JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 39, NO.
7, APRIL 1, 2021
Forward Transmission Based Ultra-Long Distributed
Vibration Sensing With Wide Frequency Response
Yaxi Yan , Faisal Nadeem Khan, Bin Zhou, Alan Pak Tao Lau, Chao Lu, Fellow, IEEE, Fellow, OSA,
and Changjian Guo
Abstract—A novel ultra-long distributed vibration sensor using
forward transmission, coherent detection, and a frequency-shifted
optical delay line is proposed and experimentally demonstrated.
In the proposed scheme, a pair of multi-span optical fibers are de-
ployed for sensing. A frequency-shifted optical delay line consisting
of an acousto-optic modulator and a time delay fiber is used at the
far end of these two fibers. Coherent detection is used to retrieve
the vibration-induced phase fluctuations of the base-band signals
as well as the intermediate frequency signals generated by the
frequency-shifted optical delay line. Two differential phase signals
can be calculated from the obtained phase fluctuations, which can
then be used to localize the vibration events by correlation opera-
tions. Localization of a few hundred Hz, around 1 kHz and tens of
kHz vibrations has been experimentally demonstrated over a total
length of 1230 km sensing fiber. Less than 125 m spatial resolution
can be obtained over the 615 km sensing range for vibrations with
larger than 1 kHz frequency by using averaging of 30 times tests
due to the nature of asynchronous operation. Frequencies from
Manuscript received August 15, 2020; revised November 12, 2020; accepted
December 9, 2020. Date of publication December 14, 2020; date of current
version April 2, 2021. This work was supported in part by National Key R&D
Program of China (2018YFB1801701), in part by Science and Technology
Program of Guangzhou (2019050001), in part by Science and Technology
Planning Project of Guangdong Province (2019A050510039), in part by Natural
Science Foundation (NSF) of Guangdong Province (2018A0303130117), in part
by The Research Grants Council (RGC) of Hong Kong, General Research Fund
(GRF: PolyU 152658/16E, 152168/17E); and in part by The Hong Kong Ph.D.
Fellowship. (Corresponding author: Changjian Guo.)
Yaxi Yan is with the Department of Electronic and Information Engineering,
The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
(SAR), China (e-mail: ya.xi.yan@connect.polyu.hk).
Chao Lu is with the The Hong Kong Polytechnic University Shenzhen
Research Institute, Shenzhen 518057, China, and also with the Depart-
ment of Electronic and Information Engineering, The Hong Kong Polytech-
nic University, Hung Hom, Kowloon, Hong Kong (SAR), China (e-mail:
chao.lu@polyu.edu.hk).
Faisal Nadeem Khan is with the Tsinghua–Berkeley Shenzhen
Institute, Tsinghua University, Shenzhen 518071, China (e-mail:
fnadeem.khan@yahoo.com).
Bin Zhou is with the South China Academy of Advanced Optoelectron-
ics, South China Normal University, Guangzhou 510006, China (e-mail:
bin.zhou@coer-scnu.org).
Alan Pak Tao Lau is with the Department of Electrical Engineering, The
Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong (SAR),
China, and also with The Hong Kong Polytechnic University Shenzhen Research
Institute, Shenzhen 518057, China (e-mail: alan.pt.lau@polyu.edu.hk).
Changjian Guo is with the South China Academy of Advanced Optoelec-
tronics, South China Normal University, Guangzhou 510006, China, and also
with the Department of Electronic and Information Engineering, The Hong
Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong (SAR), China
(e-mail: changjian.guo@coer-scnu.org).
Color versions of one or more of the figures in this article are available at
https://doi.org/10.1109/JLT.2020.3044676.
Digital Object Identifier 10.1109/JLT.2020.3044676
0733-8724 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
2241
infrasound to ultrasound can be detected. The proposed scheme has
advantages of ultrabroad frequency response, ultra-long sensing
range and simple sensing structure due to the nature of forward
transmission and coherent detection. The sensing length can be
further extended using more fiber spans and erbium-doped fiber
amplifiers, making it a promising candidate for vibration event de-
tection and localization in long-haul and ultra-long-haul fiber links.
Index Terms—Distributed fiber sensor, forward transmission,
ultra-long haul, wide frequency response.
I. INTRODUCTION
ISTRIBUTED vibration sensor (DVS) has become a core
D technological component in a wide variety of engineering
disciplines ranging from oil and gas pipelines monitoring for
leakage and intrusion, protection and harsh weather condition
detection of power lines, structural health monitoring of bridges
and dams, perimeter security protection, communication net-
work monitoring for optimized transmission capacities and even
detecting/locating seismic activities [1]–[4]. For DVS systems,
the fundamental technology can be divided mainly into two cate-
gories. The first is based on the optical time-domain reflectome-
try (OTDR) technique, including phase-sensitive OTDR [5]–[7],
polarization OTDR [8], [9] and Brillouin scattering based OTDR
[10]. Another is based on the interferometric technique, includ-
ing Michelson interferometer [11], Mach-Zehnder interferome-
ter (MZI) [12], [13] and Sagnac interferometer [14], [15], or their
combinations [16]. In OTDR based DVS systems, the backscat-
tered light is used to locate the vibration along the fiber. However,
the backscattered light has low intensity and suffers from loss
along forward and backward propagation, restricting the sensing
range of the OTDR based systems. Even though over 100 km
sensing range has been demonstrated in phase-sensitive OTDR
using the distributed amplification [17]–[19], the cost and system
complexity increases accordingly. Meanwhile, pulsed light is
used to sample the vibration, capturing fast vibration dynamics
will be another bottleneck, especially in long-distance OTDR-
based DVS systems. Frequency division multiplexing (FDM)
technique which enables multi pulses within a single round
trip time is used to extend the measurable vibration frequency
beyond the conventional limit of OTDR-based DVS systems
[20], [21]. The detectable frequency of the FDM based system
is related to the number of frequency channels. As a result, large
data amount is needed for high-speed vibration detection, which
makes the data analysis quite challenging for long haul sensing.
In interferometric DVS systems, high-speed vibration detection
 2242
is no more a problem, but Rayleigh backscattering produces
interference that might affect the interferometry performance
as well as the sensing range [13]. Interferometric DVS systems
usually use direct detection, and dual interferometers are needed
to locate the position of vibration [22], [23]. In Sagnac DVS
systems, two sets of traveling clockwise (CW) and counter-
clockwise (CCW) beams along the fiber are utilized to detect
and locate the vibration. A sensing range of 41 km and a spatial
resolution of 100 m was reported [24]. While for Michelson
interferometric DVS, a 50 km sensing range was achieved with
a location accuracy of ±100 m [11]. For MZI-based DVS, a
record 320 km sensing distance was reported [22]. It should be
noted that MZI-based DVS requires a stable reference arm as
well as polarization tracking. Time-varying phase information
can be used for vibration detection as well. However, two sets
of either homodyne or heterodyne receivers are needed [25].
In this paper, we propose and experimentally demonstrate
a novel forward transmission based ultra-long DVS with a
much simpler structure compared with the interferometric DVS
system, using coherent detection with digital signal processing
(DSP) and a frequency-shifted optical delay line (FS-ODL). A
pair of multi-span optical fibers are deployed and utilized for
sensing while a loop-back configuration is used. A commercial
phase- and polarization-diversity coherent receiver (CoRx) is
used to extract the phase information of the output signal after
forward transmission. Since both fibers are placed at the same
location, the same phase change patterns will occur in case of
disturbance events. An FS-ODL consisting of two couplers,
an acousto-optic modulator (AOM) and a time delay fiber
(TDF) is used to generate two signals of different frequencies
with a fixed time delay. Two phase differential signals can be
constructed from the phase information of these two signals
of different frequencies, which can be used to determine the
vibration position. Localization of such vibrations is achieved
by calculating the time delay between two constructed phase
differential signals using correlation operations. Averaging
can be used to further improve the localization accuracy. In
this work, we investigated the sensing performance of a 1230
km fiber link, which translates into a sensing range of 615
km. The sensing range can be further extended by using more
fiber spans with optical amplification. The proposed scheme
thus dramatically improves the range and performance of the
DVS systems and enables a new wave of ultra-long-haul and
wide-frequency vibration sensing applications.
II. OPERATION PRINCIPLE
In our sensing scheme, the fiber under test (FUT) consists
of two fibers placed together along the sensing area. At the far
end of the FUT, the forward transmitted light is split into two
beams. Beam 1 goes directly to the coupler. Meanwhile, beam
2 goes through an AOM followed by a TDF which is used as a
frequency shifter and a fixed time delay, respectively. Beam 1 and
beam 2 are coupled into the FUT again and finally detected by a
commercial phase- and polarization-diversity CoRx. When there
is no external vibration, the phase of the light at the receiver will
vary slowly due to the carrier phase noise and other phase noises
Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 39, NO. 7, APRIL 1, 2021
Fig. 1. Schematic diagram of the localization principle.
resulting from environmental fluctuations (wind flow, acoustic,
thermal variations, etc.). When an external vibration occurs at
a certain position of the FUT, the refractive index of the fiber
core and the fiber birefringence will change accordingly, which
in turn will induce a phase change in the transmission light. Due
to the loop-back configuration, a single external vibration will
affect two different positions of the FUT, marked as positions A
and B in Fig. 1. Because two fibers are very close to each other,
when light passes through position A and position B, the phase
shift caused by the external vibration will be the same. The light
at the receiver contains baseband and intermediate frequency
(IF) components due to the FS-ODL.
From Fig. 1, the propagation path for the baseband signal
is pathb : Tx→A→1→B→Rx, and the baseband signal can be
expressed as

Eb = Pb cos[ω0 t + ϕnb (t) + φ(t) + φ(t − t1 )] (1)
The propagation path for the IF signal is pathif :
Tx→A→2→B→Rx, and the IF signal can be expressed as

Eif = Pif cos[(ω0 + Δω) (t − Δt) + ϕnif (t)
+ φ(t) + φ(t − t2 )] (2)
where Pb and Pif are the powers of the baseband signal and
the IF signal, respectively. ω 0 is the angular frequency of the
incident light. Δω is the frequency shift caused by the AOM.
ϕnb (t) and ϕnif (t) are phase noises of the baseband signal and IF
signal, respectively. φ(t) is the phase change induced by external
vibrations. t1 is the time for light to propagate from A to B along
path: A→1→B. t2 is the time for light to propagate from A to
B along path: A→2→B. Δt is the time difference for light to
propagate along pathb and pathif . It is obvious that
Δt = t2 − t1 . (3)
After beating with the LO, four-channel output signals of the
receiver contain in-phase (Ix, Iy) and quadrature (Qx, Qy) com-
ponents of the beating signal between the forward transmitted
light and the local oscillator (LO), in both x- and y- polarizations.
Phase information of the baseband signal and IF signal can be
extracted since the whole optical field is obtained using coherent
detection [26]. The demodulated phase of the baseband signal
can be written as
ϕb (t) = ϕnb (t) + φ(t) + φ(t − t1 ) − ϕLO (t). (4)
The demodulated phase of the IF signal can be written as
ϕIF (t) = − (ω0 + Δω) Δt + ϕnif (t) + φ(t)
+ φ(t − t2 ) − ϕLO (t), (5)
YAN et al.: FORWARD TRANSMISSION BASED ULTRA-LONG DISTRIBUTED VIBRATION SENSING
where ϕLO (t) is the phase of the LO. After removing the
DC component (ω0 + Δω)Δt through subtracting the average
power of the signal, ϕIF (t) can be written as
ϕIF (t) = ϕnif (t) + φ(t) + φ(t − t2 ) − ϕLO (t).
We can construct two differential signals as
Δϕ1 (t) = ϕb (t) − ϕIF (t) = φ(t − t1 ) − φ(t − t2 )
+ ϕnb (t) − ϕnif (t),
Δϕ2 (t) = ϕIF (t + Δt) − ϕb (t)
= φ(t + Δt) − φ(t) + ϕnif (t + Δt) − ϕnb (t)
+ ϕLO (t) − ϕLO (t + Δt)
Under conditions that the linewidth of the light source is ultra-
narrow and the length of the TDF is negligible compared to the
coherent length of the light source, we can have approximations
below
ϕnb (t) − ϕnif (t) ≈ 0
ϕnif (t + Δt) − ϕnb (t) ≈ 0
ϕLO (t) − ϕLO (t + Δt) ≈ 0
Therefore, (7) and (8) can be rewritten as
Δϕ1 (t) = φ(t − t1 ) − φ(t − t2 )
Δϕ2 (t) = φ(t + Δt) − φ(t) = φ(t + t2 − t1 ) − φ(t) (13)
It is obvious that
Δϕ1 (t) = Δϕ2 (t − t2 )
By measuring the time delay between Δϕ1 (t) and Δϕ2 (t)
using correlation operations, the time t2 can be obtained. Thus,
the distance from point A to point B along path: A→2→B can
be calculated through L = ct2 /n, where c is the velocity of light
in vacuum and n is the refractive index of the fiber core. As the
total fiber length of the sensing system is already known, the
vibration position can thus be located.
III. EXPERIMENT SETUP
Figure 2 (a) shows the experimental setup of the ultra-long-
range DVS system. The output of an ultra-narrow-linewidth CW
tunable fiber laser (NKT X15 with 100 Hz linewidth) operating
at 1550.12 nm was split into two branches by a 50/50 coupler.
The lower branch was sent to the CoRx serving as the LO.
The upper branch was launched into the FUT directly. The
total length of the FUT was around 1230 km, containing 15
fiber spans. Standard single-mode fiber (SSMF, G.652D) was
used in all cases. An erbium-doped fiber amplifier (EDFA)
(Huawei C-band OAU) and a 4-nm bandpass filter (BPF) were
utilized in each span of the 1230-km fiber link to compensate
the loss induced by the fiber and to remove the out-of-band
ASE noise, respectively. Two 60-m fibers were wrapped around
the same PZT to cause the external vibrations. The modulation
constant of the PZT is 8.3 rad/V at 1550 nm, corresponding to
a strain-voltage coefficient of 23 nε/V . At the far end of the
fiber link, the light was split into two beams by a 90/10 coupler.
90 percent of light went through an AOM, which induced a 100
Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
2243
(6)
(7)
(8)
Fig. 2. (a) The configuration of the proposed system. AFG: arbitrary function
generator, SMF: single mode fiber, OC: optic coupler, AOM: acoustic-optic
modulator, PZT: piezoelectric ceramic transducer, BPF: band pass filter, CoRx:
coherent receiver; (b) Photo of the fiber link.
(9)
MHz frequency shift to the light, and a 5-km TDF. Meanwhile,
(10)
10 percent of the light passed directly. These two branches
(11) were then recombined and looped back using another fiber. To
investigate the sensing performance, two cases were considered
in this experiment: 1) vibrations occur at both the beginning and
(12) end of the fiber link, with the corresponding positions marked
as points A and B in black as shown in Fig. 2(a). In this case, the
distance from point A to point B is 1230 km; 2) vibrations occur
at the middle of the fiber link, and the corresponding positions
are marked as points A and B in red as shown in Fig. 2(a).
(14)
Meanwhile, the distance from point A to OC 2 and the distance
from point B to OC 3 are both 500 m. A polarization-diversity
CoRx, consisting of an optical 90° hybrid mixer (PHOTOP
C-band 2×8) and four balanced photodetectors (BPD, Finisar
BPDV2150RQ), was used for the detection of the received
sensing signals. The output signals from the BPDs were then
sampled by a real-time oscilloscope (Keysight DSAZ634A). The
sampling rate has a linear relationship with the spatial resolution:
the higher the sampling rate is, the higher the spatial resolution
can be obtained. Limited by the memory size of the oscilloscope,
the sampling rate was set to 400 MSa/s in our case. Offline DSP
was then used to analyze the collected data.
IV. EXPERIMENTAL RESULTS
Firstly, the localization performance when vibration events
occur at both the beginning and end of the fiber link is investi-
gated. Three different vibration signals are applied on the PZT
to emulate the practical vibrations with high, medium and low
frequencies. The first one is a Gauss white noise like vibration
signal with a frequency range in the order of tens of kHz (high
frequency case), which is generated using the noise function
of an arbitrary waveform generator (Agilent 33120A) with a
peak-to-peak voltage of 20 V. Because the frequency response of
PZT is not uniform, especially in the ultrahigh frequency region,
the actual vibration signal applied on the fiber is the Gauss white
noise signal superimposed with the frequency response of the
PZT. Another two electrical driving signals are user-defined sig-
nals with vibration frequency ranges of around 1 kHz (medium
 2244 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 39, NO. 7, APRIL 1, 2021

frequency case) and a few hundred Hz (low frequency case),

respectively. The peak to peak voltage of these two signals is 60
V. That means the fiber strain is 1.38 με. Fig. 3(a), (b) and (c)
show the electrical driving signals, demodulated phase curves,
constructed differential signals Δϕ1 (t)and Δϕ2 (t), and their
cross-correlation curves, respectively. Fig. 3(d) shows the spec-
tra calculated from the constructed differential signal Δϕ1 (t)
of three vibrations (blue lines) and spectra of the medium-
frequency and low-frequency electrical driving signals (red
lines). Before calculating the cross-correlation of Δϕ1 (t)and
Δϕ2 (t), a low pass digital filter was used to remove the noise.
Specifically, a 40-kHz bandwidth filter is used for the noise
like high-frequency vibration signal; a 2-kHz bandwidth filter is
used for the medium-frequency vibration signal and a 400 Hz
bandwidth filter is used for the low-frequency vibration signal.
From Fig. 3, the following observations can be made: i) External
vibrations induce obvious fluctuations on the phase curve. How-
ever, the background phase noise makes the phase curve slowly
fluctuate. ii) By constructing two differential signals: Δϕ1 (t)
and Δϕ2 (t), according to (12) and (13), the influence of the
phase noise can be significantly reduced. iii) The SNR decreases
with the decrease of frequency as shown in Fig. 3(d). For the high
frequency case, the SNR for the frequency component from 22
kHz to 25 kHz is 32 dB. For the medium frequency case, the
SNR is 24dB. For the low frequency case, the SNR is 16dB. One
factor is that the differential calculation is equivalent to a high
pass filter, which makes low-frequency signal weaker. Another
factor is the fiber link noise, which is inversely proportional to
the frequency [27], [28]. Here, the sampling rate is 400 MSa/s, so
the corresponding time resolution is 2.5 ns. From Fig. 3(a), the
time delay measured is 6.0575325 ms. Therefore, the fiber length
L = ct2 /n = 123537 m, where c = 299792458 m/s and n = 1.47.
In order to investigate the location accuracy of the sensing
system, 70 data sets averaged by 30 times were acquired for
each vibration. The localization accuracy is defined as the mea-
surement accuracy of the fiber length between two vibration
points along the fiber. Fig. 4 shows the localization results. One
can see that the calculated mean of the fiber length is 1235372
m with a difference of 3.5 m for the noise like high-frequency
vibrations. However, the calculated mean of the fiber length for
medium-frequency vibration is 1235413 m with a difference
of 126.9 m, and the calculated mean of the fiber length for
low-frequency vibration is 1235404 m with a difference of
2600.5 m. According to the theory of time delay estimation, the
SNR influences the location accuracy significantly. [29], [30] If
the SNR is lower than a threshold, the location error increases
dramatically with the reduction of SNR. Due to the differen-
tial effect as mentioned above, the SNR decreases when the
vibration frequency decreases. Besides, the effective bandwidth
and observation-time product (BT) of the differential signals
influences the location accuracy as well. The BT value should be
larger than 100 to ensure a good location estimation. Therefore, Fig. 3. Electrical driving signals, demodulated phase curves, constructed dif-
the proposed scheme has very high localization accuracy when ferential phase curves and the cross correlation curves when (a) high-frequency,
(b) medium-frequency and (c) low-frequency signals are applied at the beginning
vibration frequency is larger than 10 kHz, while the location and end of the fiber link. (d) The retrieved frequency spectra after FFT of the
accuracy decreases largely when lower vibration frequencies constructed differential phase signals when three vibration signals are applied
are considered. and spectra of medium-frequency and low-frequency electrical driving signals.

Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
YAN et al.: FORWARD TRANSMISSION BASED ULTRA-LONG DISTRIBUTED VIBRATION SENSING 2245

Fig. 4. Localization results for the 1230-km fiber link. STD: standard
deviation.

Since the localization in this work relies on the determination

of the time delay between two constructed differential signals,
it is essential to investigate the scenarios when the time delay
is small. To address this issue, we put the PZT at the middle
of the fiber link. The same vibration signals are used to drive
the PZT. Fig. 5(a) shows the cross-correlation curves of three
vibrations. It is easy to observe the time delay between two
correlation curves when vibration frequency is larger than 1 kHz,
even though the fiber length is very short. However, one can
see that for lower frequency vibrations, two correlation curves
almost overlap. For the medium frequency and low frequency
cases in our experiment, the decrease of the time delay between
two differential signals leads to the decrease of the effective
observation time, as shown in Fig. 5(c). As a result, the BT
value decreases, which may result in a degraded localization
performance, as shown in Fig. 5(b). The calculated mean of the
fiber length for high-frequency vibrations is 6095.8 m with a
difference of 2.7 m. The measurement result has accounted for
the time delay caused by 1 km sensing fiber, 5 km TDF, an
EDFA and an AOM. The calculated mean of the fiber length
for medium-frequency vibration is 6153.4 m with a difference
of 247.9 m. The calculated mean of the fiber length for low-
frequency vibration is 5737.3 m with a difference of 4225.1
m. Results confirm that less than 5 m location accuracy can be
obtained when vibration frequency is larger than 10kHz. The
location accuracy for vibrations of around 1 kHz is less than
250 m. The location accuracy for vibrations of a few hundred
Hz is less than 5 km.
Fig. 5. (a) Cross correlation curves and (b) location results when three different
In order to investigate the lower bound of the frequency vibrations are applied near the middle of the fiber link. (c) comparisons of the
response of the sensing system, sinusoidal vibrations of different observation time of low frequency signals and medium frequency signals when
frequencies with a peak-to-peak voltage of 30 V were applied at vibrations are applied at the beginning and end of the fiber link (1230 km) and
at the middle of the fiber link (6 km), respectively.
the beginning and end of the fiber link. It is worth noting that both
the demodulated phase signals and the constructed differential
phase signals can be used to analyze the frequency component frequency can decrease to 5 Hz with an SNR of more than 10 dB.
of the vibration in our system. For example, referring to (4) or The phase noise of the fiber link was measured by removing the
(5), the measured phase variation caused by external vibrations vibration from the fiber link as well. It is obvious that there exist
is actually the sum of φ(t) and φ(t − τ ), where τ is the time for several resonance peaks in the spectrum of the phase noise. We
light to propagate from point A to point B along the transmission found it was due to the noises induced by the fans of two EDFA
path. Therefore, one can directly use the demodulated phase of cluster mainframes we used in our experiment. As shown in
the light to determine low-frequency components of vibrations, Fig. 6(b), the phase noise of the fiber link with different lengths
and the measured results are shown in Fig. 6(a). The detectable was investigated. The resonance peaks exist in the spectra of the

Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
 2246
Fig. 6. (a) The retrieved frequency spectra after FFT of the phase signals
when vibrations with different frequencies are applied at the beginning and end
of the fiber link; (b) the retrieved background noise spectra of fiber link with
different lengths after FFT of the phase signals with (solid line) or without
(dashed line) HUAWEI EDFAs; (c) The retrieved frequency spectrum after FFT
of the demodulated phase of the intermediate signal when noise like vibration
is applied at the middle of the fiber link.
phase noise (shown as solid lines in Fig. 6 (b)) until the EDFA
modules are removed (shown as the dashed lines in Fig. 6(b)).
In this sensing scheme, sub Hz frequency detection can be
reached by using ultra-stable lasers [4]. However, the sum of
φ(t) and φ(t − τ ) can be seen as a filter with a free spectral
range (FSR) determined by τ , which induces null frequency
points in the frequency spectrum, as shown in Fig. 6(c). The
existence of null frequencies makes the detection of vibrations
with certain frequencies invalid. Besides, the interval of the null
frequency points Δfnull,sum = 1/τ (Please see the Appendix).
That means the interval of null frequency points will be very
small if τ is very big. The null frequency is a problem for sensors
that only using the sum of φ(t) and φ(t − τ ) for detection. For
Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 39, NO. 7, APRIL 1, 2021
the differential phase signal, the null frequencies can be written
as fnull,dif f = N/Δt. Since the length of TDF is fixed, the null
frequencies of the differential phase signals are fixed. Except that
vibrations occur at some special locations, the null frequency
point cannot simultaneously exist in frequency spectra of ϕb (t)
(or ϕIF (t)) and Δϕ2 (t) (or Δϕ1 (t)) (Please see the Appendix).
Therefore, the occurrence of the null frequency points is largely
reduced in this sensing scheme, which makes the detection of
vibration frequencies much more reliable.
V. DISCUSSIONS
In this sensing scheme, it is necessary to calibrate the total
length of the FUT prior to deployment. We measured the lengths
of the 15 fiber spans using an OTDR with ±6m distance accuracy
(100-km range). The measured lengths of the 15 fiber spans
are 70.509 km, 75.377 km, 75.811 km, 68.859 km, 90.468 km,
92.554 km, 90.506 km, 99.134 km, 70.512 km, 100.951 km,
90.435 km, 69.121 km, 69.794 km, 90.959 km, and 74.805 km,
respectively. Therefore, the total length of these 15 fiber spans
is 1229795 km. However, in this sensing system, EDFAs, BPFs,
and VOAs are used in each span. In particular, EDFAs typically
use 10-50 m of EDF for optical amplification and it may be
difficult to measure the exact length due to the optical isolators
inside the EDFAs. In our experiment, the computed total length
of the FUT is around 1235.372 km, which has accounted for the
lengths of all the EDFAs, BPFs, VOAs, TDF, AOM, couplers
and fiber jumpers. The proposed DVS uses a fiber pair as the
sensing part and hence the sensing range is half of the length
of sensing fiber while the spatial resolution is defined as half
of the location accuracy. That means the sensing range in this
experiment is 615 km. The spatial resolution is less than 2.5 m for
vibration frequency of larger than 10kHz. The spatial resolution
for vibrations of around 1 kHz is less than 125 m, and the spatial
resolution for vibrations of a few hundred Hz is less than 2.5 km.
Real-time operation is key to practical sensing applications.
The differential structure effectively mitigates most of the car-
rier phase noise appended to the sensing signal, resulting in
two clear, quasi noise free differential phase patterns with
a certain time delay (see Fig. 3). This feature may provide
great potential in the practical deployment of ultra-long range
DVS systems. Low-frequency detection is also important in
practical distributed vibration sensing applications due to the
following reasons: 1) many of the real-life events are within
the low-frequency range; 2) for submarine cables and buried
terrestrial cables, only low-frequency signals could be detected
due to rapid attenuation of the high-frequency vibrations. In
this work, however, a differential time-delay structure is utilized
at the far-end of the FUT, which serves as a high-pass filter,
making it a challenge to detect low-frequency vibrations with
high spatial resolution. This differential effect can be weakened
by increasing the length of the TDF, according to (16). The
location accuracy of low-frequency vibrations can be improved
accordingly. It is also worth noting that the length of TDF is
constrained by the system phase noise. If the length of TDF is
too long, the coherence between two differential signals will be
deteriorated as the phase noise is no longer negligible, leading to
YAN et al.: FORWARD TRANSMISSION BASED ULTRA-LONG DISTRIBUTED VIBRATION SENSING 2247

an inaccurate localization. Therefore, an ultra-stable laser with

ultra-narrow linewidth [4] helps to extend the length of TDF by
reducing the carrier phase noise. In practical applications, the
length of TDF and the linewidth of the light source should be
determined according to the sensing distance, frequency range
and required spatial resolution. Besides, for a given sensing
system, the SNR of a sensing signal depends on the amplitude
and frequency of the external perturbation. In this work, the
perturbation amplitude is linearly related to the product of
the voltage applied on the PZT and the fiber length wrapped
around the PZT. To guarantee a good SNR of the differential
signal for low frequency cases, a peak to peak voltage of 60
V was applied. If the peak to peak voltage applied on the PZT
decreases to 15 V, which corresponding to a strain of 345 nεe
applied on the 60-m fiber, the SNR will be 10 dB for the low
frequency case. Therefore, better sensing performance can be
obtained if the vibration signal has larger amplitude. Meanwhile,
longer effective signal observation time also contributes to better
location accuracy. All these make the proposed sensor useful for
some infrasound vibration detections.
In addition, the sensor has excellent localization performance
of vibrations containing larger than 1 kHz frequency component.
In real applications, events like train running on a railway track,
knocking and hammering will induce vibrations of more than
1 kHz on the sensing fiber. That means the proposed sensing
scheme has powerful application potential for these sensing
cases.
For the proposed DVS, the vibration is located using the
differential phase information. However, due to the differential
effect, the differential phase signals have low sensitivity to
low-frequency vibrations. Therefore, the demodulated phase
signals, containing the cumulative phase information that is Fig. 7. (a) Two constructed differential phase curves and (b) the cross corre-
much more sensitive to low-frequency vibrations, are used to lation curve when two independent noise vibration signals are applied at two
extend the system response to lower frequencies. Besides, since different locations of the fiber link. (c) a zoomed section of (b).
the vibration signal is continuously sampled, the upper bound
of the detectable frequency is only limited by the bandwidth of and the location interval is not too small. A proof of concept
the photodetector and the sampling rate of the analog-to-digital experiment on multi-point sensing was also conducted. Two
convertor (ADC). Therefore, ultrabroad frequency response can PZTs, which are driven by two random noise signals gener-
be realized in such a sensing system. ated separately by two arbitrary waveform generators (Agilent
In the real environment, multiple channels based on wave- 33120A and BK Precisions 4065B) both with peak to peak
length division multiplexing (WDM) can be carried on such voltages of 20 V, are placed at different positions (60 km
an ultralong fiber link. In that case, the bandwidth of the BPF apart). Experimental results of these two independent vibrations
should be increased to cover all the channels, which means the occurring at different places simultaneously are shown in the
increase of in-band ASE noise in terms of the sensing signal. Fig. 7. The sampling rate was set as 400 MSa/s. It is obvious that
The in-band ASE noise will be repeatedly amplified during there are two peaks in the cross-correlation curve. The measured
forward transmission. Therefore, the optical SNR degrades with time delay is 4.5592775 ms and 3.9676825 ms, corresponding
the increase of the sensing distance, which sets an upper limit to fiber lengths of 929821 m and 809171 m, respectively. The
of the sensing distance. However, due to CW operation, the difference of fiber length is 120650 m, which means the distance
bandwidth of the sensing signals can be very small, which helps between two vibration locations is 60325 m. With the decrease
to maintain an excellent SNR at receiver, making it possible to of interval between two vibration locations, peaks shown in
further extend the sensing range. the cross-correlation curve will get closer and finally cannot
In this work, the sensing performance of single point vibration be distinguished. Further work needs to be done to analyze the
was investigated. In real systems, multi-point vibrations are in- factors, for example, the distance interval between two points,
evitable to occur simultaneously over such an ultralong sensing the intensities of multiple vibrations, the correlativity of these
range. Since the cross-correlation based time delay estimation vibrations, etc. that influence the sensing performance.
is used to determine the locations, multi vibration points can There are several advantages of the proposed scheme. Com-
be distinguished in case these vibration signals are uncorrelated pared with the famous OTDR technique, the upper limit of

Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
 2248
the high-speed vibration detection is largely extended. The
intensity of the forward transmitting signal is much stronger
than the backscattered signal, and forward amplification can
be easily realized using EDFAs, which enables the ultra-long
haul sensing. In the OTDR based sensing systems, the phase
unwrapping failure is a problem if the vibration has both high
frequency and large amplitude. Such failure is no more a problem
in our proposed scheme as the phase of the light is continu-
ously sampled and retrieved. Compared with the interferometric
DVSs, the unidirectional forward transmission is used in our
sensing system, which avoids the Rayleigh backscattering noise
problem. Besides, for intensity detection based interferometric
sensors, since the intensity signal is not linearly related to the
external vibration, which may cause undesirable higher-order
harmonics, sometimes it is difficult to measure the vibration
correctly. However, the proposed sensing scheme uses the phase
of light, which is proportional to the vibration amplitude, to
detect the external vibration. Polarization fading effect can be
avoided because polarization diversity is used, and the eliminat-
ing of the reference arm reduces the requirement of the system
stability. Moreover, in this sensing scheme, thanks to the use of
cumulative and differential phase information simultaneously,
an ultra-broad frequency range can be realized with the number
of null frequency points largely reduced.
VI. CONCLUSION
We have demonstrated a novel optical vibration sensing tech-
nique based on forward transmission and coherent detection
using a pair of optical fibers deployed at the same location with
a loop-back configuration. An FS-ODL structure with an AOM
and a TDF is used for event localization. Ultra-broad-frequency
detection from infrasound to ultrasound can be realized in this
sensing scheme. Less than 125 m spatial resolution over 615 km
sensing range can be obtained when vibrations of larger than
1 kHz occur on the fiber link. Compared with previously reported
distributed vibration sensors, the proposed scheme has a much
simpler structure and enables ultra-long distributed vibration
sensing. We anticipate that the proposed scheme could be used
to dramatically improve the sensing range and performance of
the DVS systems, and thus enable a new wave of ultra-precise,
ultra-long-haul vibration sensing applications.
APPENDIX
In case a sinusoidal vibration is applied on the fiber, the
corresponding modulated phase term can be written as φ(t) =
φ0 sin(2πfvib t), where φ0 is the amplitude of the phase change
and fvib is the frequency of the vibration signal. As a result, for
the received signals, the phase change caused by the external
vibrations can be written as
φ(t) = φ0 sin(2πfvib t) + φ0 sin(2πfvib (t − τ ))
= 2φ0 sin(2πfvib (t − τ /2)) cos(πfvib τ ).
When the following condition fvib τ = M + 1/2 , M is an in-
teger, is satisfied, the phase change caused by external vibrations
will be zero. That means in the frequency spectrum of the phase
signal, there exist a series of null frequency points, which satisfy
Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.
JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 39, NO. 7, APRIL 1, 2021
fnull,sum = (M + 1/2)/τ . The frequency interval between two
consecutive null points is Δfnull,sum = 1/τ .
However, for the constructed differential signal Δϕ2 (t), it can
be written as
Δϕ2 (t) = φ(t + Δt) − φ(t)
= φ0 sin(2πfvib (t + Δt)) − φ0 sin(2πfvib t)
= 2φ0 cos(2πfvib (t + Δt/2)) sin(πfvib Δt). (16)
When fvib Δt = N , N is an integer, Δϕ2 (t) will be zero. That
means in the frequency spectrum of the constructed differential
signal, there also exist a series of null frequency points, which
satisfy fnull,dif f = N/Δt. The frequency interval between two
consecutive null points is Δfnull,sum = 1/Δt.
The null frequency points can simultaneously exist in the
frequency spectrum of the phase signal and the frequency spec-
trum of the constructed differential signal when the following
conditions are satisfied.
(M + 1/2)/τ = N/Δt
→ (M + 1/2)/(nL/c) = N/Δt
→ L = (M + 1/2)cΔt/nN (17)
REFERENCES
[1] X. Liu, B. Jin, Q. Bai, Y. Wang, D. Wang, and Y. Wang, “Distributed fiber-
optic sensors for vibration detection,” Sensors, vol. 16, no. 8, Jul. 2016,
Art. no. 1164.
[2] J. M. Lopez-Higuera, L. Rodriguez Cobo, A. Quintela Incera, and A. Cobo,
“Fiber optic sensors in structural health monitoring,” J. Lightw. Technol.,
vol. 29, no. 4, pp. 587–608, Feb. 2011.
[3] X. Li, Q. Sun, J. Wo, M. Zhang, and D. Liu, “Hybrid TDM/WDM-based
fiber-optic sensor network for perimeter intrusion detection,” J. Lightw.
Technol., vol. 30, no. 8, pp. 1113–1119, Apr. 2012.
[4] G. Marra et al., “Ultrastable laser interferometry for earthquake detec-
tion with terrestrial and submarine cables,” Science, vol. 361, no. 6401,
pp. 486–496, Aug. 2018.
[5] J. C. Juarez, E. W. Maier, K. N. Choi, and H. F. Taylor, “Distributed
fiber optic intrusion sensor system,” J. Lightw. Technol., vol. 23, no. 6,
pp. 2081–2087, Jun. 2005.
[6] R. Zinsou et al., “Recent progress in the performance enhancement
of phase-sensitive OTDR vibration sensing systems,” Sensors (Basel),
vol. 19, no. 7, Apr. 2019.
[7] E. Awwad, C. Dorize, S. Guerrier, and J. Renaudier, “Detection-
localization-identification of vibrations over long distance SSMF with
coherent ΔΦ-OTDR,” J. Lightw. Technol., vol. 38, no. 12, pp. 3089–3095,
2020,.
[8] L. Palmieri, T. Geisler, and A. Galtarossa, “Limits of applicability
of polarization sensitive reflectometry,” Opt. Exp., vol. 19, no. 11,
pp. 10874–10879,May. 2011.
[9] C. Cao et al., “Suppression of signal fading with multi-wavelength laser
in polarization OTDR,” IEEE Photon. Technol. Lett., vol. 29, no. 21,
pp. 1824–1827, 1 Nov. 2017.
[10] Y. Peled, A. Motil, L. Yaron, and M. Tur, “Slope-assisted fast distributed
sensing in optical fibers with arbitrary brillouin profile,” Opt. Exp., vol. 19,
no. 21, pp. 19845–19854, Sep. 2011.
[11] Q. Song, H. Peng, S. Zhou, P. Zhou, Q. Xiao, and B. Jia, “A novel weak-
scattering michelson interferometer based on PBS for long-distance dis-
turbance localization,” J. Lightw. Technol., vol. 38, no. 8, ppp. 1543–1549,
(15) Nov. 2019.
[12] B. Kizlik, “Fibre optic distributed sensor in mach-zehnder interferometer
configuration,” in Proc. Int. Conf. Mod. Problems Radio Eng. Telecommun.
Comput. Sci., Lviv-Slavsko, Ukraine, 2002, pp. 128–130.
[13] C. Ma et al., “Long range distributed fiber vibration sensor using an asym-
metric dual mach–zehnder interferometers,” J. Lightw. Technol., vol. 34,
no. 9, pp. 2235–2239, May 2016.
YAN et al.: FORWARD TRANSMISSION BASED ULTRA-LONG DISTRIBUTED VIBRATION SENSING 2249

[14] F. Teng, D. Yi, X. M. Hong, and X. J. Li, “Distributed fiber optics
disturbance sensor using a dual-Sagnac interferometer,” Opt. Lett., vol. 44,
no. 20, pp. 5101–5103, Oct. 2019.
[15] K. Wada, H. Narui, D. Yamamoto, T. Matsuyama, and H. Horinaka,
“Balanced polarization maintaining fiber sagnac interferometer vibration
sensor,” Opt. Exp., vol. 19, pp. 21467–21474, Oct. 2011.
[16] Y. Li, Z. Liu, Y. Liu, L. Ma, Z. Tan, and S. Jian, “Interferometric vibration
sensor using phase-generated carrier method,” Appl. Opt., vol. 52, no. 25,
pp. 6359–6363, Aug. 2013.
[17] Z. Wang et al., “Phase-sensitive optical time-domain reflectometry with
brillouin amplification,” Opt. Lett., vol. 39, no. 15, pp. 4313–4316, 2014.
[18] Z. N. Wang et al., “Ultra-long phase-sensitive OTDR with hybrid
distributed amplification,” Opt. Lett., vol. 39, no. 20, pp. 5866–5869,
Oct. 2014.
[19] H. F. Martins, S. Martin-Lopez, P. Corredera, J. D. Ania-Castañon, O.
Frazão, and M. Gonzalez-Herraez, “Distributed vibration sensing over 125
km with enhanced SNR using phi-OTDR over a URFL cavity,” J. Lightw.
Technol., vol. 33, no. 12, pp. 2628–2632, Jun. 2015.
[20] D. Iida, K. Toge, and T. Manabe, “High-frequency distributed acoustic
sensing faster than repetition limit with frequency-multiplexed phase-
OTDR,” in Proc. Opt. Fiber Commun. Conf. Exhib., Anaheim, CA, USA,
2016, Paper M2D.
[21] D. Chen, Q. Liu, X. Fan, and Z. He, “Distributed fiber-optic acoustic
sensor with enhanced response bandwidth and high signal-to-noise ratio,”
J. Light. Technol., vol. 35, no. 10, pp. 2037–2043, May 2017.
[22] Q. Chen et al., “A distributed fiber vibration sensor utilizing dispersion
induced walk-off effect in a unidirectional mach-zehnder interferometer,”
Opt. Exp., vol. 32, no. 3, pp. 2167–2173, Feb. 2014.
[23] S. Xie, Q. Zou, L. Wang, M. Zhang, Y. Li, and Y. Liao, “Positioning error
prediction theory for dual mach-zehnder interferometric vibration sensor,”
J. Lightw. Technol., vol. 29, no. 3, pp. 362–368, Feb. 2010.
[24] H. Wang et al., “Improved location algorithm for multiple intrusions
in distributed sagnac fiber sensing system,” Opt Exp, vol. 22, no. 7,
pp. 7587–7597, Apr. 2014.
[25] Q. Sun, D. Liu, J. Wang, and H. Liu, “Distributed fiber-optic vibration
sensor using a ring mach-zehnder interferometer,” Opt. Commun., vol. 281,
no. 6, pp. 1538–1544, Dec. 2008.
[26] K. Kikuchi, “Coherent optical communication systems,” in Optical Fiber
Telecommunications VB, The Netherlands, Elsevier, 2008, pp. 95–129.
[27] J. Dong, J. Huang, T. Li, and L. Liu, “Observation of fundamental thermal
noise in optical fibers down to infrasonic frequencies,” Appl. Phys. Lett.,
vol. 108, no. 2, Jan. 2016, Art. no. 021108.
[28] L. Duan, “Thermal noise-limited fiber-optic sensing at infrasonic frequen-
cies,” IEEE J. Quantum Electron., vol. 51, no. 2, pp. 1–6, Feb. 2015.
[29] J. Ianniello, “Time delay estimation via cross-correlation in the presence
of large estimation errors,” IEEE Trans. Acoust., Speech Signal Process.,
vol. ASSP-30, no. 6, pp. 998–1003, Dec. 1982.
[30] S. Xie, Q. Zou, L. Wang, M. Zhang, Y. Li, and Y. Liao, “Positioning error
prediction theory for dual mach-zehnder interferometric vibration sensor,”
J. Light. Technol., vol. 29, no. 3, pp. 362–368, Feb. 2011.
Yaxi Yan received the B.Eng. degree from the Harbin Institute of Technology,
Harbin, China, in 2014 and is currently working toward the Ph.D. degree in
distributed optical fiber sensing with the Hong Kong Polytechnic University,
Kowloon, Hong Kong. She worked as a Research Assistant with the Hong Kong
Polytechnic University in 2015 and 2016.
Bin Zhou received the B.S. and Ph.D. degrees from Zhejiang University,
Hangzhou, China, in 2006 and 2011, respectively. He is currently an Associate
Professor with South China Normal University. His research interests include
the areas of optical sensing, optical devices, and nonlinear optics.
Alan Pak Tao Lau received the B.A.Sc. degree in engineering science (electrical
option) and the M.A.Sc. degree in electrical and computer engineering from the
University of Toronto, Toronto, ON, Canada, in 2003 and 2004 respectively. He
received the Ph.D. degree in electrical engineering from Stanford University,
CA, USA, in 2008. He worked on source coding techniques for wireless com-
munication systems utilizing multiuser diversity for the masters’ thesis research.
He was with the Electrical Engineering Department, the Hong Kong Polytechnic
University as an Assistant Professor in 2008. He also worked with NEC Labs
America in 2006 as a Research Associate in the Optical Networking Division.
His current research interests include system designs and signal processing
techniques for coherent fiber-optic communication systems and multi-mode
fiber communication systems. He was the recipient of the numerous government
awards during his postgraduate study at Stanford University.
Chao Lu (Fellow, IEEE) received the B.Eng. degree in electronic engineering
from Tsinghua University, Beijing, China, in 1985, and the M.Sc. and Ph.D.
degrees from the University of Manchester, Manchester, U.K., in 1987 and 1990,
respectively. In 1991, he joined, as a Lecturer, with the School of Electrical
and Electronic Engineering, Nanyang Technological University, Singapore,
where he has been an Associate Professor since January 1999. From 2002
to 2005, he was seconded with the Institute for Infocomm Research, Agency
for Science, Technology and Research, Singapore, as a Program Director and
Department Manager, helping to establish a research group in the area of
optical communication and fiber devices. Since April 2006, he has been a
Professor with the Department of Electronic and Information Engineering, The
Hong Kong Polytechnic University, Hong Kong. His current research interests
include optical communication systems and networks, fiber devices for optical
communication, and sensor systems.
Changjian Guo received the B.Eng. degree from the Dalian University of
Technology, China, in 2006, and the Ph.D. degree from Zhejiang University,
Hangzhou, China, in 2011. He is currently an Associate Professor with South
China Normal University, Guangzhou, China. His research interests include
optical communication and digital signal processing.

Faisal Nadeem Khan was born in Jhang, Pakistan. He received the B.Sc. degree
in electrical engineering from the University of Engineering and Technology,
Taxila, Pakistan, the M.Sc. degree in communications technology from the
University of Ulm, Ulm, Germany, and the Ph.D. degree in electronic and
information engineering from The Hong Kong Polytechnic University, Hong
Kong. From 2012 to 2015, he was a Senior Lecturer with the School of
Electrical and Electronic Engineering, Universiti Sains Malaysia. He is currently
with the Photonics Research Centre, The Hong Kong Polytechnic University.
He has authored or coauthored more than 50 research papers in prestigious
international journals and conferences as well as written one book chapter. His
research interests include machine learning and signal processing techniques
for high-speed fiber-optic communication systems. He has been an Invited
Speaker with various prestigious international conferences including Optical
Fiber Communication 2018 and Signal Processing in Photonic Communications
2017, among others.

Authorized licensed use limited to: POLO BIBLIOTECARIO DI INGEGNERIA. Downloaded on February 15,2023 at 14:52:58 UTC from IEEE Xplore. Restrictions apply.