Seafloor acoustic sensor arrays for geophysical monitoring and exploration—as well as for marine, wide-area, and perimeter surveillance—all require long-distance cable runs for extended arrays and highly sensitive individual elements.1 In such applications, optical-fiber strain-sensing technology has generated substantial industrial and military interest. However, a number of inherent constraints have limited the performance of these systems. Besides the fundamental sensitivity of each array element, these constraints lie primarily in the lead fiber and tend to worsen with increased length.
There has been considerable effort in seeking an elegant solution to high-performance remote fiber sensing in both active and passive configurations. State-of-the-art active sensing involves single-frequency fiber lasers,3 while passive sensors typically rely on fiber interferometers wound on acoustically deformable mandrels.4 The active laser sensors are optically pumped and suffer from high attenuation at pump wavelengths in glass fibers, while the passive mandrels are bulky and optically complex. In addition, both suffer from intrinsic material Rayleigh backscatter (RBS), which introduces random noise in passive configurations. This is particularly debilitating to active sensors due to optical feedback. Further, these factors can limit the useful cable length and array size.
To overcome these issues, we have adapted techniques used in gravitational wave detection interferometry, where some of the most sensitive instruments in human history are under construction to detect minute ripples in space-time.5 The new fiber sensing methodology involves locking a radio-frequency (RF) modulated laser source, via active electronic feedback, to the resonance of a passive fiber Fabry-Pérot (FFP) interferometer fabricated in the fiber core.6,7
The passive FFP sensor is created by writing two Bragg gratings in the core of single-mode fiber with a separation of a few centimeters. This creates a resonant interferometer structure where each grating acts as a partially reflecting mirror (see Figure 1). In the presence of an acoustic signal, the optical path length (OPL) of the FFP oscillates at the acoustic frequency. This OPL change is amplified by multiple passes in the resonator, thus enhancing its sensitivity. When an RF phase-modulated diode laser is used to interrogate this resonance condition, the demodulation electronics yield a voltage signal that is proportional to the OPL displacement. This voltage is used to feedback to the diode laser to keep it in resonance with the FFP (see Figure 2).
Figure 1. The fiber Fabry-Pérot interferometer uses a UV holographic technique to write two Bragg reflectors into the fiber core. Each consists of a series of refractive index modulations. FBG: fiber Bragg grating.
Figure 2. The remote sensor system. Light from a diode laser is current-modulated prior to travelling down the delivery fiber to the remote FFP sensor. By demodulating the transmitted signal, the strain within the sensor is recovered.
This signal can also be calibrated to yield a direct measure of both static and dynamic fiber strain. If the strain signal is sinusoidal, then it would appear as a pure tone in the frequency domain, as illustrated in Figure 3. By analyzing the strain signal at acoustic frequencies, therefore, this fiber sensor can be used as an extremely acute acoustic sensor, where the dynamic fiber strain is induced by minute acoustic perturbations underwater, on the ground, or in the air.
Figure 3. (a) Dynamic strain within the fiber causes the resonance of the FFP to detune. (b) The demodulated error signal then reads out this dynamic strain signal as a function of time. (c) A Fourier transform of the error signal shows a sinusoidal strain signal as a single tone superimposed on a background of systemic noise sources.
This strain measurement technique inherits a number of key advantages from gravitational astronomy. First, it enables common-mode rejection of noise sources external to the sensing element, such as mechanical perturbations and thermal noise in the lead fiber. Second, it is immune to laser-intensity noise, and affords an enormous signal-to-noise ratio (SNR) compared with other coherent phase extraction methods. These combined advantages have dramatically improved performance.
The various noise sources are observed as an underlying noise floor in the sensor frequency spectrum. In long-range remote sensing, this is typically dominated by RBS effects. The ultimate strain sensitivity of the system is thus limited by this noise floor, as it cannot be distinguished from the actual strain signal. However, by narrowing the resonance linewidth of the FFP, and increasing the modulation depth of the laser, it is possible to further increase the SNR in this RF interrogation technique. The result is unprecedented subpicostrain sensitivity over an acoustic bandwidth up to 100kHz, in a system stretching across more than 30km of fiber.8 The enhanced SNR overcomes all RBS effects in the long length of delivery fiber, and the system is only limited by the fundamental frequency noise of the narrow-linewidth laser.
For acoustic-sensing applications such as underwater or naval hull-mounted hydrophones, and for land-based wide area surveillance, an array of fiber sensors is typically required. A sensor array can both detect the presence of signals, as well as triangulate the location of its source. Our architecture improves the sensitivity to these signals, while also increasing spatial resolution by facilitating a dramatic enlargement of the total array aperture size. Under an Australian Research Council scheme, we have commenced our development of fiber sensor arrays based on this RF interrogation technology. The project exploits optical wavelength division multiplexing in the telecommunications C, L, and S bands, with radio-frequency multiplexing for acoustic signal extraction.