pump and a continuous wave probe) around the
Brillouin frequency. The distributed Brillouin gain spectra (BGS) are obtained by detecting the local amplitude
variations of the probe wave along the sensing fiber and for each scanned frequency, the maximum amplitude of
the BGS providing the information of temperature or strain. For some application fields such as distributed
deformation measurement in large critical infrastructures, this conventional scan in frequency is too slow
(typically ranging from seconds to minutes, depending on a sensing length). In order to improve the acquisition
time, several techniques, amongst others, use a tracking of the Brillouin frequency shift (BFS) thanks to a
control loop [2,3], so that the local probe frequency is kept as close as possible to the maximum (or half-
maximum) of the resonance. In this paper, we propose a new and completely different approach that keeps the
frequency scanning in the whole elastic zone of the fiber sensor while maintaining the acquisition time as fast as
a conventional BOTDA, thanks to a unique sequence. This latter is generated by a dedicated arbitrary waveform
generator (AWG), is perfectly cyclic so that it runs in a loop without dead time, and includes a colour coding
that improves the signal-to-noise ratio: in this way, the averaging is reduced to a minimum.
2. The unique coded sequence
A unique coded sequence requires specific techniques for distributed Brillouin sensors: firstly, it is mandatory to
apply a cyclic code, looping back on itself [4]. Secondly, the cyclic code has to be colored as a distributed
Brillouin measurement is a 3D mapping (probe signal amplitude as a function of time and frequency). A first
attempt was realized with a colour cyclic code based on an equal number of different sequences and scanned
frequencies (i.e. the code length) [5]. However, these sequences were independent, with no link between them.
Fig. 1. Fast measurement principle. Top: conventional BOTDA with 4 scanned frequencies;
Bottom: the equivalent unique sequence with code length L=3.
Here we propose a refinement approach that maintains the link between all the coded sequences in order to
generate a unique sequence. Fig. 1 sketches the basic principle, compared to a conventional BOTDA: the
standard sensor scans the fiber under test (FUT) with four different frequencies with a minimum acquisition
time. Assuming that three positions are measured in the FUT, the total number of data is twelve. The
corresponding unique sequence is based on a circulant square matrix M of dimension L (L=3) [6] and contains
L+1 frequencies, so that the sequence is doubly circulant. The chronological order of the pump frequencies is
based on a repetitive pattern {f1, f2, f3, f4} where "zeros" are applied according to the coding matrix M. Three
pump pulses are continuously filling the FUT, so that the probe signal permanently records three events in the
FUT. Hence, between the start (t0) and stop times (t1), the probe signal records twelve times three events: the
unique sequence runs as fast as a conventional BOTDA. The signal processing consists in regrouping L+1 data
packets that are successively decoded by the inverse matrix M-1 (Fig. 2): the first data packet provides the
Brillouin gains
12 3
12 3
, andgg g
where
j
i
g
represents the gain at position i and for the frequency j. The next packet
is obtained by a right shift of data, providing Brillouin gains
23 4
12 3
, andgg g
and so on until all the data are
decoded. Note that all these calculations are linear and parallelizable, allowing fast decoding on an acquisition
card equipped with Field Programmable Gate Array (FPGA) and/or a Graphics Processing Unit (GPU).
Fig. 2. Decoding scheme of the unique sequence.
3. Experimental set-up
Fig. 3 shows the experimental set-up of a dual pump configuration: the output of the laser source is split in two
arms by an optical coupler. The probe wave propagates through the lower arm, where an acousto-optic
modulator negatively shifts the laser frequency by fs=35MHz. On the upper arm, the colour coded pump pulses
are generated in the electrical domain by an AWG that drives a Mach-Zehnder modulator (MZM) configured to
operate in double-sideband suppressed-carrier modulation mode. A narrow fiber Bragg grating (FBG-1) filters
out the residual power of the laser frequency. The pump waves, after amplification by an Erbium-doped fiber
amplifier (EDFA), enter into the FUT by passing through a 4-port circulator and the detected probe wave is
selected by FBG-2, used in reflection. Due to the long duration of the coded sequence, the acquisition card is
synchronized to the AWG, so that local information can be retrieved without any time drift.
Fig. 3. Experimental set-up. AWG: Arbitrary Waveform Generator; FBG: Fiber Bragg Grating; EDFA: Erbium-Doped Fiber Amplifier;
DFB: Distributed Feedback Laser: DAQ: Data Acquisition; DGD: Differential Group Delay Module.
A differential group delay (DGD) module is inserted after the MZM, with the incident angle set to 45°, so
that the two sidebands are orthogonally polarized [7]. As the probe wave experiences both gain (g) and loss (l)
processes coming from the upper and lower sidebands in frequency of the pump pulses, the Brillouin spectrum
is distorted (fig. 4), merely because stimulated Brillouin scattering depends on the polarization states of the
optical waves. It is however possible to superpose both Brillouin extrema in post-processing with the formula
g(fi)-g(fi-2fs), since they are exactly separated in frequency by 2fS. Moreover, the addition of both extrema
should be constant, whatever is the polarization state of the probe wave and provided that the effective Brillouin
linewidth ∆νeff (depending on the pump pulse duration) follows the relationship: ∆νeff ≤ 2fS.
Fig. 4. Principle of superposition of both gain and loss spectra depending on the polarization states (simulation).
Left: Brillouin spectrum; right: superposition by formula g(fi)-g(fi-2fs).
4. Measurements and results
In order to validate the concept described above, a distributed sensing along a 1.6 km single-mode fiber
terminated by a Brillouin spatial resolution test sample (with 5m, 3m 2m and 1m hot spots) is performed in one
shot. The pump pulses duration is set to 14ns with a return-to-zero (RZ) period of 170ns (one bit period
T=184ns), leaving enough time for the acoustic wave to decay, and the pump power is close to 23dBm (20dBm
per sideband). Note that the pump power is flat since the EDFA continuously amplifies the loop sequence. The
code length is L=103, and the 104 frequencies are equally spaced by 5MHz, meaning a frequency scan of about
500MHz. The total acquisition time is given by the formula (L2+L)⋅T=1.9331ms (517Hz). The decoded
Brillouin spectra are filtered in the time and frequency domains by a two-dimensional Gaussian filter [8].
Fig. 5. Brillouin gain after superposition of gain and loss spectra.
Left: distributed Brillouin gain vs distance; right: typical Brillouin spectrum.
Fig. 5 shows the distributed Brillouin amplitude, after a single-shot measurement and superposition of the
Brillouin extrema. The Brillouin gain distribution is sufficiently flat to perform Brillouin frequency shifts
measurements. Moreover, the BGS exhibit no apparent distortion that could result from the coding scheme. In
order to avoid measurement errors due to the unbalanced shape of the Brillouin spectrum, a third order
polynomial fit is applied on the upper part. Fig. 6 represents the estimated Brillouin frequency shifts (BFS)
along the whole sensing fiber, with a constant uncertainty of 1.4MHz (2σ) evaluated after 10 successive
measurements. Measurements on the Brillouin spatial resolution test sample prove the validity of the concept,
by detecting the 5m, 3m and 2m hot spots with a constant frequency shift (15MHz). In addition, the rise and fall
on the BFS correspond roughly to the pump pulse duration (the sampling interval is 4ns).
Fig. 6. BFS vs sensing distance;
left: full sensing distance; right: spatial resolution test sample with 5m, 3m, 2m and 1m hot spots.
Finally, the unique sequence is tested with two spools of single-mode optical fibers, terminated by the same
Brillouin spatial resolution test sample (total 53km). The code length is L=359, with 360 frequencies equally
spaced by 3.2MHz (total frequency scan of about 1.1GHz): with a 50kHz/µε strain dependence of the BFS, the
full elastic zone is covered (with a maximum elongation / compression of ±1%). The input probe power is set to
-13dBm, so that pump depletion is limited to 10% [9]. The pump pulse duration is set to 30ns (3m spatial
resolution) with a RZ period of 1410ns (T=1440ns), so that the acquisition time is 185ms. In this case, with a
long sensing distance, the sensor experiences polarization issues (the distributed Brillouin amplitude after
superposition of gain and loss curves is no longer flat). In order to solve this problem, a polarization scrambler is
required after FBG-1.
Fig. 7 represents the calculated BFS with an averaging of 56 successive scans, meaning a total acquisition
time of 10s. Again, the hot spots are perfectly resolved, for 5m and 3m, proving the ability of the unique
sequence to retrieve local information, despite a spread information through the total acquisition trace. Note that
without synchronization between the acquisition card (with typical time base stability of 1ppm) and the AWG,
the spatial resolution would have been strongly affected.
Fig. 7. BFS vs sensing distance.
Left: full sensing distance (53km, Av. 56); right: at the end of the sensing fiber (Brillouin spatial resolution test sample).
The last figure indicates the accuracy of the sensor at the end of 50km according to the averaging of the
unique sequence. The uncertainty is calculated with a minimum of 5 successive measurements for each sample
size. A sample of N=7 traces is just enough to mitigate polarization issues. In 10s, the sensor presents a
maximum uncertainty of 2.5MHz (2σ). The dotted line fits the 1/√N feature of the uncertainty.
Fig. 8. Uncertainty (2σ) near 53km in function of the acquisition time (1.3s corresponds to average 7).
5. Conclusion
In conclusion, we have successfully shown that it is possible to measure distributed Brillouin gain spectra in the
whole elastic zone of a single-mode optical fiber with a unique and colour coded sequence, running in loop and
as fast as a conventional BOTDA may do. The acquisition time for a sensing range of 1.6km with a spatial
resolution of 1.4m is less than 2ms, achieving a speed rate of 517 Hz with an uncertainty of 1.4MHz (2σ). The
time to perform long distance measurements (53km) is reduced to seconds, depending on the accuracy goal of
the sensor.
The authors would like to acknowledge the company OMNISENS (Switzerland) and the Swiss Commission for
Technology and Innovation (Project 18337.2 PFNM-NM) for their support.
6. References
[1] M. Nikles, L. Thevenaz and P.A. Robert, "Brillouin gain spectrum characterization in single-mode optical fibers," J. Lightwave Technol.
15(10), 1842-1851 (1997).
[2] A. Motil, R. Davidi, R. Hadar and M. Tur, "Mitigating the effects of the gain-dependence of the Brillouin line-shape on dynamic
BOTDA sensing method," Opt. Express 25(19), 22206-22218 (2017).
[3] Z. Yang, M.A. Soto, P. Ray and L. Thévenaz, "Brillouin distributed optical fiber sensor based on a closed-loop configuration," J.
Lightwave Technol. 36(5), 1239-1248 (2018).
[4] H. Iribas, A. Loayssa, F. Sauser, M. Llera and S. Le Floch, "Cyclic coding for Brillouin optical time-domain analyzers using probe
dithering," Opt. Express 25(8), 8787-8800 (2017).
[5] S. Le Floch, F. Sauser and M. Llera, "Colour cyclic code for Brillouin distributed sensors," in 24th Int. Conf. on Optical Fiber Sensors,
Proc. SPIE 9634, 963431 (2015).
[6] S. Le Floch, F. Sauser, M. Llera and E. Rochat, "Novel Brillouin optical time-domain analyzer for extreme sensing range using high-
power flat frequency-coded pump pulses,", J. Lightwave Technol. 33(12), 2623-2627 (2015).
[7] J. Urricelqui, M. Sagues, A. Loayssa, M. Alem, L Thévenaz, and M.A. Soto, "Mitigation of modulation instability in Brillouin
distributed fiber sensors by using orthogonal polarization pulses," 24th Int. Conf. on Optical Fiber Sensors, Proc. SPIE 9634, 963433 (2015).
[8] S. Le Floch and F. Sauser, "New improvements for Brillouin optical time-domain reflectometry," 25th Int. Conf. on Optical Fiber
Sensors, Proc. SPIE 10323, 103230L (2017).
[9] L. Thévenaz, S.F. Mafang and J. Lin, "Effect of pulse depletion in a Brillouin optical time-domain analysis system," Opt. Express
21(12), 14017-14035 (2013).
