The construction scale of large-scale infrastructure in China has ranked first in the world for many years. Meanwhile, due to construction quality, using environment, natural disasters, and other factors, serious accidents occur frequently. Distributed optical fiber sensing technologies employ optical fibers as signal transmission medium and sensing units to realize continuous distributed measurement of external parameters along the optical fiber. Therefore, it is the most potential non-destructive monitoring technology for large-scale infrastructure health monitoring in real time. However, distributed fiber optic sensing technologies still face various challenges such as reliability, low cost, and intelligence as they move toward the market.

At present, distributed optical fiber sensing technologies that have caught extensive attention and research include optical time-domain reflectometer, coherent optical time-domain reflectometer, phase-sensitive optical time-domain reflectometer, optical frequency-domain reflectometer, Raman optical time-domain reflectometer, Brillouin scattering optical time-domain reflectometer, Brillouin optical time-domain analyzer, and optical interferometry. We focus on introducing their working principles, system basic structures, development history, current status, and major research institutions and manufacturers at home and abroad.

Based on detailing the application requirements, principles, and methods of distributed optical fiber sensing technologies in communication system monitoring, power system monitoring, coal geology monitoring, oil and gas exploration, transportation field, transportation pipeline monitoring, aerospace equipment monitoring, and perimeter security, we provide several typical application cases.

The future main directions of development are listed:

1) Multi-mechanism integration system. Single sensing parameters make it difficult to represent the true state of the measured object, which can result in false reports and missed reports. Simultaneous measurement of multiple parameters can provide multidimensional and more comprehensive information, thereby more accurately identifying fault events. The key point of the fusion-type distributed optical fiber sensing technology is to employ different scattering lights to respond to different events in the optical fiber to achieve multi-parameter sensing.

2) Specialty sensing fiber cable technology. By changing the fiber material, structure, and packaging, specialty optical fiber cables can overcome the limitations of distributed sensors based on ordinary single-mode optical fibers, and obtain engineering applications in specific sensing parameters and performance in specific fields and scenarios.

3) Sensing signal processing and intelligent perception technology. Due to the weak intensity of scattered light compared to incident light, distributed sensing systems are limited by signal-to-noise ratio. This affects the measurement accuracy, monitoring distance, response speed, spatial resolution, and other key indicators of distributed sensing systems. Signal processing techniques to analyze and enhance collected data are important means to improve the performance of sensing systems.

4) Communication-sensing fusion system. Technologies such as wavelength division multiplexing, polarization diversity, and coherent detection from optical communication systems are applied to distributed fiber optic sensing systems. Additionally, existing optical fiber communication systems can be adopted for synchronous sensing. These are crucial steps towards the practical applications of distributed fiber optic sensing systems.

5) Distributed shape sensing technology. Leveraging distributed fiber optic sensing technology for shape sensing is an important development direction.

6) Ocean state monitoring based on existing optical cables. Existing undersea optical communication networks are employed as sensing networks to achieve intelligent perception of the surrounding environment of the cables. This enables large-scale online monitoring and early warning capabilities with relatively low investment, thus providing rapid and accurate assurance for managing major maritime incidents and maritime disaster risks.

Distributed fiber sensing and measurement techniques have been given attractive attention in recent decades due to high sensitivity, high resolution, and large capacity. They have found a wide range of applications in the structural health monitoring of civil infrastructures such as bridges and dams, power-transmission line monitoring, oil-gas extraction and pipeline leakage detection, marine geophysical exploration, dynamic measurement, fiber-optic device characterization, fault diagnosis, etc. On the one hand, distributed measurement techniques can be categorized in principle into scattering effects (including Rayleigh backscattering, Brillouin scattering, and Raman scattering) and coupling effects (polarization crosstalk). On the other hand, these techniques can be divided into optical time domain reflectometry (OTDR), optical frequency domain reflectometry (OFDR), and optical coherence domain reflectometry (OCDR).

OTDR employs the short and high power light pulse for interrogation, which is an effective tool for long distances. However, the tradeoff between sensing length and spatial resolution restricts the measurements to only meter-level spatial resolutions. OCDR utilizes the low coherence light from a broadband light source. They can offer a micrometer-level spatial resolution, whereas the measurement range is less than a few meters. OFDR is a distributed optical fiber measurement method based on the frequency-modulated continuous wave principle in the optical domain. It obtains the characteristics, such as scattering/reflection/loss and polarization features, along the optical fiber according to the mapping relationship between the Fourier transformation frequency of the interference signal and the characteristic location. In addition, the distribution of external physical fields, such as temperature/stress/strain sensing, can be further acquired. Unlike distributed measurement methods based on time-domain or coherent-domain, OFDR offers superior comprehensive properties, including high spatial resolution, high measurement sensitivity, long measurement distance, broad dynamic range, and high-speed response. However, due to the influence of phase noise, amplitude noise, and environment noise, the performance of OFDR in practice is not satisfactory.

In the past few years, various methods have been proposed to compensate for the laser source noise and environment noise to improve the performance of the OFDR. Distributed sensing based on OFDR is also developing towards high performance and multi-parameters. With the continuous expansion and deepening of the application field, OFDR is facing more daunting challenges, which put forward higher requirements for its measurement performance and anti-interference ability. Therefore, it is of great importance and necessary to provide an overview of recent research progress in existing high-performance OFDR tests and sensing techniques to guide the future development direction.

We first review the measurement principle of OFDR and summarize key technologies to enhance OFDR system performance, such as the noise sources in distributed measurement (Fig. 1), the degradation mechanisms of the spatial point spread function (Fig. 3), and the error or noise compensation techniques. Then, the measurement limit of distributed sensing based on OFDR is derived, and several methods for improving the sensing accuracy and measurement distance are analyzed (Fig. 14). Subsequently, an outline of the current development status of domestic and foreign OFDR instruments is given (Table 6). Besides, application examples are given in measuring integrated waveguide devices, polarization maintaining fibers, and inside stress sensing of optical fiber coil. Finally, several future research directions of OFDR are prospected.

OFDR systems can provide a good performance of high spatial resolution, high speed, and long measurement and sensing length. This technique can be widely applied to the fields of high-performance fiber optic component measurement and high-precision multi-parameter sensing. In the future, OFDR will continue to develop toward the goal of higher performance, stronger environmental adaptability, and higher measurement cost-effectiveness. The mixed modulation technology such as multi-domain localization (including time, frequency, and coherent domain) and multi-dimensional modulation (including amplitude, phase, and polarization modulation) can provide an effective way to break through the measurement limits and realize the ultra-high performance of OFDR technology. Furthermore, the high-precision OFDR sensing technology should be stepped up to meet the demands of multi-parameter decoupling and anti-interference ability improvement. Correspondingly, for the noise compensation algorithms at present, artificial intelligence and advanced algorithms are all important means for noise suppression capability enhancement and demodulation accuracy improvement. Besides, new requirements are put forward for the small size, low power consumption, and low cost of the core modules in OFDR instruments.

With the continuous innovation of OFDR technology theory and the progress of technology development, China's current overall technology level has achieved international parallelism. However, the typical application fields of OFDR technology need to be continuously expanded, and the advantages of the technology need to be continuously emphasized. In this context, the development of domestic OFDR technology should be highly valued and vigorously developed to realize OFDR technology independent control and localization of hardware, including continuous mode-hopping-free tunable laser source, high-speed and high-precision optoelectronic conversion, and data acquisition module. Moreover, the OFDR technology should gradually move towards engineering applications in the field rather than being confined to laboratory measurements. The environmental adaptability of OFDR instruments should be enhanced to ensure that the core technical indicators of distributed testing and sensing are not degraded in different scenarios. Finally, a high-performance distributed specialized measurement and quantitative sensing methodology should be proposed to promote application development in core fields and typical scenarios, which provides a solid foundation and strong support for satisfying the requirements of applications such as testing of military devices, exploration of oil and gas resources, and power and energy monitoring.

Over the past decades, the national demand for structural health monitoring of large infrastructures such as bridges and oil and gas pipelines has gradually increased. Based on the scattering effect in optical fibers, researchers have proposed a distributed optical fiber sensing (DOFS) system. It is not only very sensitive to external parameters such as temperature change, strain, and vibration, but also has the advantages of long-distance multipoint monitoring, low cost, corrosion resistance, radiation resistance, and large bandwidth, which makes it an important technological tool for structural health monitoring of large-scale infrastructures. DOFS is mainly categorized based on scattering mechanisms, which are Rayleigh scattering, Brillouin scattering, and Raman scattering. Compared with other DOFS, DOFS based on Brillouin scattering has high temperature and strain sensitivity, thus providing accurate measurements. In addition, it is also capable of long-distance distributed monitoring of external strains and temperature changes with high spatial resolution, which has attracted the attention of a large number of researchers and has been widely used.

However, with the increase in sensing distance in DOFS, the decrease in signal-to-noise ratio (SNR) will lead to an increase in measurement uncertainty. In addition, massive data will be generated in the process of long-distance continuous measurement, and the required measurement time will increase correspondingly. How to process massive data intelligently, quickly, and accurately to further improve system performance and obtain more accurate physical parameters is the biggest problem facing the development of the system. Currently, the development of system hardware technology is particularly insufficient in the face of massive data processing, which creates an opportunity for advanced signal processing and analysis using digital signal processing (DSP) technology, which can effectively obtain effective information from the massive data generated by the system. In the past few years, the development of powerful computer processors has laid the foundation for the development of advanced DSP technologies, and recent advances in big data and cloud technologies have provided tools for efficient storage and massive data processing. With the development, DSP technology has the advantages of smaller back-end processing time overhead and no increase in system hardware complexity.

We review the DSP techniques used for data processing in Brillouin-DOFS in recent years and focus on the applications of image and video denoising technology and machine learning information extraction and recognition technology in it, so as to provide a reference for future research of DSP technology in Brillouin-DOFS.

The multi-dimensional (time, frequency, and position) domain of Brillouin signals contains redundancy and structural similarity. However, none of the earlier denoising methods have utilized the feature. Thus, the researchers have introduced the image-video denoising technique to reduce the noise of the sensing signals. At first, some traditional image and video denoising algorithms are summarized, and the principles of the algorithms, as well as the performance of denoising effects are generally introduced. It also shows that the optimization of algorithm parameters and the transformation of 3D BGS can enhance the denoising performance. However, the traditional algorithms still affect the spatial resolution and measurement reliability. With the research and development of machine learning, neural networks have also been used for denoising Brillouin signals by the powerful nonlinear fitting ability. Neural networks have many architectures such as artificial neural networks (ANNs), convolutional neural networks (CNNs), and generative adversarial networks (GANs), and all of them are capable of fast and high-fidelity denoising.

Machine learning has a strong ability to fit complex nonlinear functions, which is very suitable for solving regression and classification problems. In addition, the machine learning algorithm is extremely short in time, showing its potential in information extraction. First, the application of traditional machine learning algorithms to the direct extraction of temperature or frequency is presented. These algorithms demonstrate much higher extraction speed than traditional fitting algorithms and have stronger robustness. With the increase in computing power, the neural network can be well-trained by simulating a large number of Brillouin gain spectra in different situations. In addition, by constructing the dataset in different cases, the corresponding purpose can be realized, such as solving the frequency extraction error caused by non-local effects. Finally, some studies on the performance evaluation of neural network model extraction and the integration of neural networks with other techniques are also presented.

DSP technology can process massive data intelligently, quickly, and accurately, so as to further improve the performance of the system. Firstly, the concept of image and video denoising makes use of the repeated structures of information in the multi-dimensional domain of Brillouin signal. Then, a variety of traditional denoising algorithms and machine learning methods have been applied. Secondly, since traditional fitting methods are time-consuming, machine learning techniques are also introduced into Brillouin-DOFS. It can directly learn the nonlinear mapping between input and output so that information such as frequency, temperature, or strain can be accurately and quickly extracted from BGS. In the future, in addition to developing more advanced techniques to achieve longer, more accurate, and faster sensing systems, how to better evaluate and interpret machine learning algorithms is also the focus of research. It is believed that Brillouin-DOFS based on DSP technology will play an increasingly important role in infrastructure, aerospace, energy transportation, and other fields.

Acoustic detection is a basic way for human beings to perceive the environment. Hydrophone technologies are key means of underwater acoustic detection and play an important role in target detection, communication, navigation, resource exploration, and marine ecological monitoring. At present, the mainstream hydrophone technologies are mainly divided into two categories of piezoelectric hydrophone and fiber optic hydrophone (FOH). The former has been widely applied, and FOH rapidly developing in recent decades features high detection sensitivity, unpowered wet-end, and convenient networking. However, these conventional hydrophones have many disadvantages. First, they are in nodal type, the multiplexing scale and array size are limited, and the largest array number is far less than 1000. Second, their array parameters (array spacing, array aperture, etc.) cannot be changed after being determined, and the target type to be located is limited, which cannot meet the detection needs of various targets. Finally, the wet-end part needs to be prepared by hand due to the complex fiber connect relationship. Therefore, the existing hydrophone technologies are difficult to meet the strict requirements of advanced marine science and future underwater acoustic detection, such as large-scale detection arrays, rapid and flexible deployment, adaptive array reconstruction, and low-cost large-scale monitoring. Meanwhile, it is extremely important to develop new hydrophone technologies.

Distributed fiber optic hydrophone (DFOH) technology is a new underwater acoustic detection technology developed in recent years. In DFOH, the optical fiber is converted into thousands of acoustic transducers, and all acoustic information can be obtained along the fiber quantitatively and spatial-continuously from the backscattering of the inquiry laser pulse. DFOH has unique advantages including densely spatial sensing, flat frequency response, flexible array reconstruction in the digital domain, and large array (tens of kilometers). Additionally, in terms of engineering applications, the wet-end of DFOH can be mechanically produced with high efficiency and good consistency, which is essential on large-scale array construction and rapid mass production. In 2019, the Naval Research Laboratory in the United State publicly stated that research was being conducted on a new generation of hydrophone technology based on Rayleigh scattering, and afterward, DFOH technology attracted widespread attention and was rapidly developed.

In DFOH, the sensing fiber is converted into acoustic transducers by utilizing the laser phase response to the external sound field, and the external sound field is continuously detected in the spatio-temporal domain, with each channel separated in the way of optical time domain reflectometer (OTDR) or optical frequency domain reflectometer (OFDR). Thus, the fundamental principle is divided into laser phase response and signal demodulation. On the former, the DFOH response mechanism is consistent with that of conventional interferometric FOH, and fiber secondary coating and wed-end structured design (Fig. 1) are also effective in improving the DFOH response (sound pressure sensitivity). On signal demodulation, DFOH is quite different from FOH and channel separating is essential, with complex backscattering mixing along the fiber. The principle details are introduced by us.

The DFOH performance has been rapidly enhanced in recent decades. The preliminary foundation of DFOH is built from phase-sensitive OTDR. The first qualitative demodulation was proposed by Taylor in 1993, and the first quantitative demodulation (Fig. 2) was realized by the Shanghai Institute of Optics and Fine Mechanics (SIOM), Chinese Academy of Sciences in 2011. Soon afterward, many demodulation methods are proposed. The DFOH concept was first proposed in 2015, when the Shandong Academy of Sciences verified the feasibility of DFOH to detect underwater sound in the laboratory, with sound pressure sensitivity of -158 dB and phase noise of -56 dB. With the joint efforts of domestic and foreign scholars, the DFOH performance indexes are greatly improved, including sound pressure sensitivity (Table 1), system noise level (Table 2), system equivalent noise, dynamic range, and response directivity (Fig. 3). Meanwhile, the effective detection range (Table 3) of DFOH passive sonar system is theoretically evaluated, and the evaluation system of DFOH performance is gradually improved.

In recent years, the dry-end technology and wet-end cables keep optimizing, laboratory tests constantly improve, and the applications are explored in underwater suspended horizontal array, lightweight towed array, and hydrophone array with submarine communication cables. On the underwater suspended horizontal array, direction and localization of underwater target and lake tests are focused, and the representative groups are from SIOM (Figs. 4 and 5), Shanghai University, and Naval University of Engineering. The lightweight towing array application is still in the exploratory stage, the flow noise and channel crosstalk are studied, and Naval University of Engineering and Zhejiang Laboratory (Fig. 6) are the most representative groups. In terms of hydrophone array with submarine communication cables, the joint team of the Norwegian University of Science and Technology and Cornell University is the biggest concern, and they detect and track whales in the Arctic with existing submarine cables (Fig. 7), which is expected to provide a new means of all-weather monitoring for target detection in vast sea areas.

As a novel hydrophone technology, DFOH has unique advantages of continuous spatial detection, flexible array reconstruction, automatable wet-end production, light weight, and low cost. In recent years, DFOH has developed rapidly and has been verified in many application scenarios. We introduce the basic sensing principle and typical demodulation methods of DFOH and review the important performance indexes and research progress, including sound pressure sensitivity, system equivalent noise, response directivity, and dynamic range. Some representative applications are also introduced, such as underwater suspended horizontal array, lightweight towed array, and hydrophone array with submarine communication cables. Additionally, the existing problems and possible development trends are discussed. We believe that DFOH will play an important role in underwater target detection, marine communication and navigation, and environmental ecological monitoring.

Force/pressure measurement has always been a focus of attention in many industrial and environmental structures, medical fields, and defense architectures. It is particularly in high demand in areas such as oil and gas wells and pipelines, geotechnical engineering, water distribution, and wastewater treatment facilities. Traditional electronic sensors are not suitable for remote monitoring, and they are sensitive to electromagnetic interference and not easily multiplexed in large-scale sensor networks. The single-point fiber optic sensor has been successfully commercialized, but in many important application areas, even dense multiplexed quasi-distributed fiber optic sensing systems cannot meet measurement requirements. Therefore, there is a strong demand for research on distributed transverse force (TF)/pressure fiber sensing. However, compared with distributed fiber sensing techniques that can measure parameters such as strain, temperature, and vibration, the basic technology for distributed TF sensing is lacking. Some indirect measurement methods using special mechanical structures to convert TF into other parameters face significant issues such as high complexity, low accuracy, and difficulty in practical application. The development of a direct distributed TF fiber sensing technology is highly desired. Previous researchers have proposed measurement techniques based on specialty fibers and single-mode fibers (SMFs) for distributed polarization properties, providing a new idea for distributed TF fiber sensing. However, due to technical limitations or performance deficiencies in the measurement systems, there have been few studies on distributed TF fiber sensing based on polarization analysis.

Based on a thorough analysis and study of previous research on distributed TF fiber sensing and potential key technologies, the authors have taken the lead in conducting research on distributed transverse pressure fiber sensing based on polarization analysis. The breakthroughs have been made in polarization-maintaining fibers (PMFs) and SMFs-based distributed TF measurement and demodulation systems, sensing medium, system performance, and typical applications. We have constructed a constructed polarization crosstalk analysis (DPXA) system without "ghost peaks", which effectively eliminates the influence of second-order crosstalk peaks on measurement accuracy (Fig. 3 and Fig. 4). We have also studied the polarization crosstalk response characteristics of PMFs and the influence of fiber coatings, demonstrating that polyimide-coated PMF is more favorable for distributed TF fiber sensing (Fig. 6 and Fig. 7). Furthermore, we have developed the equipment for PMF's axis alignment and sensing tape fabrication (Fig. 8), which enables automated 45° birefringence axis alignment for fiber sensing tape production (Fig. 9). We have verified the feasibility of distributed TF fiber sensing using the PMF and achieved high measurement resolution and repeatability (Fig. 12). We explored the feasibility of using twisted PMF and high-birefringence spun fiber (SF) as TF sensing media without dependence on the force-applying angle. By utilizing the SF, force-applying-angle-insensitive distributed polarization crosstalk measurement within 2 dB was achieved (Fig. 16). We have invented and built a high-performance distributed polarization analysis (DPA) system with full Mueller matrix measurement capability (Fig. 18). This system enables distributed birefringence measurement in a SMF with high spatial and measurement resolution (Fig. 19 and Fig. 20). We were the first to achieve direct distributed TF sensing in a SMF (Fig. 21) and obtained excellent sensing performance (Table 1, Fig. 24, and Fig. 25). We validated the feasibility of TF measurement-based monitoring deformations in SMF-embedded composite materials (Fig. 28) and determined the groove-angles for zero clamping-induced birefringence when fixing the SMF in two types of V-grooves (Fig. 30).

The above research findings provide a foundation for the measurement and demodulation techniques of distributed TF fiber sensing and provide a good overall technical reserve for its continuous promotion to practical applications. In the future, the development will mainly focus on miniaturization and integration of sensing demodulation systems, low-cost mass production of PMF and SF, optimization of fiber coating processes, or development of new coating materials to enhance the sensitivity of TF sensing, practical online applications of distributed TF fiber sensing, and exploration of new directions. Additionally, the DPA technology has also shown a potential advantage in monitoring the deformation of SMF-embedded composite materials and characterizing and optimizing birefringence properties within optical fiber devices and optical equipment. It is also a key direction for future development.

As an indispensable part of people's life, water resources can ensure stable economic development. Although the total amount of water resources is enormous in China, their spatiotemporal distribution is very uneven. For the sake of resolving the dilemma of uneven distribution of water resources in time and space, a series of large-scale water diversion projects are being built in China. Prestressed concrete cylinder pipe (PCCP) has been widely used in long-distance water delivery projects due to its advantages of high strength, good seismic resistance, and anti-leakage. However, with the increase in service life, there could be an occurrence of prestressed steel wire broken in PCCP pipelines, which may cause pipe burst accidents when broken wires accumulate to a certain extent. Therefore, it is urgent to carry out real-time online PCCP broken wire monitoring. Compared with traditional monitoring methods, the PCCP broken wire monitoring technology based on distributed optical fiber acoustic sensing (DAS) has obvious advantages such as continuously distributed sensing, long monitoring distance, and anti-electromagnetic interference. Meanwhile, the sensing optical cable can be put into the pipelines with water inside, which means that a broken wire monitoring system based on DAS can be deployed even if the PCCP pipeline is in operation. Therefore, DAS has broad application prospects in PCCP broken wire monitoring. As the broken wire monitoring system based on converged DAS gradually develops in recent years, the operation safety of PCCP pipelines will be better guaranteed.

Traditional nondestructive PCCP broken wire detection methods include manual echo detection, electromagnetic detection, and acoustic monitoring. Manual echo detection utilizes the echoes from knocking on the pipes to determine the internal defect position, so its maintenance efficiency is low, and the accuracy is uncertain depending on personal work experience. Electromagnetic detection is based on the remote field eddy current (RFEC), which was proposed by MacLean in 1951. The magnetic field signal will distort when propagating through broken wires so that the location of broken wires can be acquired. However, electromagnetic detection is still inefficient and only practical during the pipeline maintenance period. Subsequently, acoustic monitoring which is able to capture acoustic signals of broken wires through electrical or optical sensors has attracted attention. In the 1990s, Mark Holley, Robert Diaz, and Michael Giovanniello first proposed and commercialized acoustic monitoring in a section of a pipeline in Maryland. Hydrophone arrays manufactured with sensors were adopted in the program to provide comprehensive information of the pipelines so that pipe burst accidents could be predicted and handled in time. Although acoustic monitoring resolves the problems of inefficiency and non-real-time detection, traditional acoustic monitoring methods are still susceptible to ambient noises because there is a certain distance between two adjacent sensors so that the acoustic signals of broken wires will be distorted before being acquired by the sensors. In order to reduce the signal distortion brought by ambient noises, researchers turn their attention to DAS. Michael S. Higgins and Peter Paulson from Pure Technologies first utilized phase‑sensitive optical time domain reflectometry (Ф-OTDR) for PCCP broken wire monitoring and the monitoring device detected 11 broken wire incidents during four months. The positioning uncertainty of the device is about ±1.5 m. However, the frequency response of Ф-OTDR is restricted by monitoring distance so that it is hard to distinguish the broken wire signals with high frequency from other ambient noises along a long monitoring distance. Therefore, a fiber optic interferometer (FOI) can come in handy for the monitoring of high-frequency signals. FOI can detect broadband vibration signals by measuring the phase difference between the sensing light and the reference light while it is hard for FOI to locate the accurate position of the vibration. Based on the advantages and disadvantages of Ф-OTDR and FOI, there have been some converged DAS systems of these two technologies to acquire the accurate position and broadband frequency characteristics of broken wires at the same time. In 2023, Beijing Aqua Intelligent Technology Co., Ltd, Nanjing University, and Nanjing Fiber Photonics Technology Co., Ltd jointly developed a broken wire monitoring system converged by Ф-OTDR and FOI, which achieved single-ended monitoring along a section of a 20 km-long pipeline. The positioning uncertainty of the converged system is ±2 m, and the top frequency response can reach 20 kHz. Compared with the broken wire monitoring devices of foreign companies, there may still exist some gaps in the performance indicators and monitoring accuracy of domestic devices. While the industrialization of the correlation technique in China is at an early stage, relevant construction experience and characteristic data of broken wires still need to be accumulated. With the increasing construction of large-scale water diversion projects in China, broken wire monitoring based on DAS will get further development and innovation.

Broken wire monitoring is an indispensable part of PCCP pipeline health monitoring, which can guarantee people's normal lives. DAS can take advantage of the whole section of optical fiber to locate any faint disturbance along a fiber of tens of kilometers accurately. In recent years, converged DAS system has been used in PCCP broken wire monitoring to give early warning of PCCP pipelines that are at risk of bursting. However, the relevant technology has not been applied in actual projects on a large scale in China. In the future, it is essential to carry out further engineering applications, gather characteristic data of broken wire events, and improve signal processing methods.

Internet of Things (IoT) technology is an important part of the new generation of information technology. The IoT is a huge network formed by combining various information devices and sensors. This huge network is based on various types of sensors, and the sensor, as a bridge between the physical world and the digital world, is an integral part of the IoT architecture. Distributed fiber-optic sensing (DFOS) technology is widely used in many fields because of its long-distance, large-range, high-precision, and multi-point measurement capabilities.

However, most of the DFOS sensing systems use a single scattering mechanism. The parameters measured by a single scattering mechanism are limited, so it fails to fully and accurately reflect the real state of the measured object, and it is difficult to fully and effectively meet the needs of health monitoring or automatic control. In order to solve the problem that the conventional DFOS system only uses a single scattering mechanism, researchers have proposed a series of DFOS sensing systems with multi-mechanism in recent years, using the same system to measure multiple parameters. In this paper, the multi-mechanism DFOS technology developed in recent years is reviewed, and the different multi-mechanism DFOS systems and their performances are classified from the perspective of different scattering mechanisms (Table 1).

There are three kinds of scattering in optical fibers, namely Rayleigh scattering, Brillouin scattering, and Raman scattering. DFOS systems that use only one of these scattering mechanisms measure limited parameters. The multi-mechanism DFOS systems can measure more parameters by using multiple scattering mechanisms in one system, so as to reflect the state of the measured object more comprehensively.

The multi-mechanism DFOS systems are divided into five categories according to the sensing mechanism used. The system combining Rayleigh scattering with Brillouin scattering can be used not only to measure temperature, strain, and vibration but also to separate the response of temperature and strain. Systems that combine Rayleigh scattering with Raman scattering can be used for sensing temperature and vibration events. Systems that combine Brillouin scattering with Raman scattering are generally used to separate system responses due to temperature and strain.

The methods of combining scattering mechanisms in the systems are different. In this paper, these combination methods are divided into two categories: the combination based on multiplexing (wavelength division multiplexing, space division multiplexing, and time division multiplexing) and the combination of different scattered light generated by the same probe light. Multiplexing-based combination methods are straightforward in principle, but complex systems often require special sensing fibers or sacrifice measurement speed. The system using different scattered light of the same probe light has a simple setup, but special modulation and demodulation schemes are required. In addition, there may be an influence between different scattering mechanisms when the different scattering of the same probe light is used in the same system.

In addition to the combination of different kinds of scattered light in optical fibers, we also enumerate distributed sensing systems using scattered light and single point optical fiber sensing systems using interference structures or gratings. Compared with distributed sensing systems, single-point optical fiber sensing systems have the advantages of high precision and large measurement range, but the number of measured points is limited, and special optical fiber structures (such as fiber grating) are required. In practical applications, distributed fiber optic sensors or single-point fiber optic sensors can be flexibly selected according to different scenario requirements.

Finally, the prospect of multi-mechanism DFOS technology is provided. With the increasing demand for large-scale sensing and monitoring, researchers have proposed more and more multi-mechanism DFOS systems to measure more parameters and improve sensing performance. The future development of multi-mechanism DFOS systems should focus on the aspects of system complexity, sensing performance, data processing, and practical applications.

One of the natural advantages of multi-mechanism DFOS systems is the multiplexing of components in a sensor system with different mechanisms, which can significantly reduce the cost of the system when multiple sensing functions are implemented. The data processing method expands the application scenarios, increases the functions of the system, and improves the performance of the system without increasing the complexity and cost of the system hardware. Finally, how to deeply integrate the multi-mechanism DFOS systems with practical applications is an important direction. In order to achieve this direction, the design of the sensor system, the layout of the sensor cable, and the use of multi-sensor information should be considered and designed.

Information technology is the cornerstone that supports the development and social life of today's world, with its important component of sensing technology. Fiber optic sensing technology utilizes light waves as information carriers and transmission media to achieve the collection and measurement of signals in the environment. As an important branch of fiber optic sensing technology, distributed fiber optic sensing can achieve long-distance, high-resolution, and highly sensitive continuous distributed detection, obtaining two-dimensional spatio-temporal distribution information. Compared to the other two types of scattering distributed sensing, the system based on Rayleigh scattering features higher backscattering power and faster response and is more suitable for detecting dynamic and static signals such as sound waves and strain. With the increasing demands for engineering applications such as resource exploration, structural health monitoring, and underwater exploration, distributed fiber optic sensing has developed rapidly in recent years.

At present, most distributed sensing systems usually employ single-mode fiber (SMF) as the sensing medium. However, its Rayleigh backscattering signals are extremely weak, resulting in poor signal-to-noise ratio (SNR) of sensing light, which in turn causes poor SNR of demodulation signals in distributed sensing systems. Additionally, the intensity fading effect induced by high laser coherence can cause sensing blind spots, and the light intensity fading can also result in poor sensing consistency among multi-channels. Meanwhile, due to the influence of optical transmission loss, the sensing SNR of ordinary non-amplification SMF optic systems is limited at long distances. The fully continuous characteristics of backscattering signals in optical SMFs can also result in mutual limitations between the system response bandwidth and sensing distance. Therefore, scattering enhanced special optical fibers are introduced into distributed sensing systems based on Rayleigh scattering. By continuously changing the fiber material and structure, or introducing discrete scattering enhancement mechanisms, the distributed sensing limitations of ordinary optical SMFs are overcome in specific sensing parameters, sensing performance, and other aspects.

Thus, in some specific application scenarios that require high-precision detection, scattering enhanced optical fiber has irreplaceable advantages. In recent years, numerous research institutions and researchers have conducted research on scattering enhanced fiber optical distributed sensing systems and obtain significant results.

We focus on analyzing the scattering characteristics and noise suppression mechanisms of scattering enhanced microstructured sensing fibers, and elaborate on the types and precision preparation techniques of scattering enhanced fibers. Meanwhile, the performance improvement techniques of DAS and OFDR systems based on scattering enhanced microstructured fiber are summarized (Fig. 5 and Table 1), and the mechanism and typical applications of DAS SNR and sensitivity enhancement are discussed. The research progress of scattering enhanced hydrophone composite cables is elaborated (Table 2), and the construction of highly sensitive distributed hydrophone systems and their hydrophone applications are introduced (Table 3). Additionally, we summarize the high-density grating scattering enhanced microstructured fiber to achieve high-resolution, highly sensitive, and highly reliable fully distributed strain sensing based on optical frequency domain reflectometry (Figs. 10 and 11). Combined with highly reliable reconstruction algorithms, scattering enhanced microstructured spiral multi-core optical fibers are designed to achieve high-precision three-dimensional shape sensing and practical applications (Table 4).

In summary, we study the mechanism of distributed sensing efficiency enhancement from the perspective of scattering enhanced special optical fibers, introduce the automatic precision fully continuous writing technology, and focus on the principles of its optical time domain and optical frequency domain distributed sensing systems. Meanwhile, the research progress of distributed acoustic sensing and optical frequency domain reflection technology based on scattering enhanced microstructured fiber is summarized, and typical engineering applications based on the above two systems are summarized.

In the future, distributed sensing technology based on scattering enhanced optical fibers can still be improved and expanded in various aspects. For example, the material and structural parameters of scattering enhanced microstructured fiber can be optimized, and the high-efficiency and stable writing preparation process can be improved. Additionally, the scattering enhancement characteristics of optical fibers can be combined with intelligent AI algorithms to optimize sensing demodulation accuracy. Meanwhile, the high precision 3D shape sensing and scattering enhanced fiber distributed hydrophone can be further extended to various cross applications. As the scattering enhanced special sensing fibers further develop in the future, distributed sensing systems based on scattering enhanced fibers will play an irreplaceable role in most fields.

Optical fiber sensors play an increasingly important role in safety monitoring areas in the smart Internet of Things (IoT). Particularly, a fiber-optic distributed acoustic sensor (fiber-optic DAS) based on the phase-sensitive optical time-domain reflectometry (Φ?OTDR) technology provides a highly dense, cost-effective, and continuous environment measurement way over a wide range. All kinds of vibration sources can be sensed and located with high sensitivity and precision utilizing the widely laid ordinary telecommunication cables, and thus fiber-optic DAS has been applied in various ground listening applications, such as natural disaster prediction of ocean-floor seismic activity, volcanic events, and earthquake, energy exploration in oil and gas industry, and civil infrastructure monitoring in the pipelines, railways, and perimeters. It leads to a new generation of large-scale fiber-optic IoT for ground and underwater listening technology. From the current research status in China and abroad, DAS is becoming mature in its hardware performance, such as the demodulation fidelity, sensing distance, detection bandwidth, and sensitivity, which are all approaching their perfection. However, with the rapid advance of DAS applications, the complicated and ever-changing environments for large-scale monitoring have brought about challenges of high false alarm rates due to its advantages of high sensitivity. It is difficult to achieve high-precision detection, recognition, and positioning of perceived vibration and acoustic targets, which has become the biggest technical bottleneck restricting the large-scale application of DAS technology. In recent years, driven by the development of advanced signal processing and artificial intelligence (AI) technology, the signal processing methods of fully intelligent DAS with high accuracy and real-time performance in practical complex environments have become a research hotspot and focus in the field of fiber-optic sensing. The signal processing method in DAS plays a crucial and decisive role in improving the intelligent perception ability of the entire system.

We review the current research status of signal processing methods in smart fiber-optic DAS entering the deep learning stage, from mainstream supervised learning to unsupervised, semi-supervised, and transfer learning, from single-source detection to multi-source aliasing detection, and from single-task recognition or localization to simultaneous implementation of recognition and localization tasks, and we predict possible research directions for further improving the intelligent processing performance and perception ability of DAS in the future. Firstly, the typical fiber-optic DAS system structure and its vibration/sound sensing mechanism (Fig. 2), and the smart DAS and its signal processing architecture in smart city monitoring applications (Fig. 3) are introduced. Then, the signal processing methods based on deep learning are explained in detail, which includes the main stream of supervised learning methods based on multi-dimensional information extraction, and semi-supervised, unsupervised learning, and cross-scene transfer learning methods in DAS. For the supervised learning method, it includes DAS signal recognition models based on temporal information extraction, such as one-dimensional convolutional neural networks (1D-CNNs) (Fig. 4), multi-scale convolutional neural networks (MS-CNNs) (Fig. 5), multi-scale and contextual temporal relationship mining methods (Figs. 6-7), and the two-dimensional recognition models based on time-frequency (Figs. 8-11), time-space (Figs. 12-14), and space-frequency (Fig. 15) information extraction technologies. Besides, some other supervised methods are also included, for example, recognition models based on attention-based long short-term memory (Fig. 16) and the fusion of manual features and deep features. It proves that the combination of traditional empirical rules and deep learning networks can further reduce the false alarm rate of the system. In response to the problem of insufficient labeled samples in new scenarios in practical applications, several semi-supervised recognition methods based on the 1D-SSGAN (one-dimensional semi-supervised generative adversarial network), SSAE (sparse stacked autoencoder), and FixMatch models have been involved to achieve accurate recognition of DAS signals with a small amount of labeled data and a large amount of unlabeled data. Furthermore, the SNN-based DAS unsupervised learning network (Fig. 17) and the cross-scene transfer learning network based on AlexNet+SVM (Fig. 18) also appear to improve the generalization ability of DAS signal recognition methods. In order to evaluate the performance of these recognition models, we introduce seven indicators for evaluating the recognition accuracy and four indicators for the processing time of the algorithms. The above key DAS recognition methods and their performance are statistically compared in Table 2. At last, the new challenges of smart DAS sensing, from single-source detection to multi-source aliasing detection, from target recognition to localization, and from a single task to multi-task processing, as well as other methods to enhance its intelligent perception capabilities, have also been introduced.

Further improvement of signal processing and its sensing capabilities still faces new challenges and opportunities and will open a new chapter in fully intelligent DAS. Stable, accurate, real-time, and efficient signal recognition in DAS in new complicated application scenarios remains a research hotspot in the field of distributed fiber-optic sensing in the future, including: 1) improving the generalization ability of DAS recognition models in cross scenarios; 2) significant improvement in real-time processing capabilities in DAS; 3) improvement of multi-task processing ability in DAS; 4) implementation of high-performance on-chip DAS.

Fiber optic sensing technology has been widely applied in multiple fields and has received good feedback due to its advantages of strong anti-interference ability, small size, high sensitivity, long transmission distance, and intrinsic safety. Distributed sensing technology (OTDR, Φ-OTDR, and OFDR) based on Rayleigh scattering can achieve long-distance, large-scale, and multi-parameter monitoring, which has attracted more attention in applications. With the increasing demand for monitoring length and range in various application fields, the traditional methods of increasing light source power and detector detection limit have reached their peaks in increasing the system distance. The development of new scattering enhancement methods is urgent, so as to enhance the sensing distance of distributed sensing technology.

We review several ways to enhance the scattering light in fibers by enhancing their scattering coefficients and backscattering ability from the perspective of fibers, as well as the limitations and application scenarios of these methods. We also provide a detailed introduction to the latest scattering enhancement method, which enhances scattering by enhancing the backward collection coefficient and has potential development prospects in future distributed sensing.

The research progress of fiber optic scattering enhancement methods is as follows.

1) Enhancing fiber scattering by increasing the scattering coefficient. It is commonly used to increase the scattering coefficient of optical fibers through irradiation, microstructure, and nanoparticle doping to enhance the Rayleigh backscattering light of the fibers.

The irradiation method is to increase the refractive index disturbance in the fiber by ultraviolet or radiation irradiation. It is simple to operate and has continuous scattering enhancement. However, it will increase the loss of the optical fiber and reduce the sensing distance. At the same time, the preparation speed of the optical fiber is slow, requiring the removal of the coating layer and resulting in a decrease in mechanical strength. Therefore, the scattering enhanced fiber prepared in this way is difficult to apply to engineering environments.

The microstructure method refers to the formation of weak gratings, reflection points, Fabry Perot cavities, and other junction microstructures in optical fibers through ways such as ultraviolet, femtosecond, and arc discharge, resulting in significant refractive index changes. This method is flexible and has higher controllability, and it can be continuously prepared in large quantities without removing the coating layer and changing the mechanical strength of the optical fiber. However, it still increases the loss of the optical fiber and reduces the sensing distance, and the distribution of microstructures in the optical fiber is discrete, forming a minimum sensing area between two adjacent points, which reduces the spatial resolution of the distributed sensing system. This method is suitable for applications in sensing scenarios that do not require high spatial resolution.

The doping method of nanoparticles increases scattering in fibers by doping elements such as germanium, calcium, barium, gold, and magnesium. It has continuity, and the scattering enhancement is more obvious. In addition, it can be directly prepared through fiber drawing, which ensures the mechanical properties of the fiber. However, the high scattering enhancement also brings about a significant increase in losses. The losses of nano-doped fibers are generally two or three orders of magnitude higher than those of irradiation and microstructure and are generally applied in sensing scenarios with short distances and high signal-to-noise ratios.

2) Enhancing fiber scattering by increasing the backscattering collection coefficient. The method of increasing the backscatter collection coefficient to enhance fiber scattering theoretically does not increase the loss of the fiber, which mainly includes three types: plastic optical fiber, multimode optical fiber, and ultra long adiabatic tapered optical fiber.

Both polymer fiber and multimode fiber can increase the backward collection coefficient by increasing the numerical aperture, but the material absorption loss of polymer fiber itself is greater than that of quartz fiber. Therefore, it is generally applied in short-distance sensing scenarios. Multimode optical fibers have significant mode losses, and dispersion over long distances can degrade the spatial resolution of the system. It is commonly used in scenarios with lengths of kilometers.

Our team has proposed an ultra long tapered single-mode fiber that can increase the backward collection coefficient of the fiber to enhance scattering, without causing external losses. It can break through the distributed sensing long-distance limit of single-mode fiber and achieve sensing with an equal scattering signal-to-noise ratio at each point, and it can be applied to ultra long sensing scenarios with a length of above 150 km. Ultra long tapered single-mode fiber also has the advantage of enhancing the performance of fiber Bragg grating (FBG) arrays. Engraving FBG arrays on tapered optical fibers can effectively increase the remote reflection signal of FBG and expand the number of arrays, which has great development potential for future high-tech composite distributed sensors.

Temperature measurement is a common requirement in the daily life of human beings and production activities. Abnormal and abrupt temperature variations in such diverse important fields as civil engineering, industrial machinery, aerospace, and infrastructure can cause significant economic losses and are even life-threatening. High-accuracy and real-time measurements of temperature distributions are thus demanding, and different technologies have been developed accordingly. Specifically, optical fiber-based temperature sensors have been demonstrated as unparalleled modalities in harsh-environment applications, which is due to their distinct advantages over conventional electrical devices, such as small size, light weight, immunity to electromagnetic interference and chemical corrosion, and importantly, the capability for spatially distributed sensing.

As one of the most important optical fiber temperature sensing technologies, Raman distributed optical fiber temperature sensing (RDTS) systems have outstanding advantages of electromagnetic compatibility, long sensing distance, and wide temperature range. Thus, they are widely adopted in the safety monitoring of infrastructure structures, the oil and gas industry, fire detection, and many other fields. The first RDTS system can be dated back to 1985, when a sensing distance of 1 km is realized with a spatial resolution of 3 m based on the first generation of opto-electronic devices. Recent years have witnessed tremendous growth and advancement in light sources, detectors, sensing elements (i.e., optical fibers), and signal demodulation techniques, which leads to new generations of RDTS systems with significantly enhanced performance measures. It is of significance to summarize and discuss the existing research progress and future development trends for further development in RDTS systems.

We review the advanced RDTS technology based on the performance improvement methods mainly in three aspects of system optimization, temperature demodulation, and data processing. Firstly, the system optimization is focused on the system structures and components. The structure optimization is mainly conducted as the double-ended or loop structure, which involves the adoption of optical switches or an additional mirror at the optical fiber end to reduce the uncertainty of measured temperature in the long-term utilization. The components optimization mainly involves the optimization of the light source, detector, and sensing fiber. In the light source, related research concentrates on the wavelength and pulse width selection, application of pulse coding technology, and employment of new-type light sources. In terms of detectors, with the rapid development of single-photon detectors, their applications in ROTDR systems are bound to bring greater performance improvement. In terms of sensing fiber, the adoption of dispersion shift fiber (DSF), dispersion compensation fiber (DCF), and low water peak fiber provide a new idea for long-distance sensing. Few-mode fibers (FMFs) combine the advantages of single-mode and multi-mode fibers and yield better performance in RDTS systems. Secondly, in temperature demodulation, a variety of calibration and compensation methods are summarized in the problems of optical power fluctuation, Rayleigh scattered light residue, and differential temperature sensitivity for Stokes and anti-Stokes. Thirdly, the data processing mainly involves the applications of various denoising algorithms to improve the spatial resolution and signal-to-noise ratio (SNR) of the systems.

Subsequently, a global market survey of RDTS systems is summarized. The main research and development institutions and manufacturers at home and abroad are listed by combining the performance comparison of their typical products. The typical applications in various engineering scenarios are presented.

The current research achievements of RDTS systems can be comprehensively summarized from three perspectives including system optimization, temperature demodulation, and data processing. In terms of system optimization, pulse coding technology can enhance the SNR by over 10 dB without compromising the system's spatial resolution. The application of genetic optimization algorithms to find the optimal pulse coding sequence maximizes the advantages of pulse coding, leading to more than seven-fold improvement in temperature resolution. By introducing chaotic laser sources into the RDTS system and combining relevant demodulation algorithms, the spatial resolution can be increased from 50 m to 0.3 m. The utilization of FMFs provides a viable solution to overcome both the low SRS threshold of single-mode fibers (SMFs) and the large modal dispersion of multi-mode fibers (MMFs). Compared to SMF (MMF), this approach significantly enhances system temperature (spatial) resolution with minimal influence on spatial (temperature) resolution. In the context of temperature demodulation, various calculation methods of temperature calibration are introduced to mitigate the effect of system power fluctuations, signal crosstalk, and losses on measurement results. Appropriate calibration methods can improve temperature accuracy by over 14 ℃. Regarding data processing, in recent years, algorithms based on image processing and artificial intelligence have been successively applied to RDTS systems, both improving SNR and contributing to spatial resolution enhancement. Under the applications of these algorithms, spatial resolution can increase by five times, and temperature resolution reaches the order of 0.01 ℃.

Furthermore, we summarize the domestic and international market trends of the RDTS systems and major vendors and compile information on typical products offered by these vendors. However, compared to foreign products, there is room for further improvement in the technical indicators of domestic products.

With an increasing number of proposed ideas, there is considerable potential for enhancing the performance of RDTS systems. Achieving miniaturization by device integration and enhancing portability is a long-term theme for the development. The application and optimization of various artificial intelligence algorithms will also equip the systems with faster processing speeds and superior performance. Additionally, rational design optimization of sensing optical fibers and other system-related aspects could further expand the systems' applications in extreme environments such as ultra-high and low temperatures, and nuclear radiation. Many potential applications of this deserve further exploration.

Optical pulse coding (OPC) has caught much attention in optical fiber sensing in recent years, especially when combined with phase-sensitive optical time domain reflectometry (Φ-OTDR).

In the 1970s, optical fiber sensing technology emerged rapidly with the development of optical fiber communication technology, and it employs optical fiber as the sensing medium or optical transmission path to sense changes in the surrounding environment by the characteristic changes of light waves. With the increasing demand for sensors in society and the continuous maturation of optical sensing technology, optical fiber sensing systems have been widely adopted. These systems based on the light scattering principle can carry out long-term monitoring in harsh environments and achieve the measurement of physical quantities with large spatial scales or high spatial densities by continuous sensing points in optical fibers. Meanwhile, they have become a key component of borderline security, firefighting early warning, pipeline monitoring, transportation line supervising, and large-scale structural health monitoring among other fields.

Based on Raman scattering, Brillouin scattering, and Rayleigh scattering, a variety of optical fiber sensing schemes can be implemented. Rayleigh scattering is a kind of elastic scattering caused by refractive index changes in the optical fiber and has a faster response speed compared with the other two scattering methods. Additionally, based on the interference effect, Rayleigh scattering-based optical fiber sensing is more sensitive to the changes in the measured parameters. Φ-OTDR based on Rayleigh scattering is one of the most important applications of distributed acoustic sensing (DAS) and quasi-distributed acoustic sensing (Q-DAS), with fast response and high sensitivity. Despite the sound performance of Φ-OTDR, it is still affected by some of its factors, such as signal-to-noise ratio (SNR), spatial resolution, and transmission distance. The mutual constraints among these factors can limit the Φ-OTDR performance. By coding the probe pulses injected into the fiber, the SNR of the sensing signal can be significantly increased without increasing the peak power of the pulses, thus avoiding nonlinear effects. Meanwhile, the single-pulse response can be obtained after decoding at the receiving end, and the spatial resolution of the system is determined by the length of a single pulse rather than the entire probe pulse sequence, thus maintaining the spatial resolution and receiving a high-SNR sensing signal. In most cases, OPC is a viable solution to meet the demands of high accuracy, long distance, and high sensitivity sensing because it can overcome the limitations among various key parameters.

Regarding the combined applications of OPC technology and Φ-OTDR, the development of optical pulse coding technology in optical fiber sensing is firstly reviewed, and its applications in sensing systems based on Raman scattering, Brillouin scattering, and Rayleigh scattering are introduced. Meanwhile, we present the representative studies of researchers in China and abroad and conduct a comparison of the performance enhancement brought by different coding schemes and traditional schemes. The development of the technique is summarized as shown in Tables 1-4, with the system performance of the different schemes compared. Then the coded Φ-OTDR technical schemes proposed by our group are presented in more detail, including Φ-OTDR based on unipolar and bipolar Golay coding, and the suppression of interference fading and frequency drift therein. Finally, the Φ-OTDR technical route based on orthogonal codes with the same carrier proposed by our group is highlighted.

In recent years, under the joint efforts of several research teams at home and abroad, optical pulse coding technology has been successfully integrated with optical time domain reflection technology in depth, which has led to remarkable development in the direction of optical fiber sensing based on optical time domain reflection technology. By various innovative ways of combining optical pulse coding technology with Φ-OTDR, the constraints among the key performance parameters of Φ-OTDR can be overcome. Optical pulse coding can help Φ-OTDR achieve distributed optical fiber sensing with long distance, high SNR, high spatial resolution, and quasi-distributed optical fiber sensing with long distance, high SNR, and large bandwidth. The acoustic wave sensing technology based on optical pulse coding can still be further extended to engineering applications, such as vehicle positioning, seismic wave detection, and perimeter security. It is worthwhile to deeply explore high-level applications of the technology in engineering fields in the future.

Based on Rayleigh, Brillouin, and Raman scattering, and weak reflection arrays in optical fiber, distributed optical fiber sensors (DOFSs) can achieve real-time monitoring with long range and high spatial resolution for multiple parameters such as optical fiber loss, temperature, strain, vibration, and sound. As a result, DOFSs catch more and more attention. The received signals from the sensing fiber of a DOFS are normally very weak and thus the received signal-to-noise ratio (SNR) is quite small. Although the SNR can be enhanced by increasing the optical pulse power launched into the sensing fiber, it is generally limited by fiber nonlinearity and then is upperbounded. A pulse with long duration can also be employed to improve the SNR but the spatial resolution is sacrificed. A better alternative to enhance the SNR without spatial resolution loss is to adopt an optical pulse sequence with some coding at a fairly low power to avoid fiber nonlinearity. Therefore, it has become an essential technique to enhance the performance of a DOFS by a long coded pulse train in DOFS. As technical characteristics of various DOFSs are different, applicable coding scheme has to be carefully designed for a particular DOFS. Design considerations may include several aspects such as code sequence, modulation format, detection scheme, and decoding methods. Hence, it is important and necessary to summarize the existing research on DOFS coding techniques for performance enhancement to guide the future development of this field.

In principle, the response of a DOFS with coding can be considered as the convolution of the coding sequence with the impulse response of the sensing fiber. The aperiodic autocorrelation of the coding sequence is utilized to construct an impulse function, and the impulse response of the sensing fiber can be recovered by correlating the DOFS response with the coding sequence itself at the receiver site. Therefore, the aperiodic autocorrelation characteristics of the coding sequence are critical, which can be evaluated from three aspects of coding gain, spatial resolution, and crosstalk suppression. A sequence with better aperiodic autocorrelation performance is always pursued. The sensing principle of a DOFS also exerts some effects on coding sequence selection. Unipolar sequences are frequently employed in DOFSs with intensity detection. A typical unipolar sequence is Simplex sequence. Bipolar sequences such as Golay complementary sequences can also be converted to unipolar sequences, and they are popular in DOFSs with phase detection and can be implemented by binary phase shift keying modulation via Mach-Zehnder modulator (MZM) (Fig. 3). A widely employed bipolar sequence is Golay complementary sequence. Polyphase unimodular sequences have also been proposed recently in DOFSs with phase detection for much better crosstalk suppression capability (Figs. 5-8). Such sequences have been realized via modulation by an acoustic-optical modulator (AOM) (Fig. 4).

Various DOFSs with coding techniques have been proposed. For Rayleigh scattering sensors with incoherent optical sources, Simplex sequence, Golay complementary sequences, CCPONS, and other unipolar sequences have been put forward to improve performance such as dynamic range, spatial resolution, and measurement speed. For Rayleigh scattering sensors with coherent optical sources, both unipolar and bipolar sequences have been proposed. Multi-input-multi-output (MIMO) technique has also been demonstrated in a Rayleigh scattering DOFS with polarization multiplexing and coding to increase measurement bandwidth. For Raman scattering sensors, Simplex sequences and other unipolar sequences are popular. The coding sequence performance is frequently degraded by the transient effect of erbium-doped fiber amplifier (EDFA). Many schemes have been proposed to demonstrate their anti-degradation capability. Coding techniques have also long been explored in Brillouin scattering sensors to improve the performance in measurement accuracy, spatial resolution, and measurement speed. In addition to conventional correlation based decoding schemes, deconvolution-based decoding techniques have also been presented. Weak fiber Bragg grating array (WFBGA) is an emerging DOFS, with coding techniques explored in such a DOFS. For WFBGA with interrogation based on intensity, Golay complementary sequences with return zero (RZ) code format have been discussed. For WFBGA with interrogation based on phase, MIMO techniques with Golay complementary sequence and polyphase unimodular sequence using polarization multiplexing have been demonstrated, with much better crosstalk suppression performance (Figs. 9-10).

After decades of development, DOFSs have been widely employed in various areas and a lot of applications have been developed based on DOFSs. Those applications have raised increasingly higher requirements for DOFS performance. Coding technique is an important technical method to enhance the performance. We analyze the underlying principle of coding technique and manifest the connection between sensing performance and characteristics of coding sequences. Meanwhile, features, performance, and implementation of some widely used sequences in DOFSs are summarized. We analyze the technical characteristics and applicable coding schemes of DOFSs based on Rayleigh scattering, Raman scattering, Brillouin scattering, and WFBGA, and summarize the improvement of SNR, spatial resolution, sensing bandwidth, and sensing range by employing the coding techniques. As the DOFS technology is advancing, coding techniques will be further developed, which calls for the research on higher-performance coding sequences and their implementation schemes. Additionally, coding techniques will also integrate other techniques such as frequency division multiplexing, wavelength division multiplexing, spatial division multiplexing, and channel equalization to explore the fiber characteristics of low loss and high bandwidth. Finally, the space for DOFS performance improvement will be expanded.

Brillouin optical time-domain analyzer (BOTDA) has been widely studied and employed in academia and industry due to its capability of providing information about the spatial distribution of targeted quantities over a long optical fiber. Several advanced techniques have been proposed during the last decade to further enhance the signal-to-noise ratio (SNR) beyond the conventional single-pulse BOTDA. As one of the most efficient approaches, optical pulse coding technology has been extensively developed, and it launches one or several trains of pulses into the sensing fiber, with the exploitation of code types such as Golay, Simplex, Cyclic, genetic-optimized codes, and their derivatives. Meanwhile, this technology can assist BOTDA to further achieve considerable SNR improvement without compromising the spatial resolution and measurement time and thus has significant superiority and application prospects. Despite many corresponding advances, there are still a series of challenges in cost reduction and performance improvement. Therefore, it is important and necessary to summarize the existing research to inspire the future development of this field more rationally. Focusing on the BOTDA sensors based on optical pulse coding, we introduce the principle of several mainstream optical pulse coding technologies employed in distributed fiber sensors, and discuss the research progress of BOTDA sensors based on optical pulse coding in recent years.

The fundamentals of conventional optical pulse coding technologies applied in BOTDA sensors are reviewed in detail, including unicolor unipolar codes (Golay, Simplex, Cyclic, and deconvolution-based codes), color codes, and bipolar codes. The theoretical coding gain of such present mainstream coding schemes is also summarized, all of which can be approximately regarded as proportional to the square root of the code length.

It is significant and necessary to improve and optimize the coding system for reducing the influence of system imperfections and get as close as possible to the theoretical coding gain. With the development of BOTDA sensors based on optical pulse coding technology in the past decades, a series of optimization schemes have been proposed in coding parameters, post-processing algorithms, coding system designs, and coding types (Fig. 8). This makes the basic theory of pulse-coded BOTDA more mature and complete, and provides theoretical guidance for realizing the highest possible sensing performance of pulse-coded BOTDA. The performance optimization of pulse-coded BOTDA sensors from the above four aspects is discussed in detail. Additionally, the analysis results of performance and shortcomings of different mainstream unicolor unipolar coding schemes in practical applications are summarized in Table 2, which demonstrates the comparative application advantages of the deconvolution-based optical pulse codes over the other current codes.

In addition to the optimization measures of BOTDA sensors based on optical pulse coding, the combination of optical pulse coding technology and other advanced technologies such as differential pulse-width pair, optical pre-amplification, pulse pre-pump, distributed optical amplification, and digital signal processing has been put forward to further enhance the comprehensive performance of pulse-coded BOTDA sensors. The performance of reported BOTDA sensors based on optical pulse coding is sorted chronologically and several representative results are summarized in Table 3. The bipolar complementary Golay code based on a three-tone probe yields the highest overall performance thanks to its high coding gain, low optical noise, and high robustness.

Optical pulse coding technology has been studied for decades and has fully proven to work well in BOTDA sensors with long sensing distance (> 25 km) and high spatial resolution (<5 m). Compared with the optimized single-pulse BOTDA sensors, the overall performance of pulse-coded BOTDA sensors with a proper code type is greatly improved with the assistance of optimized system parameters and post-processing algorithms. The present mainstream coding schemes adopted in distributed optical fiber sensors include unicolor unipolar coding, color coding, and bipolar coding, whose theoretical coding gain can all be approximately proportional to the square root of the code length. The analysis of unicolor unipolar coding schemes is relatively mature. With the aspects of hardware and software costs, robustness to baseline fluctuations, tolerance to signal-dependent noises, tolerance to non-uniform code envelop, and ability of arbitrary energy boost considered, the deconvolution-based code exhibits performance advantages in nearly all aspects. This provides an economical and effective solution to sensing performance improvement for any given sensing device, without any auxiliary hardware or additional measurement time. Based on the existing research results, how to overcome the optical noises and the higher-order non-local effects to obtain a longer coding length to further improve the SNR has become a current challenge.

As opposed to unicolor unipolar coding, since color coding and bipolar coding are researched at a later stage and are more costly in practical applications, the solution of reducing the cost of coding without compromising its performance is a current challenge. Further in-depth analysis and exploration of color coding and bipolar coding will help evaluate and compare the performance of different coding schemes more comprehensively and rigorously for adapting to different application requirements.

Furthermore, the combination of optical pulse coding technology and other advanced technologies such as distributed amplification technology is a common strategy for performance enhancement. By taking distributed Raman amplification technology as an example, the optimization standards of optical pulse coding technology and distributed Raman amplification technology are different, and the direct superposition of their respective optimization schemes when such two technologies are applied simultaneously may cancel out the benefit. Till now, there is still a lack of specific optimization standards and targeted optimization schemes for the combination of such two technologies. Therefore, it is of significance for further improving the performance of distributed sensors to conduct in-depth investigations on their targeted optimization schemes when the optical pulse coding technology and other advanced technologies are adopted simultaneously.

In-situ spatial distribution acquisition of biochemical substances is particularly important for gas-liquid distribution monitoring, cell analysis, tumor detection, drug design, and other fields. Optical fiber biochemical sensors are ideal tools for biochemical detection due to their unmarked, in-situ, fast, and accurate properties. However, existing optical fiber biochemical sensors only obtain the content of a single point of biochemical substances, making it difficult to obtain the spatial distribution information. The distributed biochemical sensing method with hundreds or thousands of sensors continuously distributed along the optical fiber axis can achieve this goal. We start with quasi-distributed optical fiber biochemical sensing and comprehensively review the latest progress of distributed optical fiber biochemical sensing in gas sensing, refractive index (RI) sensing, and biochemical sensing. Finally, the development prospects and current challenges of distributed optical fiber biochemical sensors are discussed. The research on distributed optical fiber biochemical sensing is expected to lead the current study of single-point discrete optical fiber biochemical sensing to the multi-point continuous distribution development. Additionally, it has the potential to be a new powerful tool in fields such as chemistry, biology, and medicine.

For quasi-distributed biochemical sensing, the gas sensing development based on multiplexing mainly focuses on reducing noise and improving the limit of detection (LOD). With the deepening sensor research, the utilization of different sensors and the combination of multiple methods have greatly improved the quasi-distributed gas sensing accuracy, quantity, and efficiency. In terms of quasi-distributed fiber optic RI sensing, the RI sensitivity has greatly improved with the optimized sensor preparation process. Based on the multi-channel advantages, it is possible to achieve quasi-distributed RI sensing in multiple regions. However, due to the sensor size, quasi-distributed RI sensors cannot achieve spatial distribution recognition in solutions. Therefore, the research on distributed RI fiber optic sensors is very necessary. Quasi-distributed fiber optic biochemical sensing has achieved simultaneous sensing of multiple biological tissues or chemical substances. However, like quasi-distributed RI sensors, quasi-distributed fiber optic biochemical sensors cannot achieve position monitoring of biological tissues. This can be addressed in distributed fiber optic biochemical sensing.

Compared to the quasi-distributed sensing, distributed gas sensing has clearer requirements in sensing distance and spatial resolution. Gain modules such as EDFA are gradually added to sensing systems for long-distance detection. Meanwhile, the sensing method has gradually changed from quasi-distributed multi-channel sensing with multiple gas cells to distributed single channel sensing with multiple gas cells. Additionally, distributed RI sensing improves its spatial resolution to the millimeter level, which is of significance for detecting concentration distribution and substance localization in solutions. However, due to the short development period of distributed RI sensing, there are still powerful development prospects in spatial resolution and sensitivity. For the newly developed distributed biochemical sensing, distributed biochemical research mainly focuses on pH measurement. With the improved spatial resolution, it is possible to achieve micro localization of biochemical substances such as tumor cells. The sensing spatial resolution of distributed biosensors based on OFDR can be improved to less than 100 micrometers or even a few micrometers. When the spatial resolution approaches cell size, fiber optic probes can be adopted for individual cell localization. We believe that these manifestations are essential for the localization and subsequent treatment of tumor cells. It can be foreseen that distributed biochemical sensors based on OFDR will become the most active research field in the entire distributed fiber optic biochemical sensing.

We start with quasi-distributed fiber optic biochemical sensing and comprehensively review the latest progress of distributed fiber optic biochemical sensing in gas sensing, RI sensing, and biochemical sensing. Quasi-distributed fiber optic biochemical sensors are based on methods such as time division multiplexing, space division multiplexing, wavelength division multiplexing, and frequency division multiplexing, and they can be utilized to characterize the performance of multiple sensors in an optical system by multiplexing. These methods have advantages in measuring multiple gases, biochemical substances, or multi parameters. However, due to the sensor size and structure, it is hard to achieve high spatial resolution, such as distributed sensing at the micron level. The distributed fiber optic biochemical sensor based on OTDR or OFDR demodulates the backscatter in the optical fiber, which can not only obtain information about the biochemical substance contents but also obtain spatial distribution information of these contents. More importantly, OFDR and OTDR technologies can choose appropriate spatial resolution based on the differences of sensing targets. However, there are still many challenges in distributed fiber optic biochemical sensing at present. The first challenge to be addressed is the sensing sensitivity. Compared to traditional split and quasi-distributed sensing, the sensitivity of distributed sensors is still 1-2 orders of magnitude lower. Furthermore, how to improve the backscatter signal-to-noise ratio of sensing fibers is also a problem to be overcome currently. Although distributed RI sensing based on OFDR has gradually become a hot topic, it is currently limited by the evanescent field excitation method. Additionally, the relatively short sensing distance limits the application scenarios. Therefore, in future research, some sensitization methods of fiber optic single point biochemical sensors can be referenced, such as gold plating, silver plating, nanoparticles, and two-dimensional materials. Meanwhile, based on distributed RI sensing, it is necessary to further expand the research on distributed biochemical sensing by combining immune reactions, chain reactions, and other biochemical reactions. By utilizing the high spatial resolution characteristics of OFDR, it is ultimately possible to perceive the spatial distribution of biochemical substance contents in subcellular structures such as nucleic acids, DNA, ions, and enzymes. This will potentially become a powerful new tool in life sciences.

Distributed fiber optic sensing technology based on optical frequency domain reflectance (OFDR) has found extensive application in areas such as monitoring the health of structures and measuring temperature/strain in harsh environments. It has proven advantageous due to its ability to provide high spatial resolution, compact design, lightweight nature, and excellent immunity to electromagnetic interference. However, since the backward Rayleigh scattered light used for localization in OFDR is usually weak, the reduction in similarity (SD) between the reference spectrum and the measurement spectrum due to noise can significantly impact the robustness and accuracy of the system's measurements, especially in situations involving long distances, high temperatures, or a significant number of range strains. To address this problem, in this paper, we develop a tuning nonlinearity compensation model for tunable laser sources, finding that the residual tuning nonlinearity may lead to a random position deviation (PoD) for each sensing gauge. Based on the PoD statistical analysis, we build a system for evaluating the SD between the reference and measurement spectra. Combining with Kalman prediction and local search, the proposed method can match the reference and measurement spectra efficiently and accurately, resulting in compensation for the random PoD introduced in the sensing gauge of interest. We hope to extend the sensing range while realizing increased spatial resolution, robustness, and speed.

The research on tuning nonlinearity starts from the schematic diagram of a polarization diversity OFDR system. By examining the origins of its residual tuning nonlinearities, we employ statistical techniques to explore how they impact the PoD in each sensing gauge. The analyses illustrate that the innate noise from the tunable laser, similar to the outer strain or temperature variations, could contribute to the PoD. In particular, because of the statistical portrayal of the residual tuning nonlinearities, the additionally generated PoDs exhibit an approximately standard distribution. Based on this finding, we further design a process based on Kalman filtering (KF) and local search to compensate for the random PoDs from tuning nonlinearities, wherein two judgment conditions (JC1 and JC2) determine whether to enter/break the local search loop. Compared with other post-filtering methods, this method updates the measurement information by satisfying JC1 < T_JC1 or minimizing JC2. This procedure is closer to real sensing scenarios and therefore improves the SD. Besides, we start the local search loop from the center (j = ±1) with higher probabilities to the distal (j = ±M) and break the loop once JC1 < T_JC1. Thus, the presented strategy could accelerate the search process.

We compare the distributed sensing results recovered by the proposed method with the existing methods (Fig. 5). It is evident that the currently available approaches have limitations in terms of measurement length and strain/temperature measurement range due to the residual tuning nonlinearities. In contrast, the presented method can recover the strain/temperature distributed along the fiber axis without observing outliers, suggesting it can sufficiently compensate for the innate SD degradation due to the residual tuning nonlinearities. In particular, the robustness of the proposed method has a significant advantage when the measured strain or temperature is beyond 5000 με or 300 ℃, respectively. Additional examinations of the PoD random variations caused by the tuning nonlinearities and external stress indicate that the amplitude and range of the former are weaker (Fig. 7), implying that it is typically confined and temporary. The requirement to implement the adaptive judgment conditions JC1 and JC2 is verified in parallel. The distributed fiber optic strain/temperature sensing equipment and its software can achieve a sensing distance of greater than 150 m and a spatial resolution of 5 mm (Fig. 9), and the completion time of a single measurement under the full sensing range and the highest spatial resolution is less than 6 s. The system could measure strains varying from 2000 to 10000 με at about 140 m. A lateral comparison of each curve reveals that the shape of the data sets is similar, and the height of the "platform" is directly proportional to the applied strain. It is evident that the system effectively measures the magnitude and location of the sensing event; a horizontal comparison of the data sets demonstrates that the shape of the data sets is comparable, and the height of the "high platform" is linearly correlated to the applied strain.

In conclusion, the random PoD due to the residual tuning nonlinearities is theoretically verified to decrease the SD between the reference and measurement spectra in OFDR systems. A novel local search and dynamic prediction method based on KF is then proposed. This method can effectively compensate for the random PoD by local search and accelerate the search process by the KF prediction. Experiments show that the proposed method can significantly improve the robustness of the sensing system under the limited range (temperature of 450 ℃ and strain of 10000 με) sensing application. Moreover, it can compress the computation to 5.8%-28.6% of that without dynamic prediction operations.

The phase-sensitive optical time domain reflectometer (Ф-OTDR) system can quantitatively characterize the vibration events along the sensing fiber by extracting the phase information of Rayleigh scattered light, which features a fast response up to milliseconds or sub-milliseconds and a detection sensitivity of nanostrains. Ф-OTDR has been widely used in sensing scenarios such as seismic wave monitoring, perimeter intrusion monitoring, and pipeline leak monitoring. Due to the interference fading effect in the Ф-OTDR system, the phase information extracted at the lower signal amplitude is distorted, which results in the incorrect response of the intensity and frequency of the vibration event. Then, it introduces frequent false alarms in practical engineering applications. Over the past decade and beyond, tremendous efforts have been devoted to addressing this issue, typically including the utility of special optical fibers (i.e., seven-core fiber and periodic ultra-weak Bragg grating array), complex demodulation algorithm (i.e., phase-shifted double pulse method and tracking algorithm), and multi-frequency pulse modulation [i.e., phase-shifted double pulse method and multi-branch acoustic-optic modulator (AOM) modulation]. From an applicative point of view, a simple multi-frequency pulse modulation Ф-OTDR system for suppressing the interference fading effect, which features flexible controlling of the frequency component of the probe optical pulse without sacrificing the response bandwidth and spatial resolution, is still a research gap to date.

In this paper, for the first time (to the best of our knowledge), an AOM for generating multi-frequency probe light is employed in the Ф-OTDR system. The modulation frequency interval and number can be flexibly controlled by an arbitrary waveform generator (AWG) within the operating bandwidth of the broadband AOM. Subsequently, the continuous multi-frequency probe light is modulated into pulsed light through a general AOM. The multi-frequency beat signals can effectively suppress the inherent interference fading effect along the sensing fiber by appropriate filtering, demodulation, and multiplexing; ultimately realizing the high-fidelity demodulation of the Ф-OTDR system. We believe that the proposed scheme can provide a practicable way toward a simple and compact structure, precise phase delay control, and flexible and controllable frequency components without sacrificing response bandwidth and spatial resolution to suppress the interference fading effect in Ф-OTDR.

The multi-frequency modulation principle and experimental schematic diagram are shown in Fig. 2 and Fig. 3, respectively. We utilize the AWG to generate three unequal interval radio freqency (RF) signals, including the frequency components of 340, 390, and 450 MHz. The multi-frequency modulation RF signal is simultaneously loaded on the broadband AOM. Then, the multi-frequency continuous light is modulated by an AOM with a fixed drive frequency of 300 MHz. The multi-frequency probe light has the characteristics of a down-shift frequency of 300 MHz, pulse width of 100 ns, and repetition frequency of 10 kHz. A 2 Gsa/s data acquisition card acquires the multi-frequency beat signals. The 40, 90, and 150 MHz beat signals are obtained by digital bandpass filtering, respectively, as shown in Fig. 4. Figure 5 verifies that the beat signals of different frequency components present various intensity distributions, which also further indicates the different interference fading locations versus different frequency components. A piezoelectric ceramic transducer (PZT) simulates vibration events across the sensing fiber. The experimental results of the vibration events cannot be effectively reconstructed by the generic method with a single-frequency probe Ф-OTDR, as shown in Fig. 6. It can be seen that the reconstructed phase information originating from 90 MHz beat signal is distorted at the positions of the low amplitude points. Figure 7 shows the experimental results of the vibration events via the proposed multi-frequency modulation Ф-OTDR. More specifically, the high-fidelity reconstructed phase information is obtained by multiplexing three-frequency beat signals based on the amplitude intensity evaluation criteria. The demodulation results of 25 Hz are consistent with the frequency of the PZT vibration signal. The axial strain generated by the fiber is 160 nε. The beat signals of three different frequencies are multiplexed by amplitude intensity evaluation for the two reference points of any phase reconstruction interval. The suppressed results of the interference fading effect are evaluated using 10% of the normalized intensity as the threshold. Figure 8 shows that the probability of the interference fading effect is effectively reduced from 17.541% to 1.123%. In addition, π phase shift is introduced into the spectrum of beat signals of 40, 90, and 150 MHz, respectively. The generating time-domain signal intensity distribution is inconsistent with those of the original signals, as shown in Fig. 10. Figure 11 shows that the probability of interference fading effect is further reduced to 0.045% after all six sets of beat signals are multiplexed.

In conclusion, we propose a novel multi-frequency pulse modulation Ф-OTDR scheme based on broadband acoustic-optic modulation. The proof-of-principle experimental results indicate that the probability of coherent fading can be reduced from 17.541% to 0.045%. Furthermore, this method offers the advantages of a simple and compact pulse modulation structure, precise phase delay control, and a flexible and controllable pulse frequency component without sacrificing the response bandwidth and spatial resolution in Ф-OTDR. We believe that our work provides a practical way toward distributed fiber optic acoustic and vibration sensing, such as seismic wave monitoring, perimeter intrusion monitoring, and pipeline leak monitoring.

Optical fiber distributed acoustic sensing (DAS) system is a type of acoustic sensor built upon optical fibers and optoelectronic technology. It primarily utilizes optical coherence detection technology to convert acoustic vibration signals into optical signals, which are then transmitted through optical fibers to a signal processing system, thereby obtaining valuable acoustic wave data. They possess advantages such as high sensitivity, wide dynamic range, resistance to electromagnetic interference, structural flexibility, scalability into large-scale arrays, and suitability for extremely harsh conditions. Consequently, since their advent, driven by significant military and civilian applications, they have rapidly developed into an important direction in modern optical sensing and acoustic sensing technology. In recent years, DAS has been widely applied in fields such as oil and gas pipeline monitoring, national defense security surveillance, power cable monitoring, and structural health monitoring of large-scale infrastructure. With the expanding applications of DAS, there is a growing need for acoustic measurement solutions with higher sensitivity and reduced noise floors. To meet these demands, researchers have employed techniques to reduce system phase noise and suppress the fading phenomenon in the context of coherent optical time-domain reflectometry (COTDR) systems. The traditional COTDR systems struggle to achieve ideal signal demodulation due to the presence of signal fading phenomena. We propose a phase demodulation technique based on the multi-frequency optimized diversity (MFOD) algorithm.

This experiment employed a multi-frequency coherent optical time-domain reflectometry (MF-COTDR) system based on the frequency-shifted loop structure. Firstly, a detailed description of the phase demodulation method in the traditional COTDR systems was provided. This was followed by the utilization of multiple matching bandpass filters to separate detection pulses of different frequencies, achieving phase demodulation in the MF-COTDR system. Subsequently, we designed an MFOD algorithm based on the MF-COTDR system and established a coherent detection system for MF-COTDR, incorporating a frequency-shifted loop structure and a polarization diversity receiver. This reduced the influence caused by signal fading due to polarization fading and phase fading. Then, we verified the influence of the MFOD algorithm on the signal demodulation performance of the MF-COTDR system and tested the frequency response range of the MF-COTDR system.

Since the MFOD algorithm filters out phase signals in frequency bands unfavorable for gain aggregation, the phase variation information exhibits significantly improved SNR, rendering the phase fluctuations of the fiber under test (FUT) extremely weak in the absence of external disturbances (Fig. 4). Subsequently, a comparative analysis is conducted on the phase demodulation signals of single detection frequency COTDR, MF-COTDR, and MF-COTDR based on MFOD. The results indicate that the adoption of the MFOD algorithm not only suppresses signal fading but also reduces the noise floor of the COTDR system (Fig. 5). Further testing of the frequency response range of the MF-COTDR system demodulated by the MFOD algorithm shows that the system can achieve excellent broadband frequency response (Fig. 6).

We propose an optimized phase demodulation algorithm for the MF-COTDR system based on the frequency-shifted loop structure. This algorithm can selectively and automatically identify favorable phase demodulation signals from multiple detection frequencies with statistically independent coherent rayleigh noise patterns. The adoption of the MFOD algorithm enables the creation of a distributed sensing system with a uniform noise floor across all sensing channels. Through experiments, the impact of the MFOD algorithm on the overall noise floor of phase demodulation signals of the COTDR system is evaluated. The results show that when combined with the MFOD algorithm, the overall noise floor of the MF-COTDR system is significantly improved; the average signal-to-noise (SNR) is increased by about 9 dB, and the minimum strain resolution reaches 33.3 pε/Hz^1/2. In addition, we study the response of the MF-COTDR system based on the MFOD algorithm to vibration signals in different frequency ranges. The results show that the MF-COTDR system has larger response bandwidth and better linear response. The research has a good reference value for improving the phase signal demodulation performance of the COTDR system and promoting its application in practical engineering.

Acoustics detection is significant for national marine military defense, resource exploration, and disaster monitoring. Hydrophone technology is a key means of underwater acoustic detection and has been extensively researched. Compared to traditional electronic hydrophones, fiber optic hydrophone (FOH) has advantages such as resistance to electromagnetic interference, small size, light weight, and high detection sensitivity, making them a research hotspot for the new generation of hydrophones. Compared to traditional point-type FOH arrays, FOH based on fiber-optic distributed acoustic sensing technology (DAS), has advantages such as large capacity, high consistency, miniaturization, and low cost. It has gradually become a research hotspot in recent years. In 2017, due to its unique advantages, the DAS technology was first introduced into the field of underwater acoustic detection. This new type of FOH was continuously wound with one fiber, without any fusion splice and other optical components, which greatly reduces the complexity of the system and the manufacturing process and shows the potential application as the distributed FOH array. In 2022, a short tested distributed fiber-optic hydrophone towing array was proposed and demonstrated in a sea trial. We elaborate the comprehensive study on the development of a large-scale distributed fiber-optic hydrophone towing array with 192 independent sensing units, including the optimization of array-structure design, the integration of attitude perception system, the large-scale manufacture technology, the demodulation methods, and the calibration testing. The lake trial test with the performance evaluation on the array towing attitude, the noise test, the spatial gain, and the localization of the artificial target is presented as well.

The distributed FOH array is composed of an acoustic sensing cable sandwiched between front and rear vibration-isolation cables. Each vibration isolation cable is 25 m long, with different designed density ratios. The acoustic sensing cable consists of two 50 m long sections, including the high acoustic pressure sensitivity sensing units, the vibration damping modules, the attitude perception modules, the signal transmission fibers, and the Kevlar tension ropes, with a neutral buoyancy in water. The acoustic sensing unit is specially designed with a fiber evenly wrapped on the composite material structure to enhance acoustic pressure sensitivity. The 50 m long sensing cable includes 96 sensing units, which are wound by a single fiber. Based on the automation of the fiber length control, the constant tension maintenance, and the winding curing process, we achieve the highly efficient and consistent manufacture of the large-scale sensing unit. The attitude perception module has an oil pressure hole to sense the hydraulic pressure and evaluate the depth of the array with an error of less than 2 cm. Besides, the local incline angle can be acquired by the attitude perception chip embedded in the module. With the home-developed signal acquisition terminal and display software, the attitude angle and the depth of the cable can be obtained in real time.

The lake trial test of the developed large-scale distributed fiber-optic hydrophone towing array is carried out in Dongjiang Lake in Hunan Province. When the towing speed is 6 kn, the depth at the front of the array is 14.7 m, while the depth at the tail is 31.3 m. The inclination angle of the acoustic section of the array is about 7.8°, maintaining a good level (Fig. 7). The measured results of the array depth match well with the simulation results. The variation of towing noise with the towing speed shows that the hydrophone towing array has good noise suppression capabilities. The phase time domain spectrum of the 600 Hz signals from 12 channels of the array, as well as the PSD results of the 400 Hz, 500 Hz, and 600 Hz signals are presented in Fig. 9. Although different channels suffer different levels of noise, the response to the line spectrum signal exhibits excellent consistency. Traditional beamforming (CBF) signal processing is performed and the spatial gain of the whole array can be achieved as 16.87 dB (Fig. 10). The arrival angle estimation (DOA) is also conducted at 400 Hz. The relatively bearing time trajectory of the hydroacoustic target is obtained by continuously recording the DOA data (Fig. 11), which indicates that the distributed FOH towing array can achieve target trajectory tracking.

We elaborate on the design, manufacture, and testing process of a large-scale distributed fiber-optic hydrophone towing array, introduce the uDAS signal demodulation system, and analyze the lake trial. The conclusions are listed below:

1) The large-scale distributed fiber-optic hydrophone towing array has an average acoustic pressure sensitivity of -127.44 dB (re rad/uPa) with a standard deviation of 1.2 dB, within the frequency range of 20-1000 Hz.

2) The large-scale distributed fiber-optic hydrophone towing array has real-time attitude perception capability and can maintain good array posture underwater.

3) At a towing speed of 6 kn, the large-scale distributed fiber-optic hydrophone towing array shows good towing noise suppression capability. By array signal processing, the spatial gain of the hydrophone array reaches nearly 17 dB and the DOA estimation is achieved. It can achieve underwater acoustic target location and trajectory tracking.

Distributed optic fiber sensor (DOFS) is widely used for status monitoring and damage detection of aerospace vehicles due to its ability to achieve large-area and high-density sensing of structures. However, in the face of uncertainties caused by the harsh service environment of aerospace, a phenomenon of strain reading anomalies (SRAs) occurs in DOFS measurements. These SRAs result in significant strain peaks occurring in localized regions or at specific moments in time, thereby posing challenges for DOFS to accurately measure physical quantities and making it even more difficult to interpret these measurements. To minimize the negative effects of SRAs, some researchers have adopted a series of data processing methods, such as polynomial fitting method, spectral shift quality (SSQ) method, geometrical threshold method (GTM), and polynomial interpolation comparison method (PICM). Although these data processing methods are effective in reducing random errors in measurement data, they fail to completely remove the phenomenon of SRAs, and there is still a risk of removing highly reliable measurement readings. Meanwhile, the above methods still use the fixed threshold method to detect and determine the anomalies, and the determination of the fixed threshold relies on manual experience, which has low detection efficiency and a high false alarm rate, thus limiting its application in complex service environments. Therefore, we propose an intelligent adaptive post-processing method for detecting and quickly removing SRAs from DOFS.

The proposed algorithm, namely the adaptive geometrical threshold offset method (AGTOM), adopts the K-means clustering method to adaptively determine thresholds for distinguishing differences of thresholds caused by various structural features and service conditions. A continuous geometric correction is implemented on the distorted strain curves to effectively eliminate SRAs. To verify the effectiveness of the proposed method, a case study is conducted on the processing of DOFS measurement data collected during the pressure cycling test of a fuel tank. The Pearson correlation coefficient (PCC) is utilized to evaluate the correlation between the post-processing curves and normal strain curves. Besides, a comparison is conducted with other post-processing algorithms (GTM and PICM) to highlight the advantages of the proposed method.

Based on their different response characteristics, SRAs can be classified into two categories: harmless strain reading anomalies (HL-SRAs) and harmful strain reading anomalies (HF-SRAs). For the HL-SRAs, AGTOM consistently yields optimal post-processing results with PCC values not less than 0.965. It is followed by GTM, whose PCC values are all not less than 0.798. However, GTM interferes when HL-SRAs are coupled with NaN values. In addition, PICM achieves promising processing results only in the first typical case (i.e., sparsely distributed HL-SRAs). In the remaining three typical cases, PICM still produces distortions with a PCC value not greater than 0.512. Importantly, both GTM and PICM exhibit distorted post-processing curves when HL-SRAs are coupled with NaN values. For HF-SRAs, AGTOM also yields the highest post-processing results, with no PCC value lower than 0.917. The susceptibility of PICM to curve distortion accurately reflects the difference between HF-SRAs and HL-SRAs because the main characteristic of the former is that strain values follow an erroneous strain response or frequent sudden changes. It is difficult to determine the change in strain increment using PICM because it detects and removes SRAs by comparing the fitted value with the original value. Compared with PICM, GTM takes into account the sudden changes of the strain increment, resulting in improved post-processing results when HF-SRAs consist of densely changed SRAs. However, similar to HL-SRAs, both GTM and PICM show worsened post-processing results when NaN values interfere with HF-SRAs, indicating lower algorithmic robustness for GTM and PICM compared to AGTOM

The proposed algorithm AGTOM is able to distinguish the differences in thresholds due to different structural characteristics and service environments. The K-mean clustering algorithm uses an internal evaluation metric, namely Davies-Bouldin index (DBI), to characterize the clustering effect of strain increments. The threshold is determined by obtaining the optimal k value. For the HL-SRAs, both GTM and AGTOM methods can achieve satisfactory processing results. However, PICM is susceptible to interference when facing densely distributed and coupled HL-SRAs, leading to serious distortions in its post-processing curves. For HF-SRA, the post-processing curves of the other two algorithms are distorted to varying degrees, except for AGTOM, which exhibits the highest PCC compared to the normal strain curve. For both HL-SRAs and HF-SRAs, GTM and PICM are interfered with when SRAs are coupled with NaN, indicating that the algorithmic robustness of both is lower than that of AGTOM. To further validate the effectiveness of AGTOM, it will still be necessary to test AGTOM by applying it to different experimental scenarios in the future.

We investigate intelligent processing and imaging methods for distributed acoustic sensing vertical seismic profile (DAS-VSP) data, focusing on longitudinal waves and converted waves. Meanwhile, we discuss DAS-VSP morphology component analysis for noise reduction, intelligent separation of multiple waves in DAS-VSP data, and regularization methods using deep learning for DAS-VSP data, and study a multi-wave VSP imaging method based on minimum travel time. The proposed method combines the advantages of VSP-CDP conversion and conventional ray-based Kirchhoff migration and utilizes minimum travel time information to determine the reflection wave paths in the VSP data. By controlling the focusing imaging near the reflection paths using travel time tables, this method reduces the curvature compared to traditional seismic migration methods and calculates the coverage during the imaging to resolve uneven imaging amplitudes. By the actual data processing of offshore inclined well DAS-VSP, the DAS-VSP P-P wave and P-S wave imaging profiles are obtained simultaneously for the first time in China. Combining targeted processing, the researchers achieve imaging of both P-P and P-S waves from DAS-VSP data. The results indicate that the DAS-VSP from deviated wells provides conditions for multi-wave imaging and yields higher signal-to-noise ratio (SNR) imaging data for P-P waves and P-S waves. Multi-wave data is more conducive to oil and gas prediction and identification, and artificial intelligence (AI) processing and multi-wave imaging methods provide new technical means for DAS-VSP in oil and gas exploration and development.

The process of the proposed VSP imaging method based on minimum travel time is demonstrated, and multi template fast advancement algorithm is employed to calculate the travel time table for each shot and receiver pair. Further, the two travel time tables are summed and sorted from small to large ones by depth. Given the number of grids (migration aperture), the wave field data are projected at the corresponding position according to the travel time and stacked. Meanwhile, we repeat the projection for each location, followed by calculating the coverage folds to average the seismic amplitude anomaly due to uneven coverage. Finally, we stack them all to form an imaging profile. This process is applicable for both the P-wave and converted wave imaging. This method combines the advantages of common depth point conversion and migration and focuses imaging near the reflection path, thus reducing migration artifacts, calculating the number of coverage times during the imaging, and addressing abnormal imaging amplitudes due to uneven coverage.

Our data are located in the Pinghu Oil and Gas Field in the East China Sea, which is excited by air gun source and received by DAS. The converted wave imaging process extracts a shot line in the well trajectory direction for testing (Fig. 1). The maximum offset is 4190 m, the shot point distance is 50 m, and the total number of shots is 148, with the measured optical cable depth of 3357 m, and maximum offset of -1533 m, and DAS receiver channel distance of 2 m. Figure 1 shows the upgoing converted wave ray and polarization direction. It indicates that the polarization of the upgoing converted wave in the well trajectory direction is perpendicular to the optical cable, the well trajectory is in the opposite direction, and the upgoing converted wave is parallel to the optical cable.

We study a VSP imaging method based on grid ray tracing, which has the following advantages:

Distributed acoustic sensing (DAS) technology regards the same optical fiber as both a sensing and transmission medium, enabling real-time capture of vibrations and acoustic signals in the vicinity of optical fibers. This technology features robust environmental adaptability, high sensitivity, and exceptional resistance to electromagnetic interference. DAS can transform optical cables spanning tens of kilometers into evenly distributed arrays for vibration and acoustic sensing, thus achieving spatial sampling resolutions of less than 1 m. Consequently, it has caught significant attention in fields such as oil and gas exploration and marine geophysics. However, the maximum sensing range of DAS is constrained by optical source bandwidth and stability, with the current maximum sensing range limited to approximately 100 km. Therefore, employing DAS for intercontinental undersea long-distance communication cables poses significant challenges. We introduce a DAS seabed in-situ monitoring system integrated with an underwater experiment platform, coupled with relevant offshore trials. This deployment extends DAS capabilities to more remote and deeper-sea areas. The analysis of offshore trial results validates the system feasibility. In the future, consideration can be given to combining submersible technologies for laying sensing optical cables on the seabed, which holds substantial significance for advancing marine geophysical research.

We employ pressure chamber technology to construct a deep-sea DAS system capable of withstanding pressure at depths of up to 10000 m and conduct rigorous pressure testing in the laboratory, subjecting the system to pressure of up to 110 MPa. Subsequently, the system is deployed on a deep-sea in-situ experimental platform equipped with autonomous underwater mobility capabilities and is placed at a depth of 1423 m for in-situ testing. Throughout the sea trial, the base station undergoes various operational states including submersion, landing, moving, and raising, with distinct vibration data generated by each of these states. By analyzing the data collected by the DAS, we can discern the base station's temporal and spectral characteristics in different operational conditions. These research findings confirm the feasibility of the design and deployment of deep-sea DAS systems in the extreme deep-sea environment.

When the base station is stationary, DAS can capture the background noise of the sea, with an average noise level of 4.64×10^-4 rad/Hz, which closely aligns with laboratory measurements. The operational states of the base station can be primarily categorized as follows. ① When the base station submerges, there is a significant energy transfer upon contact with the seafloor, resulting in strong vibrations. Subsequently, the remaining modules of the base station enter the water one by one to generate a series of relatively weaker vibrations. ② When the base station lands on the seafloor, by comparing two consecutive impact signals, the energy of the second signal is observed to attenuate significantly compared to the first one, which indicates successful soft landing of the base station. ③ Before the base station is raised, the payload should be released. DAS records the produced vibration signals when the base station releases its payload. It is noticed that when the base station is in weight balance, the payload release primarily generates high-frequency signals above 30 Hz, but when it is not in balance, the signal energy is mainly distributed below 20 Hz. ④ When the base station moves underwater using servo propellers for direction control, compared to the stationary state, the rotation of the propellers and the movement of the base station introduce stronger interference below 60 Hz, and these interference signals' harmonics are also present in the high-frequency range.

DAS technology is a novel seismic monitoring technique emerging in recent years, and features high spatial density, long-range capabilities, and dynamic measurements. Meanwhile, it offers sensing distances of up to several tens of kilometers with spatial resolution of a few meters. This technology is known for its simplicity, low development and maintenance costs, and ability to provide real-time data transmission, and it has the potential to significantly reduce observation costs, thus becoming a promising choice for deployment in critical marine regions. We present the deep-sea validation testing of the deep-sea DAS systems. During a 21-day in-situ sea trial at a depth of 1423 m, over 600 GB experimental data are recorded. Analysis of the vibration events generated during the base station's submersion, settling on the seabed, payload release, and movement processes confirms its capability for deep-sea operations and vibration signal detection. Additionally, it validates the operational reliability and stability in the deep-sea environment, laying a strong foundation for future trials at depths of up to 10000 m.

As a novel distributed sensing system, the distributed optical fiber vibration system (DOFVS) has been widely applied in recent years due to its advantages of real time, high accuracy, and strong robustness. DOFVS has many application fields, such as structural health monitoring, pipeline leak detection, and perimeter security. In recent years, DOFVS performances such as spatial resolution, monitoring distance, and accuracy have been improved with the demodulation algorithm development and system structure optimization. Meanwhile, with the development of technologies such as deep learning and artificial intelligence, DOFVS also gradually becomes intelligent. To achieve accurate automatic pattern recognition of vibration signals, we combine the DOFVS with an unmanned aerial vehicle (UAV) video monitoring system. The proposed system employs convolutional neural networks to realize pattern recognition in optical signals and video signals simultaneously. Our scheme increases the number of recognizable sensing events and improves recognition accuracy, expanding the intelligent application scenarios of DOFVS.

We propose a multi-dimensional sensing event recognition scheme based on convolutional neural networks, combining the DMZI-based DOFVS and a UAV video monitoring system. The proposed scheme adopts Resnet 50 as the feature extraction backbone network to extract features of the optical signals and video signals. The optical signals are transformed from 1D time-domain signals to 2D time-frequency signals by short-time Fourier transform. The 2D time-frequency images are then segmented based on power distribution to reduce image noise, and the images are fed into a 2D Resnet 50 network to obtain the confidence of the recognized sensing events. The 3D video signals are fed into a SlowFast model with a 3D Resnet 50 as the feature extraction network to obtain the confidence of the recognized sensing events for video signals. Finally, the confidence vectors obtained from both optical and video signals are multiplied and normalized, and the event with the highest confidence is output as the final judgment event. To verify the feasibility of the proposed method, we conduct experiments to recognize nine types of sensing events, and the average recognition accuracy and system response time of the proposed scheme are obtained.

The proposed scheme overcomes the limitation of recognizing multiple events when only recognizing optical signals. The employed dataset consists of two parts: one is the 2D time-frequency images corresponding to optical signals with 1800 images for each sensing event, and the other is video data obtained from UAV with 140 segments of 20 s videos for each intrusion event (Table 3). Both parts are divided into training, validation, and testing sets in an 8∶1∶1 ratio. To validate the feasibility and effectiveness of the proposed solution, we compare the results of recognizing optical signals alone, results of video signals alone, and the fused recognition results (Table 4). Optical signals achieve high recognition accuracy on events with more obvious time-frequency features, such as climbing, cutting, and pulling. However, the events with similar features have low accuracy, such as crashing, kicking, and waggling. Similarly, the accuracy of UAV video signals for events such as climbing, knocking hard, and pulling is low. When optical signal recognition and video signal recognition are applied separately, neither of them achieves sound pattern recognition results. After confidence fusion, the proposed method achieves 99.58% recognition accuracy for nine sensing events in the testing set. Moreover, the recognition of optical signals and video signals can be performed simultaneously, and the system response time can meet the real-time detection needs.

We propose a multi-dimensional DOFVS pattern recognition scheme based on convolutional neural networks (CNNs), which combines two models including a 2D time-frequency signal recognition model based on the Resnet 50 and a 3D video signal recognition model based on the SlowFast model. This scheme not only expands the features of the optical signal by time-frequency transformation but also automatically extracts and classifies features using CNNs. The impact of low robustness of manual feature extraction schemes can be reduced. Meanwhile, the 3D video signal recognition is combined with optical signal recognition to enable the detection of nine types of events including climbing, crashing, cutting, kicking, knocking hard, knocking lightly, pulling, waggling, and no intrusion. The effectiveness of the proposed scheme is verified via experiments, which demonstrate that the average accuracy of the nine events is 99.58% and the recognition time is 0.16 s to achieve real-time synchronous response to event changes. Compared with traditional single optical signal recognition, the proposed scheme greatly expands the event types that can be recognized in the DOFVS field. Therefore, this scheme will further improve the DOFVS stability and reliability in practical engineering applications in the future.

The signal-to-noise ratio (SNR) is a crucial performance metric in Brillouin distributed optical fiber sensors. Ensuring accurate noise characterization is essential for effective targeted denoising. However, collecting real noise data poses practical challenges. Gaussian noise, traditionally used in supervised methods, is somewhat effective but lacks accuracy. In this paper, we propose to utilize a self-consistent generative adversarial network (SCGAN) to model real noise distribution using collected Brillouin gain spectrum (BGS) data. This enables us to generate noise data for training denoising convolutional neural networks (CNNs). By training the SCGAN to replicate real noise intricacies, we can effectively train a CNN to discern between signal and noise, resulting in more precise noise reduction. By addressing the limitations of conventional Gaussian noise models, our method bridges the gap between artificial noise simulations and complex real-world BOTDA system noise patterns. This innovative approach has the potential to significantly enhance noise reduction techniques for BOTDA systems, improving accuracy and efficiency.

While generative adversarial networks (GANs) have showcased their effectiveness in modeling intricate noise distributions from extensive datasets, they harbor a notable training limitation. GANs optimize their generator networks by minimizing dissimilarities between generated and real samples. Unfortunately, this process might inadvertently prioritize prevalent training data patterns, sidelining other potential variations. To transcend this limitation, this paper introduces a SCGAN as a solution for noise modeling. Going beyond conventional GANs, SCGAN introduces a novel approach. It supplements the adversarial loss with three additional loss functions, effectively offering more guidance and constraints during network training. This augmentation facilitates a more holistic approach to noise modeling by steering the network towards a broader representation of noise patterns. To substantiate the differentiation between Gaussian noise and SCGAN-generated noise, we employ histogram statistics and amplitude spectrum analysis. Subsequently, both types of noise are harnessed to train three state-of-the-art denoising CNNs. The performances of networks are then compared across experimental BGS encompassing varying temperatures and SNRs. This approach reflects a holistic exploration, encompassing both noise modeling and denoising neural network evaluation.

To enable a thorough comparative analysis between SCGAN-generated noise and Gaussian noise, we employ histogram statistics and the Kolmogorov-Smirnov test for both noise sources. Furthermore, a two-dimensional Fourier transform is executed to acquire the noise amplitude spectrum, with the findings visualized in Figs. 10 and 11. These analyses distinctly display the divergences between Gaussian noise and real noise. To effectively showcase the enhanced SNR brought forth by our method, we assess denoising neural networks trained with distinct noise sources across various temperature settings and averaging times. The outcomes are tabulated in Table 1 and Table 2. Importantly, networks trained using SCGAN-generated noise consistently exhibit elevated SNR values compared with their Gaussian noise-trained counterparts. Following the acquisition of temperature data, we compute the corresponding root mean square error (RMSE) and standard deviation (SD). Figures 7 and 8 provide the comprehensive outcomes achieved by different neural networks trained with varying noise sources under diverse temperature conditions and SNRs. Remarkably, networks trained with SCGAN-generated noise consistently outperform their counterparts, delivering superior denoising outcomes characterized by precision and stability. These results underscore the efficacy of SCGAN-based noise training in achieving remarkable noise reduction, generating highly accurate and dependable measurement outcomes across a spectrum of temperature conditions and averaging times.

We introduce the utilization of SCGAN for modeling real noise data and generating paired noise data tailored for supervised training. The research entails a comparative study involving three supervised denoising neural networks—DnCNN, ADNet, and BRDNet—trained with both Gaussian and SCGAN-generated noise. The outcomes distinctly illustrate the method's efficacy in noise reduction for Brillouin distributed optical fiber sensor data, while preserving intricate details. Notably, networks trained on SCGAN-generated noise exhibit superior proficiency in identifying noise features, leading to enhanced measurement outcomes. This advantage remains consistent even under conditions of low averaging times, suggesting the potential for heightened data acquisition rates. Importantly, this paper pioneers the application of generative adversarial models in the domain of Brillouin distributed optical fiber sensor denoising, presenting a novel frontier. Leveraging the diverse arsenal of generative adversarial data generation methods, the technique introduced here has the potential for broader adoption in the realm of distributed optical fiber sensing. This pioneering approach sets the stage for substantial advancements in the accuracy and efficiency of noise reduction methods, ultimately contributing significantly to practical sensor data acquisition rates.

Raman distributed optical fiber sensing technology has scientific significance across various fields due to its unique ability to perform distributed measurements of spatial ambient temperature fields. However, the spatial resolution of existing Raman distributed optical fiber sensing with a detection range extending to kilometers is constrained to the meter level due to the physical limitations of the optical time domain reflection positioning principle, which relies on the pulse time-of-flight method. Therefore, we introduce an innovative Raman distributed optical fiber sensing approach based on a multi-order time-domain differential reconstruction correlation method. In this novel method, we replace the conventional pulse laser with chaotic, amplified spontaneous emission (ASE), and noise signals as the sensing sources, and then employ a multi-order time-domain differential reconstruction technique to reconstruct the Raman anti-Stokes scattering signal. This reconstruction process enables us to extract intensity information from each sensing fiber position point and thus capture the random fluctuation characteristics of detection signal timings. To further optimize the proposed approach, we employ a correlation compression demodulation method to unveil the correlation between the spatial position of the Raman scattering temperature-modulated light and the detection signal. Notably, our scheme diverges from the traditional optical time-domain reflection positioning principle, opting instead for the correlation positioning principle. This shift allows us to overcome the physical constraints associated with the pulse width of conventional light sources, thereby elevating the spatial resolution of the sensing technology from the traditional meter-scale performance to the centimeter level.

The simulation model adopts intricate pulse signals as the detection signals, including chaotic pulse signals, noise pulse signals, and ASE pulse signals. This model bifurcates into two beams using a haloscope, with one beam serving as the reference signal and the other functioning as the detection signal, and subsequently they enter the sensing fiber via a wavelength division multiplexer (WDM). During experimentation, a small segment of the sensing fiber experiences ambient temperature changes, designated as the detection fiber (FUT), while the remaining portion is still in a constant temperature environment. A WDM filters out Raman anti-Stokes scattering signals at 1450 nm wavelength. Subsequently, the Raman anti-Stokes scattered signal undergoes multi-order time-domain differential reconstruction followed by correlation with the reference signal, which facilitates the demodulation of temperature change information within the detection fiber. In the initial multi-order time-domain differential reconstruction, the Raman anti-Stokes scattered signal excited within the sensing fiber is subjected to the process. Based on the random amplitude characteristics of the complex signal timing, this procedure allows for the isolation of the Raman anti-Stokes signal from each data point location on the sensing fiber. Consequently, each data point of the reconstructed signal exclusively contains the scattered signal intensity information of an individual location point, as opposed to encompassing the light intensity information of all signals within the length corresponding to the pulse width in the OTDR positioning principle under the traditional scheme. Finally, the system employs relevant compression demodulation technologies, enabling the compression of scattering intensity information from all data points within FUT to the FUT's start and end positions. The precise compression facilitates the accurate determination of detailed FUT positions.

The chaotic signal exhibits a higher incoming fiber optical power and more significant random amplitude fluctuations, facilitating the extraction of Raman anti-Stokes light intensity at various fiber points by multi-order time-domain differential reconstruction. This contributes to an enhanced signal-to-noise ratio (SNR) within the Raman distributed sensing system, grounded in a complex signal correlation method (Fig. 2). All three signal sensing schemes adeptly pinpoint the FUT's position and length information. Notably, at temperature mutation points within the FUT region, we observe a pair of positively and negatively correlated peaks. Remarkably, the SNR of the chaotic signal-based sensing scheme surpasses that of both the ASE signal sensing scheme and noise signal sensing scheme (Fig. 4). The spatial resolution of the Raman distributed optical fiber sensing system is predicated on complex signal correlation hinges on the half-height full width (FWHM) of the autocorrelation function of the sensing detection signal. Our analysis of the autocorrelation results and spectral characteristics of three complex signal timings reveals that the employed FWHM of the three complex signal autocorrelation functions is 0.01 ns. Consequently, the theoretical spatial resolution is 1 mm (Fig. 5). We also observe a positive correlation between the differential order and the peak-to-peak value of positive correlation, which signifies a substantial improvement in the system's SNR within a certain range of increasing order. The optimal SNR for the chaotic sensing scheme occurs when the differential reconstruction order is set to the 5, surpassing the peak coefficient of 5.04 dB observed in the first-order case (Fig. 6). Furthermore, we find that the FUT's positively correlated peak-to-peak values demonstrate linear correlation with temperature (0.323 for every 1 K temperature increase in the above-mentioned simulation conditions). The correlation highlights the utility of positively correlated peaks for precise temperature change demodulation. We successfully achieve accurate demodulation of spatial localization and temperature for FUTs with 353 K and 373 K, featuring two 0.05 m lengths and 0.05 m intervals (Fig. 7).

We present a novel Raman distributed fiber sensing technology, which utilizes the multi-order time-domain differential reconstruction correlation method to enhance the spatial resolution and SNR performance in traditional Raman distributed optical fiber sensing systems. Furthermore, the correlation compression demodulation method is adopted to elucidate the spatial distribution of the Raman scattering temperature-modulated light field. Significantly, our approach supersedes the conventional optical time-domain reflection localization principle by adopting the correlation positioning principle, thus overcoming the spatial resolution constraints imposed by the pulse width of traditional light sources. As a result, the traditional meter-level sensing spatial resolution is theoretically elevated to the centimeter level. Meanwhile, we extend the theoretical framework of the differential reconstruction scheme to encompass any order and scrutinize the influence of differential order on the sensing system's SNR. Numerical simulations demonstrate that our method can extract more temperature information, encompassing the detection fiber region without compromising the sensing spatial resolution. Furthermore, it amplifies the SNR of the sensing signal by multiplying the scattered signal correlation peak's amplification effect. Additionally, it reduces the temperature signal crosstalk in the non-detection optical fiber region during the demodulation, enhancing the SNR of the Raman distributed optical fiber sensing system. Finally, we demonstrate the significant advantages of chaos signals in Raman sensing and introduce a fresh research perspective for Raman distributed optical fiber sensing technology.

We aim to address limited data acquisition in fiber optic sensing technology, especially in phase-sensitive optical time-domain reflectometry. A data augmentation method based on conditional generative adversarial networks (GANs) is proposed to generate a large number of training samples and improve the detection capability and performance of the classifier model.

The experimental data collection is conducted using a phase-sensitive optical time-domain reflectometer (Φ-OTDR). First, the collected real data are adopted as input to the conditional GAN. The GAN model automatically extracts signal features and generates realistic signal data with the assistance of input conditions, with the specific experimental flow shown in Fig. 7. Second, the generated data and original data are separately fed into classifiers such as decision trees, support vector machines, and convolutional neural networks for classification. By comparing the detection results of the generated and raw data across different classifiers, the effectiveness of the data augmentation method is evaluated, and the specific comparison results are shown in Fig. 12. This comprehensive approach can assess the influence of the generated data on the classifier performance to address limited data acquisition in fiber optic sensing technology.

The experimental results demonstrate that the detection results of the generated data significantly improve across decision trees, support vector machines, and convolutional neural networks. The generated data enhance the detection capability and performance of the classifier models, achieving the target identification in Φ-OTDR. Furthermore, improvements in the conditional GAN can generate more realistic signal data, further enhancing the model performance.

We successfully address the data acquisition limitations in Φ-OTDR by a data augmentation method based on conditional GAN. The generated data improve the detection capability and performance of the classifier models. The research findings provide new insights and methods for small-sample detection, and also valuable references for the applications of other fiber optic sensing technologies.

In recent years, with the rapid development of internet businesses and the emergence of emerging industries, large-capacity and high-speed optical access network systems have been deployed and extensively studied. Passive optical network (PON) technology has become a cost-effective optical access network solution due to its natural advantages of low power consumption and flexible bandwidth access. However, because of the passive nature of PON systems, in the event of fiber faults, the fault points cannot actively report fault information to the central office, making real-time monitoring and fault location maintenance work difficult. Additionally, there is mutual interference between users in the centralized tree-shaped network architecture of PON systems, which makes the point-to-point fault monitoring scheme of traditional optical network systems unable to accurately locate the specific link and fault. Therefore, studying effective PON system monitoring plans has also become one of the focuses of future PON system research. PON system fault detection methods, such as improved schemes using wavelength-tunable OTDR, Brillouin OTDR, and optical feedback with chaotic lasers, require adding difficult-to-install and/or expensive equipment to the system, which increases the system cost to some extent. Alternatively, although existing optical encoding schemes have good scalability and low upgrade cost, they have not considered the problem of system performance degradation caused by interference between different users. In this article, a PON monitoring system based on multi-wavelength grouping is proposed. By assigning users prone to interference to different groups and deploying corresponding reflecting Bragg gratings at the end of the link, different wavelengths are used to detect and distinguish between user groups, thereby reducing the probability of interference. We hope that our scheme and model can provide useful assistance for the design of PON fault monitoring systems.

A user distribution model for PON is developed to calculate the interference probability function of users, and an optimization approach for multi-wavelength packet scheduling is mathematically formulated. The simulation results of this model have verified the feasibility of multi-wavelength packet scheduling as a means of reducing inter-user interference. The system performance is simulated by MATLAB based on the received signal power and the function of various noise components, and appropriate transmission power and pulse width parameters are obtained for different numbers of users and distribution ranges. The feasibility of a fault monitoring system for PONs based on multi-wavelength packet scheduling is simulated by OptiSystem and verified.

Figure 3 shows the probability of interference occurrence under the multi-wavelength grouping scenario, where it can be effectively avoided when the number of equally spaced users is less than the wavelength number. The research results on the pulse width of the system show that the signal-to-noise ratio (SNR) has ideal values when the pulse width range is -10≤lgT_C ≤-8 (Fig. 4). When the pulse width T_C ≥10^-7 s, the signal-to-interference ratio (SIR) gradually approaches a lower boundary value (Fig. 6). The pulse width cannot be too narrow because reducing the pulse width requires a higher accuracy requirement for the system. Therefore, we suggest setting the pulse width to -9≤ lgT_C ≤-8 in a PON fault monitoring system based on multi-wavelength grouping. Figure 5 shows that the SNR improves linearly with the transmitted power. It is found that when the number of users increases, the SNR exhibits a decreasing trend (Fig. 5). This is because an increase in the number of users within the same interference range causes an increase in both beat noise and shot noise, leading to a decrease in SNR. Additionally, when the maximum user spacing l_e = 1 km and the number of users K increases from 64 to 128, the increase in interference causes SIR to decrease from 40.6 dB to 34.9 dB. This is because the decrease in user separation distance leads to an increase in interference, which in turn leads to a decrease in SIR. A PON fault monitoring system based on multi-wavelength grouping, with wavelengths of m=4 and users of K=16, is constructed in OptiSystem software, and at a transmitted power of P_s=4 dBm and pulse width T_C=10^-8 s, the identification accuracy for 3 dB pulse width corresponds to 2 m (Fig. 7).

A PON fault monitoring system based on multi-wavelength grouping is proposed. First, the interference caused by user distribution in PON systems is analyzed, and the probability of user distribution is derived. Subsequently, the optimal solution method for multi-wavelength grouping is mathematically modeled, and the theoretical impact of this grouping method on interference probability is obtained. It is proved that the multi-wavelength grouping method can effectively reduce the interference probability between users. In addition, the anti-interference performance of the PON fault monitoring system based on multi-wavelength grouping is investigated. The influence of different system parameters on SNR and SIR is simulated. For different user numbers and distribution ranges, selecting appropriate pulse widths and output powers can effectively improve system performance. It is found that when the pulse width is set to -9≤lgT_C ≤ -8, it can effectively suppress noise interference in the signal, thereby improving SNR, while also effectively reducing interference between users and ensuring that SIR remains within an acceptable range. Finally, by using the OptiSystem software, the system is simulated under conditions of wavelength number m=4, user number K=16, transmit power P_s =4 dBm, and pulse width T_C=10^-8 s. The simulation results show that the recognition accuracy corresponding to a pulse width of 3 dB is 2 m, and the results in the detection signal recognition module are consistent with the set parameters, verifying the effectiveness of the system. This work provides guidance for the design and parameter selection of PON fault monitoring systems.

The phase optical time domain reflector (Φ?OTDR) can quantitatively measure the external perturbation information by extracting the phase signals from the coherent Rayleigh curve. During solving the phase, the arctangent operation result is limited in [-π, π]. Phase unwrapping is inevitable to obtain a correct signal waveform. However, the traditional phase unwrapping algorithms require that the absolute difference between adjacent phase values does not exceed π. In Φ?OTDR, increasing the emission frequency of the optical pulse can enhance the sampling rate of the perturbation signals, thereby reducing the absolute difference. Nevertheless, the total optical fiber length is determined by the measured object. Once this length is determined, the emission frequency of the optical pulse is limited, which means that the sampling rate of the perturbation signal cannot be further increased. At this point, if the external perturbation signal changes fast, the phase signal cannot be correctly acquired. Therefore, the time-division dual-frequency light is introduced into coherent Φ?OTDR. However, the initial phases in the directions of both pulse time and fiber length are inconsistent, the introduction of time-division dual-frequency light cannot directly enhance the sampling rate of the perturbation signal, which influences the precise reconstruction of phase signals. Additionally, optical pulses with different frequencies amplify the number of fading positions, thereby heightening the challenge of choosing the reference position which is adopted to eliminate the inconsistency of the initial phase in the pulse time direction. Therefore, in coherent Φ?OTDR with the time-division dual-frequency light, a new method is needed to eliminate the inconsistency of the initial phase in the directions of both pulse time and fiber length and thus truly obtain sampling rate multiplication of the perturbation signal and precise reconstruction of phase signals.

The introduction of time-division dual-frequency light into coherent Φ?OTDR satisfies the requirement of sampling rate multiplication for uniform sampling on the sampling sequence of the pulse. However, the probe pulses of different frequencies complicate the distribution of coherent Rayleigh curves, phase curves, etc. Therefore, the true implementation of sampling rate multiplication requires more complex processing. To conveniently select the reference position of coherent Φ?OTDR with the time-division dual-frequency light, we calculate the distance of modulus value at each fiber sampling position for each frequency component, multiply and normalize the distance value at each fiber sampling position, and confirm the location of the perturbation signal based on the normalized curve. Then, we calculate the minimum value of the modulus at each fiber sampling position for each frequency component, and then multiply and normalize the minimum values of the modulus at each fiber sampling position. The reference position is just the fiber sampling position where the maximum value of the normalized curve is closest to the left side of the perturbation source. Correspondingly, the wrapped differential phase is obtained based on the reference position. Since two different frequency lights make the inconsistency of the initial phase more complex, it is best to eliminate the inconsistency of the initial phase caused by two different frequency lights spontaneously. Therefore, in coherent Φ?OTDR with the time-division dual-frequency light, for each frequency component, we select the pulse time when the perturbation is equal to a static event and the perturbation begins to change. Additionally, the wrapped differential phase at this pulse time is chosen as the reference phase. Then, the wrapped differential phase at each pulse time is subtracted by the reference phase of the same frequency pulse light. Meanwhile, the phase change after unwrapping at the pulse time of the reference phase is again adopted as the new reference phase to eliminate the noise and corresponding phase unwrapping error introduced by the reference phase. The unwrapped phase change at each time is subtracted by the new reference phase of the same frequency component.

Two different frequency lights with a pulse interval of 10 μs and an emission frequency of 50 kHz are introduced into coherent Φ?OTDR, and a Burst perturbation signal acts on the optical fiber. To obtain intermediate frequency signals of two frequency components, we filter the collected coherent Rayleigh scattering curves by 40 MHz and 80 MHz bandpass filters respectively. Based on the intermediate frequency signals, the modulus values of the two frequency components are calculated separately. Then, based on the product of the distance of modulus value [Fig. 8(b)], the left side of the perturbation position is accurately determined to be 1.036 km. Based on the product of the modulus minimum value, the reference position is quickly determined to be 1015.68 m [Fig. 8(b)]. For each frequency component, a wrapped differential phase at the pulse time closest to the perturbation change is selected as the reference phase, and then the wrapped differential phase of the same frequency component is subtracted from the reference phase. The difference values are cross-recombined into a new sequence. After unwrapping, it is just the phase change [Fig. 8(b)]. Furthermore, the phase change at the pulse time of the original reference phase is taken as the new reference phase, and the phase change of the same frequency component is subtracted from the new reference phase to obtain a new phase change. The new phase change exhibits a continuous linear profile along the fiber [Fig. 9(b)]. Finally, the precise Burst signal is extracted, and the maximum difference between adjacent phases is 4.5105 rad if the signal is down sampled. The fitting chi-square coefficient of the sinusoidal part is 0.9998, and the root mean square error is only 0.3872 rad.

In the phase optical time domain reflectometry, the traditional phase unwrapping algorithms require that the absolute difference between adjacent phases does not exceed π. It makes the sampling rate increase of perturbation signals crucial for precise reconstruction of the phase signals. However, the optical fiber length limits the increase in the sampling rate of the perturbation signals. Therefore, the time-division dual-frequency light is introduced into coherent Φ?OTDR. To eliminate the phase distortion caused by inconsistent initial phases in direct unwrapping, we perform the new method of choosing the reference position and dual static compensation. In the experiment, when the external perturbation is a Burst signal with a frequency of 700 Hz, the absolute difference between adjacent phases of the single-frequency probe pulse reaches 4.5105 rad. By the proposed method, the Burst signal is accurately retrieved, and the root mean square error of the sinusoidal part is only 0.3872 rad, which means that the sampling rate of perturbation signals is doubled and the phase signal is precisely reconstructed in coherent Φ-OTDR with the time-division dual-frequency light.

Due to the low scattering coefficient of Rayleigh backscatterings and fading noise, a distributed acoustic sensing (DAS) system suffers from detection noise of high level and non-uniform distribution along sensing fiber. Advanced fiber microstructure manipulation technology gives birth to a type of quasi-distributed acoustic sensing (qDAS). This technology employs a fiber array consisting of equally spaced weak reflection points such as weak fiber Bragg gratings (wFBG) or microstructures introduced by laser exposure. By the detection of interrogation optical pulses rather than Rayleigh backscatterings, the qDAS is free of fading noise. Additionally, the reflectivity of a weak reflection point is typically 3.16×10^-5 (-45 dB). Such reflectivity not only improves the signal-to-noise ratio of detected optical signals compared with that in a DAS system, but also reduces transmission loss and gets rid of crosstalk induced by multiply-reflected optical pulses due to low reflectivity. Therefore, this type of qDAS is promising in large-scale and high-fidelity sensing applications. Unfortunately, the interrogation optical pulses generate Rayleigh backscatterings during transmission in a fiber array. The Rayleigh backscatterings are collected together with the interrogation optical pulses. By taking an optical pulse of 100 ns in width as an example, the reflectivity of the induced Rayleigh backscatterings reaches 2×10^-6 and is comparable to that of a weak reflection point (3.16×10^-5). Since Rayleigh backscatterings carry sensing information, they introduce crosstalk between adjacent sensing channels in a qDAS system, which severely distorts the sensing signals and degrades the accuracy of sensing event positioning. Taking a wFBG-based qDAS system as an example, we study a method that suppresses the crosstalk induced by Rayleigh backscatterings. We hope that our study can help improve the sensing fidelity, sensing scale, and sensing positioning accuracy of a qDAS system, and expand the applications of qDAS.

The phase interrogation scheme is presented in Fig. 1. A dual-pulse direct detection scheme with a high heterodyne frequency is employed for phase interrogation in a wFBG-based qDAS system, and the polarization switch method is adopted to eliminate polarization fading noise. A phase modulator is applied after a laser source to suppress crosstalk induced by Rayleigh backscatterings. Driven by an electrical cosinusoidal signal, the phase modulator introduces a cosinusoidal phase modulation Ccos2πfmt to the coherent continuous light wave emitted from the laser source. Two optic probe pulses with equal pulse width W=100 ns, frequency difference Δν=20 MHz, and optical path difference n_eL_MZ are launched in a pair into a wFBG array, where n_e=1.5 and L_MZ=29.8 m are effective refractive index and fiber length difference of M-Z interferometer that generates two optic probes respectively. By carefully designing modulation frequency f_m to satisfy πfmτ=kπ,k=0,1,2,?, crosstalk can be suppressed, where k is integer, τ=neΔL/c, and ΔL=2L-LMZ. Additionally, larger magnitude C of phase modulation leads to better crosstalk suppression.

The validity of the proposed approach is experimentally verified. A fiber array consisting of seven wFBGs is under interrogation and a PZT is placed between the second and third wFBGs to simulate a cosinusoidal acoustic signal oscillating at 1 kHz. wFBGs are equally spaced by L=15.5 m to make ΔL= 1.2 m. An electrical cosinusoidal signal oscillating at f_m and a magnitude of 2 V is applied to the phase modulator to induce cosinusoidal phase modulation. By the design of f_m=166 MHz and ΔL=1.2 m, πfmτ=0.996π is realized. The demodulated signals along the fiber array are obtained, and the mean power spectrum density (PSD) over ten measurements at 1 kHz along the fiber array is presented (Fig. 7). Compared with the crosstalk of -21.82 dB without cosinusoidal phase modulation, the crosstalk in the case of cosinusoidal phase modulation is reduced to -35.22 dB. Under power boost of 10 dB, the electrical cosinusoidal signal is applied to the phase modulator, and the PSD at 1 kHz along the fiber array is also presented (Fig. 7). It is shown that the crosstalk is further suppressed by 23.85 dB and is -45.67 dB. The experimental results reveal that the cosinusoidal phase modulation approach can reduce the crosstalk, and the larger magnitude C of phase modulation leads to better crosstalk suppression.

We propose a cosinusoidal phase modulation approach to suppress crosstalk induced by Rayleigh backscatterings in a qDAS system. The hardware implementation of this technology is simple, and only a phase modulator and a cosinusoidal electrical signal source are applied behind the laser. Cosinusoidal phase modulation is adopted to the high-coherent laser, and crosstalk can be suppressed by carefully setting the phase modulation frequency. Larger phase modulation amplitude leads to better crosstalk suppression. The validity of the proposed approach is confirmed by 23.85 dB crosstalk suppression in the experiment. The proposed cosinusoidal phase modulation approach for crosstalk suppression is suitable for a variety of phase demodulation methods such as high-frequency heterodyne, PGC and 3×3. Meanwhile, this method will suppress crosstalk in a qDAS system based on weak reflection point, reduce system detection distortion, and improve qDAS detection ability. In addition to crosstalk, Rayleigh backscatterings act as intensity noise in a qDAS systems and deteriorate detection noise. The proposed cosinusoidal phase modulation approach also optimizes detection noise by suppressing the intensities of Rayleigh backscatterings. Relevant studies have been carried out, and the noise suppression effect will be quantified and evaluated in our future work.