Coherent OTDR Using Flexible All-Digital Orthogonal Phase Code Pulse for Distributed Sensing

A coherent optical time-domain reflectometer (COTDR) using flexible all-digital orthogonal phase code pulse is proposed for distributed acoustic sensing. All-digital orthogonal phase code pulse with frequency shift and time shift is used as probe. Coherent detection and balance photodetector are used to amplify interference signal and get rid of its direct-current (DC) component. The new scheme needs only single channel detection while keeping sampling frequency. Amplitude triangle modulation and frequency linear sweep modulation waveform are used for system performance investigation. The experiments on 15.4 km optical fiber showed that waveform information can be recovered well. The all-digital orthogonal phase code pulse will provide a flexible solution for different application requirement.


I. INTRODUCTION
Distributed acoustic sensing (DAS) system has a wide application on intrusion detection [1], seismic waves measurements [2], structure health monitoring [3] and pipeline monitoring [4], and so on. Comparing to optical frequency domain reflectometry (OFDR) [5]- [8], phase-sensitive optical time domain reflectometry (OTDR) based on Rayleigh backscattering (RBS) is a more common configuration for DAS. Quantitative waveform recovery requirement of DAS makes high performance phase extraction process play an important role to provide a linear response [9].
Phase extraction is usually realized with the phase generated carrier (PGC) [10], [11], the path-matched interferometry [12], and phase diversity detection [13], [14]. Among them, phase diversity detection is a promising DAS phase demodulation method for its ability to eliminate the common The associate editor coordinating the review of this manuscript and approving it for publication was Cesar Vargas-Rosales . mode noise generated on the sensing fiber. A. Masoudi et al. used a 3 × 3 coupler to construct an unbalanced Michelson interferometer and output three interference signals with 120 • phase shift in sequence [15]. J. Jiang et al. realized coherent phase diversity detection by utilizing a 3 × 3 coupler and pulse pair containing two pulses with difference frequency [16]. But the perturbation on the two arms of the interferometer may produce phase noise and reduce signal to noise ratio (SNR). The spatial resolution is also fixed by the optical path difference of interferometer arms. Three channels simultaneous high-speed detection also increase the system data acquisition hardware difficulty and cost. Z. Wang et al. reduce the simultaneous detection to two channels by utilizing a 90-degree hybrid, which realizes a 90 • phase shift between two interference signals [17]. A. E. Alekseev et al. proposed a single-channel detection by combining an intensity modulator and a phase modulator to realize differential phase-shift keying modulation [18]. Three probe pulses are injected into the sensing fiber in sequence to VOLUME 8, 2020 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ emulate analog phase diversity demodulation. The first part phase of pulses are zeros while the second part of pulses are delayed by relative phase shift 2π/3.However, the obtained three interference signals are non-simultaneous detection and the acoustic-induced phase itself may vary between them. In addition, its measurement bandwidth is also reduced by 2/3. In order to overcome the problem of measurement bandwidth, A. E. Alekseev et al. then adapted a dual pulse and heterodyne detection method with each pulse having different carrier frequency [19]. The phase is demodulated through a I/Q quadrature demodulator with two analog harmonic radio signals. Experiment was demonstrated with PZT transducer being locating at the 1 km from the beginning of sensing fiber. X. He et al proposed a similar method with a digital I/Q quadrature demodulating algorithm [20]. They used two acousto-optic modulators and a delay fiber to generate heterodyne dual pulse. The demodulation experiment was demonstrated on 0.47 km sensing fiber and spatial resolution is fixed by the delay fiber. Y. Muanenda et al. combined the dual pulse with phase generated carrier technique [21]. They used an acousto-optic modulator, a phase modulator and a delay fiber to generate homodyne dual pulse. The demodulation experiment was demonstrated on 1 km sensing fiber and the 2kHz frequency vale information was retrieved correctly, however, the recovered waveform fidelity was low. Direct detection is adapted by the above dual-pulse-based phase extraction schemes. Thus, the sensing fiber length will be limited by weak backscattering signal. Y. Shan et al. combined ultra-weak fiber Bragg grating array (UWFBG) and distributed pump using a Raman fiber laser to enhance the light signal intensity, which helped the experiment demonstration on 42 km sensing fiber. However, the spatial resolution is fixed by the interval of UWFBG and prevents the possibility from using the dark optical fibers deployed for telecommunication.
In this paper, we proposed a coherent OTDR using flexible all-digital orthogonal phase code pulse for distributed sensing. We use a DPMZM to realize flexible all-digital orthogonal pulse with frequency shift and time shift. Coherent detection and balance photodetector are used to coherently amplify the RBS signals and get rid of the DC component of interference. The new scheme only needs a single channel detection while keeping sampling frequency. Orthogonal signals are obtained by digital mixing. We carried out experiment on 15.4 km sensing optical fiber with 10 m optical fiber wrapped on a piezoelectric ceramics (PZT). Modulation signals with amplitude triangle envelopment modulation and frequency linear sweep modulation are loaded on the PZT. The experiment results showed that applied modulation signal waveforms are recovered well. In addition, all-digital orthogonal phase code pulse provides a flexible solution for different requirements from potential applications, such as spatial resolution, signal intensity adjusting related with spatial interval, and so on.

II. PRINCIPLE
As shown in Fig. 1, the light is modulated into digital orthogonal phase code pulse, which is a triple-pulse contain three sub-pulses locating with frequency shift f 1 , f 2 and f 3 , and time shift W 0 . The initial phases of the three sub-pulses are coded with 0 • , 0 • and 90 • . The pulse width of the three sub-pulses are all W . The two sub-pulses with frequencies f 1 and f 3 emitted simultaneously while another sub-pulse with frequency f 2 emitted W 0 + W earlier than them. The triplepulses are emitted with T repetition time.
The triple-pulses can be generated with a dual parallel Mach-Zehnder modulator (DPMZM) driven by two voltage waveforms expressed as: where V D is the amplitude of the voltage signal which need be much less than the half-wave voltage V π for linearly modulation [22] The coherent detection scheme is shown in Fig. 2. The light with an optical frequency of f 0 from a narrowband continuous wave laser goes through a 1 × 2 fiber coupler. The light is divided into the local reference light and the signal light. The local reference light goes through a polarization controller and enters the 2 × 2 fiber coupler. The signal light goes into a   DPMZM which is driven by an arbitrary waveform generator (AWG). The triple-pulses are generated and then amplified by the Erbium-doped fiber amplifier (EDFA). Then the amplified pulses are injected into the sensing fiber through the circulator. When the pulses are transmitted in the sensing fiber, the RBS is generated at each scattering points in the fiber. The generated RBS light comes back along the sensing fiber, goes through the circulator and interfere with the local reference light in the 2 × 2 fiber coupler. The interference light signals are received by the photoelectric balance detector (PBD) and transformed into electronical signal. DC component of interference signal is discarded by PBD and the alternating component is sent to the data acquisition (DAQ) card. The collected data is transmitted to the computer and processed.
The carrier frequencies of the interference signal are equal to the shift frequencies f 1 , f 2 and f 3 . And the phases of the coherent OTDR signals are the difference value between the phase corresponding to position of sensing fiber and the local reference fiber phase. Thus, the signal can be expressed as: where, f 1 , f 2 and f 3 are the carrier frequencies of the coherent OTDR signals, for simplifying we let Z is the coordinate of the scattering point of sub-pulses with f 1 and f 3 . φ Loc is the phase of the local reference light. φ 1 (Z ), φ 2 (Z + L), φ 3 (Z ) are the phase corresponding to the position Z and Z + L on the sensing fiber. L is corresponding to the pulse interval W 0 . Fig. 3 shows the demodulation procedure. The signal is filtered by three bandpass filters to be decomposed into three signals: The envelopes of S 1 and S 3 are removed and the results are then mixed with S 2 respectively. The phase difference induced by frequency shift can be neglected since the shift frequency is far smaller than optical frequency. Therefore, the orthogonal signal I and Q can be obtained after the lowpass filters: where, n is the fiber effective refractive index; λ is the vacuum wavelength of the laser; (Z ) is the phase difference VOLUME 8, 2020 corresponding to the distance L. (Z ) is extracted by finding the angel of complex number Q + iI and minus 2π ft. When strain ε induced by acoustic event is applied to the fiber between the Z and Z + L position point, tiny change in fiber length and fiber refractive index are introduced. The differential phase variation (Z ) is proportional to the strain ε by: (10) where µ the material Poisson's ratio; p 11 , p 12 are elements of elastic-optic coefficient matrix.

A. SYSTEM SETUP
We established experiment setup according to Fig.2. The narrowband laser generates CW light with 1550.12nm wavelength, less than 3kHz linewidth and 40mW output power. The bandwidth of DPMZM (Photoline MXIQ-LN-40) is 40GHz. The rise time of this modulator is less than 25ps, which is negligible compared to the light pulse width. The dual-channel arbitrary waveform generator (AT-AWG-GS 2500) has 14-bit vertical resolution, 2.5GS/s sampling rate. The frequency shift of the sub-pulses are set as f 1 = 46MHz, f 2 = 65MHz and f 3 = 84MHz for satisfied the effective spectrum range. The sub-pulse width is W = 100ns and the interval time is W 0 = 100ns. The repeat period of digital orthogonal phase code pulse is T = 200µs. Thus, the sample rate of distributed sensing is 5kHz. A 10-meter fiber wrapped on a PZT is utilized for simulating the strain perturbation caused by acoustic field waveform. The 10-meter fiber is connected to the end of a 15.4km long optical fiber. A function waveform generator (Agilent33521A) generates voltage signals to drive the PZT. The working frequency range of the BPD is from 30kHz to 150MHz. And the data is sampled by a DAQ card with 1GS/s sampling rate and 12-bit resolution.

B. EXPERIMENTAL RESULTS
An amplitude modulation (AM) signal is applied to the 10-meter fiber. The carrier signal is a sinusoidal signal with fixed frequency of 200Hz and the amplitude of 1V. The amplitude modulation signal is a triangular-wave with the frequency of 10Hz and the AM ratio of 10%∼100%, which changes the waveform amplitude from 0.1V to 1V. Fig. 4(a) showed the spatial-temporal domain retrieving result. And an AM sinusoid waveform can be recognized at 15.4km. The location information and waveform of signal are detected correctly. The detail recovered waveform is shown in Fig. 4(b), the orange discrete points are the measured results and the green curve is the fitting curve of the retrieved waveform. The waveform is recovered well with fitting R-squared value 0.9939 and root mean squared error 0.0625. The zoomed local graph in Fig. 4(c) gives a more direct observation with retrieved signal waveform having little distortion. Fig.4(d) gives the power spectral density of recovered signal. The measured carrier frequency and the AM frequency of the fitting curve are 200.1Hz and 10.0Hz, which are consistent with the loaded signal. The SNR achieves 19.4dB. The center frequency in the spectrum of the AM signal is the carrier frequency, while the space between this center line to the nearest lines on left or right is the AM frequency.
In the practical application scenario of distributed acoustic sensing, the acoustic frequency will provide important information for calculating the velocity of a moving sound source based on the sound Doppler effect. We then investigated frequency detection performance of the system A signal with fixed amplitude and linear frequency modulation is loaded to the sensing fiber. The fixed amplitude is 1V. The frequency sweep from 100Hz to 500Hz in 400ms linearly. The time-frequency analysis diagram of the retrieved waveform instant center frequency at 15.4km is shown in Fig. 5(a), from which obvious periodical frequency linear variation can be seen. The repetition period can be recognized as 400ms which is consistent with the signal loaded. The length of the hamming window used in short-time Fourier transform is 256 points while the step is 1 point. The measured center frequency and the fitting curving are showed in Fig5(b).
The stepped increasing blue line shows the measured center frequency over time, while the red line shows the linear fitting curving. The interpolate frequency resolution 4.9Hz is limited by the Fourier transform length 1024 points when executing short-time Fourier transform. As shown in Figure 5(b), there are obvious deviations from the set frequency nearby the 100 Hz and 500Hz. The deviations from the set frequency nearby the 100 Hz and 500Hz are caused by the discontinuity between front part and end part of frequency sweeping cycles, i.e., when hamming window move to the location with frequency being round 100Hz or 500Hz, the discontinuous point exist in the intercepted window signal. Excluding the data in the front part and end part of the frequency sweeping cycles,