Optical communication networks have contributed much to the recent dramatic change in information technologies. These changes have been led by the tremendous advances in optical fibers, lasers, integrated photonics, wavelength division multiplexing (WDM) technology, and other photonic devices and technologies. It is important to note that these same advancements provide the building blocks for another type of photonic network: a distributed and multiplexed fiber-optic sensing network.
Distributed sensors can be embedded into a variety of structures or materials to improve their reliability, safety, and security. Fiber Bragg gratings (FBGs), for example, work well as sensors for temperature and/or strain.1 These devices, which use WDM technologies, have been integrated into multiplexed sensor networks that diagnose the status of structures such as bridges, highways, and buildings.2,3 Structures that incorporate sensors can perform self-diagnosis, monitoring their own condition to warn of impending failure or the need for routine maintenance. Advantages of such "smart" structures include less down-time, less frequent maintenance, and improved safety. Similarly, smart materials, formed of composite structures incorporating embedded sensors, have been designed for use in aircraft, spacecraft, and ship bodies, for example.
Fiber Bragg grating sensors only supply data at discrete points, however. Many practical applications require fully distributed sensing in which the entire length of the fiber generates data. To achieve distributed sensing, one can use intrinsic properties of the fiber, such as scattering and mode coupling.4 The frequency shift of Brillouin scattering, for example, can be used to detect sensing strain and/or temperature. Raman scattering also can be used for temperature sensing.
Most distributed sensing systems based on stimulated Brillouin scattering use pulsed sources for the probe beam. The problem is that the properties of the pulsed approach limit the spatial resolution of the sensor to no better than several meters. This is not sufficient for smart-materials applications.
We recently developed a continuous-wave method to measure the Brillouin gain spectrum distribution along an optical fiber. Using our technique to manipulate optical coherence, we demonstrated distributed strain sensing with 1-cm spatial resolution. This resolution is 100 times higher than that of the conventional pulsed lightwave technology. optical coherence manipulation
For distributed sensing systems, mature time-of-flightbased techniques exist that can locate the point along a fiber at which a signal originates. Time domain techniques do not provide the spatial resolution or system flexibility necessary for some applications, however, leading engineers to investigate alternatives. Our group has developed a new technique for distributed sensing schemes. The method is based on manipulating the shape of the optical coherence function of the system. It uses a continuous, rather than pulsed, laser, and it includes no moving parts. As an added benefit, this approach does not require data analysis to determine the point along the fiber at which a signal originates.
In an interferometer, the phase difference between the waves propagating through the two arms generates an interference pattern. Pure sinusoidal waves yield an obvious sinusoidal interference pattern. If waves are frequency-modulated to be non-sinusoidal, then the visibility of the pattern changes as a function of the differential time delay between the two waves. The interference pattern is most visible when the lightwaves are strongly correlated. The visibility of the interference pattern as a function of the differential time delay is called the optical coherence function.
Another way of thinking about the optical coherence function is as the Fourier transform of the power spectrum of a laser. By definition, modulating the optical frequency of a laser source changes its power spectrum. If the power spectrum includes many frequency lines, for example, then the optical coherence function becomes a series of sharp peaks, like delta functions. Using frequency and/or phase modulation in appropriate waveforms, we can synthesize arbitrary shapes for the optical coherence function.5,6
Our distributed sensor can be configured as an interferometer, such as a Michelson, Mach-Zehnder, or loop types in which the sensing fiber forms one arm. By changing the waveform of the frequency modulation or by additionally modulating the phase of the beam in one arm of the interferometer, we can sweep the peak(s) of the optical coherence function over a specific distance down the sensing fiber. In our distributed fiber Brillouin strain sensing system, for example, we use a loop interferometer in which the pump and probe beams travel through the fiber loop in opposite directions from other. We apply a sinusoidal frequency modulation to the pump and probe beams to synthesize the coherence peak in the sensing region.
We also have synthesized and scanned a delta-function-like coherence peak using stepwise waveforms for the frequency and the phase modulation. In addition, we can synthesize triangle and square shapes. We have applied these synthesized coherence functions in distributed sensing systems based on a variety of techniques, such as optical reflectometry for diagnosis of optical subscriber networks7, or distributed lateral force sensing8-11, which is applicable to smart structures and security systems. Brillouin strain sensing
Stimulated Brillouin scattering can be thought of as the diffraction of light from the refractive index changes in a medium, caused by the effects of acoustic waves. To trigger SBS, we pump the mediumin this case, the fiberwith a beam, then probe it with a counter-propagating second beam. The pump wave generates a gain for the probe with a center frequency downshifted by 11 GHz from that of the pump. This frequency shift is known as the Brillouin frequency. Longitudinal strain and/or changes in temperature along the fiber change the Brillouin frequency; for example, a frequency change of 500 MHz corresponds to 1% strain. The stimulated wave undergoes amplification across a selected frequency band, known as the Brillouin gain spectrum (BGS). If we measure the BGS at different points along the fiber, we can use it to obtain the Brillouin frequency shift and thus the change in strain or temperature.
The optical spectral width of the lightwave for generating Brillouin scattering has to be narrower than that of the BGS; otherwise, the measured BGS broadens, making it difficult to determine the spectral center precisely. Since the typical value of the BGS width is 30 MHz, the limit of the spatial resolution in the pulsed-lightwave technique turns out to be 1 m. This limit prohibits its application to smart materials, which require higher spatial resolution.
Figure 1. The correlation-based continuous-wave technique creates a position-selective generation of stimulated Brillouin scattering. By varying the frequency of the probe modulator, the system creates a Brillouin gain spectrum (BGS). By varying the modulation frequency of the laser (which alters the frequency of both the pump and probe lightwaves), we shift the correlation peak along the fiber, obtaining BGS data as a function of the position.
To circumvent the resolution limit in the pulsed-lightwave technique, we have developed a new technique for measuring Brillouin gain spectrum distribution along an optical fiber.12-14 We induce stimulated Brillouin scattering (SBS) by triggering interference between the pump and probe lightwaves. In our technique, we control the coherence of the pump and probe beams in order to localize the SBS at a specific position in an optical fiber where the correlation between the two lightwaves is high (see figure 1).
Figure 2. Light from the 1.55-µm distributed feedback laser diode (DFB-LD) is divided by a coupler into the pump and probe lightwaves, which are then modulated by electro-optic modulators (EOMs). The pump is also amplified in an erbium-doped fiber amplifier (EDFA) and launched into the fiber under test. The other beam is modulated to create sidebands around the original frequency; the first lower sideband acts as the counter-propagating probe beam. Interference between the pump and probe lightwaves trigger SBS at the spot where the fiber is under strain. By discovering where the correlation between the lightwaves is high, we can determine where along the fiber the SBS is generated.
Using a coupler, we divide the light of a 1.55-µm frequency-tunable distributed-feedback (DFB) diode laser into pump and probe lightwaves. The pump beam is chopped by a radio-frequency-driven electro-optic modulator (EOM) that is amplified by an erbium-doped fiber amplifier (EDFA) and launched into the fiber under test (see figure 2). Another EOM modulates the probe beam with a microwave frequency v, so that sidebands are generated around the original frequency vo. The first lower sideband at vo v, serving as the probe, propagates against the pump in the fiber under test, and reaches the detector. An optical filter eliminates the other sidebands.
An important feature of our system is that the pump and probe beams undergo identical frequency-modulation at the laser. As a result, SBS occurs exclusively at the correlation peak position, where the two lightwaves are highly correlated. At other positions, the SBS is prohibited because of the relative spectral broadening between the pump and probe. The probe power increment caused by the Brillouin gain is obtained through lock-in detection. We obtain the BGS by varying the frequency v of the modulator that generates the probe. We can shift the correlation peak along the fiber by changing the modulation frequency m at the laser. By repeating the BGS measurement while changing the frequency m, we obtain the BGS as a function of the position.
Figure 3. The data plot of Brillouin frequency shift as a function of position demonstrates that a continuous-wave SBS technique achieved 3-cm spatial resolution.
We applied strain to a short section of the fiber, and measured the strain distribution with a resolution on the order of centimeters (see figure 3).15 Such high resolution cannot be obtained by conventional pulsed-lightwave techniques. We believe that our CW technique will allow even higher spatial resolution and that this technique is suitable for the smart material applications. Toward this end, we demonstrated a system that successfully measured strain distribution along the surface of a pipe 15 cm in diameter, with a spatial resolution of 1 cm.16 This is also the first demonstration of smart materials with such a high spatial resolution.
Systems that use correlation-based continuous-wave techniques or synthesis of optical coherence function techniques provide high resolution for distributed sensing. Our fiber-optic Brillouin distributed strain-sensing system demonstrated a 1-cm spatial resolution, which is 100 times higher than that of the conventional time domain techniques. We continue to research methods for creating fiber sensing networks that can improve the reliability, safety, and security of structures and devices for the good of society. oe
1. J. Dakin and B. Culshaw edt., Optical Fiber Sensors, p. 369, (1997).
2. Proceedings of 14th Intern. Conf. on Optical Fiber Sensors, Venice (2000).
3. Proc. SPIE #4204 (2000).
4. J. Dakin and B. Culshaw, Optical Fiber Sensors, p. 309 (1997).
5. K. Hotate, Optical Fiber Technology, 3, pp. 356-402, (1997).
6. K. Hotate, T. Saida, and Z -Y. He, Proc. SPIE #3407, pp. 366-373, (1998).
7. T. Saida and K. Hotate, IEEE Photonics Technol. Lett., 10, pp. 573-575, (1998).
8. T. Saida and K. Hotate, IEEE Photonics Technol. Lett., 9, pp. 484-486, (1997).
9. K. Hotate, X. Song, and Z-Y. He, IEEE Photonics Technol. Lett., 13, pp. 233-235, (2001).
10. Z-Y. He and K. Hotate, SPIE Fiber Optic Sensor Technology II, Boston, pp. 4204-19, (2000).
11. Z-Y. He and K. Hotate, Optics Lett., 24, pp. 1502-1504, (1999).
12. K. Hotate and T. Hasegawa, IEICE Trans. on Electron., E83-C, pp. 405-412, (2000).
13. K. Hotate and T. Hasegawa, 14th Intern. Conf. on Optical Fiber Sensors, Venice, pp. 651-661, (2000).
14. K. Hotate and M. Tanaka, IOOC/OECC, Sydney, pp. 271-274, (2001).
15. K. Hotate and M. Tanaka, 14th Intern. Conf. on Optical Fiber Sensors, Venice, pp. 647-650, (2000).
16. K. Hotate and M. Tanaka, CLEO/Pacific Rim Vol.I, Makuhari, pp. 496-497, (2001).
profile: sensitive solutions from optical fibers
From the time he received his doctorate of engineering in electronics engineering from the University of Tokyo (Tokyo, Japan) in 1979, Kazuo Hotate involved himself with optical fibers. Upon graduation, he joined the university's faculty as a lecturer and pursued his interest in finding applications for optical fibers in space.
For his work on measurement and analysis of optical fiber characteristics that initial year, Hotate received an Achievement Award. Over his 23-year career, Hotate has written and coauthored several books on optical fibers and optical-fiber sensors as well as more than 180 journal papers and international conference presentations.
"I have been especially interested in the application of optical-fiber sensing technology in space," says Hotate. Today, his laboratory team concentrates on spatial resolution in distributed fiber-optic sensors. "Our sensors can detect changes every centimeter along our optical fibers," explains Hotate. "In a highway tunnel, for example, they could tell exactly where a fire was, and from which direction the wind was blowing, just by the temperature variations along the fiber." Hotate says many corporations have contacted him concerning the technology, and he predicts that it will soon be at work watching over aircraft wings, rocket fuel tanks, and other structures.
Hotate will be general chair of the 16th International Conference on Optical-Fiber Sensors in 2003.
Integrated CCD eliminates sensor tradeoffs
By Helen Titus, Eastman Kodak Co.
Today's vision-system designer has a choice between two image-sensor alternatives: complementary-metal-oxide-semiconductor (CMOS) sensors and charge-coupled-device (CCD) detectors. Each presents its own tradeoffs. While the CMOS sensor offers a high degree of functional integration, its imaging performance is limited compared with that of a CCD. And while the CCD can offer the ultimate in output quality, it has historically been a single-function device, requiring many external support components.
Highly integrated CCDs now offer the best of both worlds, delivering strong imaging performance while incorporating many functions that previously required the digital camera designer to integrate discrete components. These integrated sensors simplify and increase speed in digital camera design while simultaneously reducing device size, power consumption, and cost. For example, it is now possible to buy a megapixel CCD that features 10 bits of dynamic range at 40 MHz, 7.4-µm square photodiodes with integral microlenses, throughput rates up to 48 frames/sec, and the ability to operate in progressive scan or interlaced mode with single or dual outputs. voltage control
One of the most complex aspects of digital camera design is the need to control multiple clock signals at various voltages to drive the camera's electronic shutter and transfer the image through and out of the CCD. The integration of an electronic shutter driver, vertical clock drivers, a fast dump driver and other functions on the same die as the CCD greatly simplifies this task.
Integrated clock drivers generate the proper voltages for a CCD's internal gates and control the shifting of electrical charge from the sensor's photodiodes to and through the vertical shift register (VCCD). By integrating voltage translation circuitry on chip, a CCD image sensor can limit its voltage requirements to a single 5-V supply, thus simplifying the external circuitry needed.
For example, in the KAI-1020 CCD image sensor (Eastman Kodak Co.; Rochester, NY), an integrated electronic shutter driver triggered by a 4-µs, 5-V pulse on the SH input provides a method for precisely controlling image exposure time without additional mechanical components (see figure). The on-chip electronic shutter driver takes care of translating the 5-V pulse to the 30-V substrate voltage needed for this function. The electronic shutter eliminates the need to wait until previously acquired images are completely read out of the VCCD. Integrated clock drivers are also used for the two phases of vertical clocks that first shift the charge from the photodiode into the vertical CCD and then through the VCCD during readout.
An integrated fast dump driver, in turn, allows an entire line of an image to be quickly drained without clocking it through the horizontal shift register (HCCD). The fast dump feature is useful for subsampling an image to achieve higher frame rates. For example, in high-speed applications that can tolerate the loss of some image resolution, the designer may choose a subsampling factor of two, which means dumping the even numbered lines and reading out only the odd numbered lines. This scenario almost doubles the frame rate.
The fast dump driver may also be used to boost frame rate through a process known as subwindowing, which allows the system to focus on only a particular region of interest in an image while discarding the rest. If, for example, only the center 512 lines of an image is of interest, the designer could activate the fast dump driver and clock the VCCD for 256 lines, turn off the driver and clock VCCD and HCCD for 512 lines, and finally reactivate the driver and clock the VCCD for 240 lines.
Other valuable on-chip functions include a correlated-double-sampling (CDS) unit, which greatly simplifies the analog signal processing performed in a digital camera. Designing high-speed CDS circuitry can be especially difficult, so the integration of this feature within an image sensor is particularly useful in digital-camera design.
Kazuo Hotate, Masato Tanaka
Kazuo Hotate is professor and Masato Tanaka is a graduate student at the University of Tokyo.