Research Progress and Development Trend of the Multi-Illuminator-based Passive Radar （in English）

1. Introduction

Passive radar, also known as a passive coherent location system, is a brand-new bistatic or multistatic radar system that uses the Illuminators of Opportunity (IOs) to detect moving targets^[1-7]. The available IOs include terrestrial broadcast, television, and cellular signals. Passive radar is commonly equipped with two types of receiving channels: surveillance and reference channels. The surveillance channel is used to collect the target echo, while the reference channel is utilized to receive the direct signal from the IO. Passive radar first cross-correlates the signals of surveillance and reference channels to obtain the Range-Doppler (RD) map and then executes the detecting and tracking algorithms to observe moving targets. The working principle of passive radar is shown in Fig. 1. When modern frequency resource becomes increasingly crowded, passive radar is an appealing and advantageous surveillance technique because it does not require frequency allocation. This radar is also featured in silent operation, which offers an important sensing technique for the green development of the new era and future battlefield operations with low-to-zero radiation power.

Figure  1.  Detection schematic diagram of passive radar

DownLoad: Full-Size Img PowerPoint

Passive radar has gained close attention from academia, industry, and applied fields for its unique advantages over active radar. The exclusive merits of passive radar include (1) superb performance in low-altitude areas, (2) strong detection capability for low-velocity and small RCS targets, (3) flexible deployment, and (4) low network-based surveillance cost. Passive radar also has a broad perspective in military and civilian integrated fields due to its exclusive merits. In military fields, passive radar can be deployed in vital and sensitive areas, such as borders, offshores, islands, and reefs. The silent operation of passive radar promotes the early warning and anti-strike capabilities of the air-to-sea defense network. In civil applications, passive radar can be deployed to survey general aviation aircraft and illegal Unmanned Aerial Vehicles (UAVs) at airports or important venues. When passive radar is combined with the 2D video surveillance method, a 3D ground-to-air monitoring network can be constructed to provide a new surveillance alternative and location-based service for smart cities.

Dr. Wan (2012) concluded the status and trends of the UHF/VHF band signal-based passive radar at home and abroad^[3]. To show the theoretical and application progress of passive radar over the past years, this paper first reviews the evolving history, the latest developments of key technologies, and several experimental systems of passive radar. The recent research results on passive radar by Wuhan University are also elaborated. The challenges and development trends of passive radar are finally summarized.

2. Overview of Researches and Developments of Passive Radar

As a mature technique, passive radar was created before World War II and has evolved with time. The Daventry experiment in England in 1935 revealed that the illumination from the shortwave BBC Empire transmitter can be used to detect aircraft targets 10 km away. During World War II, Germany designed a practical passive radar system called “Klein Heidelberg”^[8], which receives the direct wave and target reflection from the transmission of England’s “Chain Home” active radar. “Klein Heidelberg” locates the aircraft by combining the Time-Difference-of-Arrival (TDOA) and the Direction-of-Arrival (DOA). This system has also successfully forecasted the bombing plane cutting through the English Channel.

During World War II, the innovation of the transceiver switch promoted the development of active radar, and the early hardware specifications and signal processing capabilities of monostatic active radar are better than that of passive radar. Thus, passive radar has undergone a long and tortuous period. The research interest in bistatic radars reemerged in the 1980s. During this time, an analog television signal-based passive radar system set up by Griffiths et al.^[9] at University College London has successfully detected moving aircraft. In the 1990s, Howland^[10] successfully tracked moving targets via analog television signals. The Lockheed Martin Company of the United States launched an FM-based commercial PR system called “Silent Sentry” in the late 1990s, which is a milestone product that designates the passive coherent location as the main detection schematic of passive radar.

Electronic countermeasures, anti-radiation missiles, and stealth technologies have rapidly developed over the past two decades. The bistatic or multistatic radar is advantageous over monostatic radar in solving these new challenges and has received renewed attention worldwide. During this period, the number of universities and research institutions that focus on passive radar research has steadily risen, and the depth and breadth of passive radar research have been markedly strengthened. The currently available IOs comprise broadcast and television signals^[11-18], cellular signal^[19,20], wireless local area network^[21], navigation and communication satellite signal^[22-25], as well as illuminations of active radar^[26,27]. Owing to the development of digital radio technology, the bandwidth of the digital wireless signal is normally irrelevant to the broadcasting content and is a suitable IO for target detection. Thus, passive radar has undergone its golden times.

Digital broadcast and television signals are currently becoming increasingly popular. The wide coverage and stable waveform characteristic of the digital broadcast and television signals grant their competence over analog broadcast and television signals. Thus, the digital broadcast and television signals are practical IOs for passive radar, and this type of passive radar has been the most popular research topic. Most countries have independently proposed their standard of digital broadcast or television signals covering HF, VHF, and UHF bands^[28-30]. The popularity and commercialization degree of digital broadcast or television signals have affected the development of this type of passive radar. Moreover, the digital broadcast or television signals of different countries are distinct in base-station placement, transmitting power, and signal format. Thus, this type of passive radar should consider the national conditions to promote system design. According to the publicly reported literature, studies of the passive radar in many European countries have a history of more than ten or even decades, and their technologies are at the forefront of the world. However, studies on the digital broadcast and television signals based on passive radar in China are delayed due to the low popularity of these signals, and most of the early investigations remain at the simulation level^[31-33]. Fortunately, the domestic research on this type of passive radar has been accelerated recently due to the ongoing coverage of digital broadcast and television signals. Several domestic experimental systems have also been developed, and their feasibility has been comprehensively verified by field experiments^[34-38].

3. Developments in Key Technologies

As mentioned above, passive radar relies on third-party noncooperative IOs for target detection. The uncontrollable transmitted signal, the complex propagation environments, and the serious interference of multipath clutter complicate the acquisition of target information. Many research institutions have made comprehensive explorations and obtained substantial advancements on the above issues in recent years. The following focuses on the research advancements of the main technologies in passive radar signal processing.

(1) Reference signal reconstruction. In passive radar, the reference signal is mainly used for time-domain clutter suppression and RD map calculation. Hence, the quality of the reference signal directly affects the detection performance. The reconstruction technique based on “modulation + remodulation” is usually adopted for the digital television-based passive radar^[39-48]. The modulation mode of the digital television broadcast signals can be divided into two types. The first is the Cyclic-Prefix Orthogonal Frequency Division Multiplexing (CP-OFDM) techniques, such as the Digital Video Broadcasting-Terrestrial (DVB-T) signal and the China Mobile Multimedia Broadcasting (CMMB). The second is the Time-Domain Synchronous Orthogonal Frequency Division Multiplexing (TDS-OFDM) techniques, such as the Digital television Terrestrial Multimedia Broadcasting (DTMB). The reconstruction method is tailored in accordance with the signal modulation modes. Baczyk et al.^[39] and Searle et al.^[40] successfully recover the transmitted DVB-T signal by applying the pilots and the cyclic prefix to synchronization, channel estimation, and equalization. Using the signal detection principle, Refs. [41,42] successfully optimize the reconstruction methods in Refs. [39,40], respectively, which can significantly improve the quality of the reconstructed reference signal even in cases of low Signal-to-Noise Ratio (SNR). Refs. [43,44] provide some suitable adjustments considering parameter selection and signal processing structure to the reconstruction method of DVB-T, and the transmitted second-generation Digital TV Broadcast (DVB-T2) signal is successfully reconstructed. The reconstruction processes of CMMB signals are demonstrated in Ref. [45], and their effectiveness is validated by simulated and experimental data. Compared with CP-OFDM modulation, the TDS-OFDM signal has no pilot and uses a Pseudo-Noise (PN) sequence instead of a cyclic prefix as a guard interval. Hence, the aforementioned reconstruction methods are no longer applicable. The reconstruction method of the DTMB signal is investigated in Ref. [46], and the PN sequence is used for signal synchronization, channel estimation, and equalization. Finally, simulation and measured data verify that this method can obtain a relatively pure reference signal. The baseband sampling rate is equivalent to the signal bandwidth; thus, the reconstruction method proposed in Ref. [46] may lead to the aliasing of high-frequency signal components in practical applications. Ref. [47] studies the DTMB reconstruction method under the oversampling condition by using the improved least square algorithm for channel estimation to address the aforementioned problem. Compared with the reconstruction method of baseband data rate, oversampling processing can improve the quality of reconstructed signals. The reconstruction method proposed in Ref. [48] is applicable for all OFDM modulated signals. However, multiple sensors are required, which substantially increases the system complexity. Moreover, the SNR of the received signal from the echo sensor must be higher than a certain level to maximize the performance of this method.

Compared with the passive radar on a fixed platform, numerous Doppler spread clutter are found in the reference signal of passive radars based on mobile platforms (airborne and vehicle). Thus, the reference signal reconstruction may encounter additional challenges. Fortunately, the antenna diversity technology has been successfully applied to solve the reference signal reconstruction problem in the airborne passive radar^[49]. In addition, the improved blind equalization method is an effective processing method to extract the pure reference signal^[50]. Overall, the reconstruction methods for digital television broadcast signals are mainly divided into two categories according to the signal modulation mode. These methods reconstruct the reference signal from the perspective of signal recovery. The optimal reference reconstruction method, which combines the reconstructed signal quality with the time-domain clutter suppression performance, must be further studied to improve the detection performance of passive radars.

(2) Multipath clutter rejection. The strong direct-path signal and the multipath clutter reflected by ground objects (generally named multipath clutter) will cover up the target and reduce the system sensitivity. Therefore, effective methods should be conducted to reject the strong clutter in surveillance channels to reduce the influence of the residual clutter and its sidelobes after matched filtering on target detection. The clutter rejection method in passive radars can be roughly divided into two categories, namely the time-domain and spatial-domain methods. Time-domain methods mainly include conventional adaptive filtering and time-domain projection algorithms. Time-domain adaptive filtering algorithms mainly comprise the Least Mean Square (LMS), the Normalized LMS, the Gradient Adaptive Lattice (GAL), and the Recursive Least Square (RLS) algorithm. Among them, the first three algorithms are characterized by their low computational complexity and slow convergence. On the contrary, the convergence rate of the RLS algorithm is relatively fast, but its calculation complexity is relatively high. Represented by the Extended Cancellation Algorithm (ECA), the time-domain projection algorithm includes the ECA, the Batch vision of ECA (ECA-B), and the Sliding ECA (ECA-S). ECA-B uses piecewise processing based on ECA, which can reduce computational complexity and improve robustness in time-varying environments. ECA-S can effectively eliminate the modulation peaks of slow target echoes induced by the segment operation of ECA-B by sliding operations^[51]. The performance of the aforementioned time-domain clutter rejection methods is limited by non-ideal factors, such as the fractional delay, carrier frequency offset, and channel frequency response. Ref. [52] effectively reduces the impact of these non-ideal factors and improves the robustness of clutter rejection by introducing sub-band processing. With the popularity of digital signals based on OFDM modulation, the subcarrier domain clutter rejection method has been developed^[53,54]. The commonly used subcarrier domain algorithms mainly include Average Channel Response Filter on subCarrier (ACRF-C), ECA by subcarrier, and Reciprocal Subcarrier Detector. Ref. [55] comprehensively evaluates the performance of these subcarrier domain algorithms and reveals their inner mechanisms and mutual relationships. Refs. [56,57] combine the segmented smoothing idea with the ACRF-C algorithm, which can effectively improve the clutter rejection performance for nonstationary and Doppler extended clutters.

In addition to the above methods, the target information extraction methods without clutter suppression are also discussed in Refs. [58,59], in which the orthogonality between OFDM symbol subcarriers is utilized. Additionally, the spatial information can be used in multichannel passive radars to form deep nulls in the clutter direction adaptively and achieve clutter rejection^[60-62].

Passive radar on mobile platforms (airborne and vehicle) has gradually become a new research focus due to its mobility and flexible deployment capability^[63-66]. Unfortunately, the movement of the platform will induce Doppler expansion of clutter and submerge low-speed targets. Simultaneously, passive radar is also affected by the time-varying characteristics of transmission waveform and the array calibration error. In this condition, the performance of the Displaced Phase Center Antenna (DPCA) and the Space-Time Adaptive Processing (STAP) is limited. The multilevel extended clutter suppression idea of cascading the adaptive array error calibration, the conventional clutter suppression methods (ECA and LMS algorithms), the reciprocal filter, and the DPCA (or the STAP) algorithm perform well in practical applications^[67-70].

Overall, many studies on clutter suppression methods have been conducted, but most of these studies mainly focus on a single method. Consequently, the parameter optimization and combination strategies of different clutter suppression methods and the coupling factors between clutter suppression and reference signal reconstruction must be further explored.

(3) Target detection. The passive radar receives the direct wave signal as well as the target echo from noncooperative illuminators and processes them coherently to realize target detection and extract target parameters. The main detectors of passive radar can be divided into the following three categories: Cross-Correlation Detector (CCD), Generalized Likelihood Ratio Test (GLRT) detector, and singular value detector. The cross-correlation detector is similar to the matched filter, which calculates the cross-ambiguity function of the reference and surveillance signals to realize the target detection ^[71-76]. In the presence of unknown parameters in the signal model, the generalized likelihood ratio detector can be obtained by substituting the maximum likelihood estimation of the unknown parameters into the likelihood ratio test^[77-81]. The third type of detector is the singular value decomposition detector based on modern random matrix theory^[82-84]. This type of detector considers the low-dimensional characteristics and the time-varying power of the target signal. The target SNR is generally improved to ensure detection performance. Such an improvement can be achieved by using a variety of methods, such as single-frame long-time coherent accumulation or multiframe noncoherent accumulation. The current research hotspot in the process of long-time accumulation is the effective compensation for the target range and Doppler migrations to improve the robustness of the algorithm to maneuvering targets^[85-88]. Researchers at the University of Rome, Italy, took the polarization diversity passive radar based on FM broadcast and DVB signals as the research object and proposed the polarization generalized likelihood ratio detection method. Their experimental results show that this method can effectively improve the target detection performance^[89-91]. In addition, the multifrequency joint detection method can effectively alleviate the impacts of non-ideal factors, such as the unstable transmission waveform (including FM broadcast signal) and the flicker of Radar Cross-Section (RCS), and significantly improve the detection probability of targets^[92-94].

(4) Target tracking. The performance of the passive radar with single bistatic pair is limited. The spatial and frequency diversities can be deployed to expand the dimension of information acquisition and obtain accurate target states through target tracking and information fusion in multistatic passive radars. Thus, the multistatic passive radar has attracted extensive attention. Among them, the recent research advancements mainly involve target tracking in Single Frequency Network (SFN) passive radar, cognitive tracking of passive radar, and data fusion of multiband passive radar. Multiple transmitters transmit the same signal at the same frequency simultaneously; thus, the association between measurements and transmitters is blurred. Refs. [95-100] study the ambiguity problem in SFN passive radars, and the data association between measurements, transmitters, and targets are correctly obtained, which finally leads to good target tracking results. Refs. [101,102] propose a concept of symbiotic radar, which uses part of the communication network for passive radar detection. This concept independently plans the transmitter of the communication network and the receiver of the passive radar through the tracking effect, thus finally realizing the purpose of cognitive tracking. Refs. [103-106] proposed a method to realize target tracking and data fusion by combining UHF and VHF band passive radars and performed numerous experiments to verify the feasibility of the system. The external illuminators present new challenges to the target tracking of passive radar. Therefore, the current research focus is to improve the mainstream tracking algorithm to solve these problems. However, the large-scale network partition handover and intelligent tracking in networked passive radar must still be further studied.

(5) Passive radar imaging. The research on passive radar systems in target detection, tracking, and positioning has made remarkable progress. Compared with target measurement and tracking, target images can provide rich and fin features, which are conducive to the classification and recognition of targets. Therefore, the research on passive radar imaging systems has attracted increasing attention from scholars at home and abroad. Compared with the imaging system of active radars, bistatic imaging has certain system advantages. First, the positions of transmitters and receivers can be flexibly configured to meet the practical application needs, such as forward-looking, combined, interference, and multipolarization imaging. Second, the passive radar system is easy for multistation networking and multifrequency information fusion. This system can fully obtain the target scattering information of different bands and perspectives, establish the target three-dimensional observation information, and realize the target three-dimensional imaging. Refs. [107,108] theoretically analyze the spatial resolution performance of Passive Bistatic Inverse Synthetic Aperture Radar (PB-ISAR). On this basis, the field experiments of target imaging, including aircraft and cargo ships, are conducted. The multibandwidth synthesis is realized by using three adjacent DVB-T signals to improve the imaging quality. The experimental results verify the accuracy of the theoretical model. Ref. [109] imaged the same cooperative ship by using DVB-T and DVB-S signals simultaneously. The experimental results show that the target size and main scattering points can be effectively estimated and distinguished under the two illumination sources, which lay a foundation for further information fusion. Ref. [110] effectively fused the spatial diversity information into the imaging process of the DVB-S signal to improve the imaging effect. The feasibility of Passive Bistatic Synthetic Aperture Radar (PB-SAR) has also been verified by experiments due to the development of mobile platform passive radar systems. Refs. [111-113] show the imaging results of airborne PB-SAR on the ground scene based on DVB-T signals. Refs. [114,115] present the imaging results of multistatic PB-SAR based on satellite navigation signals.

Most passive radar imaging systems are currently performed under the framework of ISAR/SAR imaging technology. As its imaging basis, the matched filtering is a linear process, which has the advantages of simple implementation and stable performance. However, the matched filtering performance is usually limited by the insufficient signal bandwidth of IOs, which affects the resolution of adjacent targets. The passive radar imaging algorithm based on sparse representation can effectively overcome the shortcomings of traditional imaging algorithms and achieve high-precision imaging. Ref. [116] utilized multiple non-adjacent DVB-T signals for bandwidth synthesis to achieve high-resolution ranges. However, the discontinuity between signals introduces high grading lobes. The compressed sensing method is used in Ref. [116] to reduce the grating lobe effectively and improve the imaging quality significantly. Ref. [117] used the simultaneous orthogonal matching pursuit algorithm to alleviate the influence of target scattering coefficient flickering with carrier frequency and observation angle on imaging results in multistatic airborne PB-SAR systems and realized accurate imaging of observation scenes.

4. Typical Passive Radar Systems and Experiments

Radar research is an experimental science, of which experimental verification is an essential part. With the popularization of digital broadcasting, television, and communication signals and the breakthrough of key technologies, several experimental passive radar systems have been successfully developed and reported publicly.

4.1 Typical Passive Radar Systems Abroad

In recent years, the German Academy of Applied Sciences (FGAN-FHR) has developed a series of passive radar experimental systems, including DELIA (DAB Experimental radar with LInear Array), PETRA (Passive Experimental TV RAdar), and CORA (COvert RAdar) systems. The parameters of these experimental systems have been introduced in Ref. [3]. In addition, FHR has cooperated with the Norwegian National Defense Research Institute to develop the LORA11 (linear array covert radar) system based on DVB-T signals^[118] (Fig. 2(a)). LORA11 uses 11 vertically polarized disk-cone antennas to form a receiving array with an array element spacing of 0.36 m, which can achieve 60° azimuth coverage. Experimental results show that the system can effectively detect aircraft within 10 km in the altitude range of 600–700 m.

Figure  2.  Typical passive radar systems

DownLoad: Full-Size Img PowerPoint

The German Hensoldt company has successively launched several versions of the TwinVis system^[119,120] (Fig. 2(b)) since 2008. Hensoldt showed the latest version of TwinVis at the 2018 Berlin international air show. The system can simultaneously use 16 FM signals and 5 DAB/DVB-T signals to achieve target detection. Simultaneously, the system can set up a processing center to fuse the information from four passive radars to obtain complete target information. The system adopts a three-layer receiving array configuration of 7-7-16 elements from bottom to top, which can realize 360° coverage of the target and the measurement of the target height. The experimental results show that the maximum detection distance of the system is more than 200 km, and the range accuracy is better than 250 m.

The AULOS system (Leonardo S.P.A. company in Italy) uses FM and DVB-T signals as illuminators^[121,122]. In this system, the FM signal is mainly used for monitoring the long-distance and wide-range air targets, and the DVB-T signal is utilized for surveillance of low, slow, and small targets in the air and coastal monitoring. The system includes two versions: fixed and mobile platforms (Fig. 2(c)). The mobile platform version adopts an 8-element uniform dipole circular array as the receiving array to realize 360° coverage of the detection area. The fixed platform version has a receiving array with four array elements, which can simultaneously cover a 120° target detection area and measure the pitch angle of the target. The experimental results show that the system can detect air targets at a distance of 200 km and effectively monitor the coastline.

The Patria company of Finland released the MUltiStatic Coherent Location (MUSCL) system^[123] (Fig. 2(d)) in 2018. This system uses FM and DVB-T signals to form a multistatic receiving mode, which can achieve 360° coverage of targets within hundreds of kilometers, track more than 100 targets simultaneously, and distinguish fixed-wing aircraft, propeller aircraft, and helicopters.

In addition to the above VHF/UHF band passive radar experimental systems, the passive radar systems based on Global Navigation Satellite System (GNSS), Long-Term Evolution (LTE), WiFi signal, and other illumination sources are also currently being developed and gradually applied^[124-129].

4.2 MIPAR Passive Radar Series of Wuhan University

Relying on the national key discipline of radiophysics and supported by a series of national key projects, the radio wave propagation laboratory of Wuhan University has continuously cultivated the field of passive radar for more than ten years. The passive radar system and its networking technology based on digital broadcast television signals of China have been systematically studied. The multiband multiarray Multi-Illuminator-based Passive Radar (MIPAR) system has been developed, and long-term field experiments have been conducted to verify its effectiveness and robustness. MIPAR has been successfully applied in many national key demand fields. The relevant systems and typical detection results are introduced as follows.

(1) HF band passive radar. Digital Radio Mondiale (DRM) digital broadcasting standard is the only standard of global shortwave digital sound broadcasting determined by ITU-R. In 2011, the team of Wuhan University conducted the first domestic passive radar research experiment based on DRM signal, which confirms the feasibility of this technology^[18,130,131]. The influence of ionospheric traveling interference, radio frequency interference and other factors on the system is then comprehensively analyzed ^[132-134], and a radio frequency interference suppression method based on maximum likelihood estimation is proposed. Aiming at the problem of target positioning under the combination mode of sky and ground waves, the team proposed a new positioning method based on the arrival elevation of direct waves and analyzed the positioning accuracy of this method^[135]. Numerous shortwave broadcasting, communications, and radar signals are observed in the shortwave band. Owing to the continuous deepening of theoretical research and the increasing improvement of the experimental system, the team also performed target detection experiments based on multi-illuminators of HF bands in several places across the country in recent years. Experimental results show that the radar system can realize over-the-horizon detection and continuous tracking of moving targets. Fig. 3 shows the results of aircraft target detection in a test. The change trend of target bistatic range and velocity obtained by radars are in good agreement with the corresponding Automatic Dependent Surveillance-Broadcast (ADS-B) information, which proves the accuracy of radar detection results. High-frequency passive radar integrates advantages of traditional high-frequency sky/ground wave and bistatic/multistatic passive radars. This radar can also realize long-range target early warning and demonstrates broad application prospects.

Figure  3.  Comparison between HF-band passive radar detection results and ADS-B information

DownLoad: Full-Size Img PowerPoint

(2) VHF/UHF band passive radar. The team of Wuhan University has successfully developed a multichannel VHF/UHF band passive radar system. The system has a compact structure, flexible architecture, and stable performance. This system can complete the task of 24/7 all-weather real-time target monitoring. The system can be compatible with DTMB/CMMB (China), DVB/DVB-T2 (European)/FM/CDR, and other signals. According to the actual detection requirements, the system can choose the uniform circular array as the receiving array to realize the omnidirectional coverage of the detection area or the uniform linear array as the receiving array to realize the reliable monitoring of key areas. Using this experimental system, the team has performed numerous experiments on UAVs, general aviation aircraft, civil aircraft, and other types of targets in more than ten places across the country, breaking through a series of key technologies in signal and data processing, such as reference signal reconstruction, clutter suppression, target detection, target tracking, and data fusion^{[47,52,87,136-144]}. The developed radar system also conducted application demonstrations in multiple general aviation airports in Hubei, Henan, and Sichuan provinces. In 2017, the team first reported the detection results of UAVs using passive radar based on digital TV signals at the IEEE International radar conference, showing the potential application of this technology in the field of “low, slow, and small” target monitoring^[141]. The team takes advantage of the easy networking of passive radar to perform a multistation networking fusion experiment to build a tight low-altitude surveillance network. By integrating the information obtained by multiple transceiver pairs, the radar effectively compensates for the detection blind area, expands the detection range, and improves the detection accuracy^[142]. The experimental results show that the horizontal positioning accuracy of the system for general aviation aircraft is better than 50 m, the speed accuracy is better than 1 m/s, and the data update rate is less than 1 s. Typical detection results are shown in Fig. 4(b). The team also performed time-division multifrequency experiments^[143], which use a single receiver to obtain the spatial diversity information of the target, to further reduce the cost and improve the system flexibility. Simulated and experimental results indicate that its target detection performance is close to that of the simultaneous multifrequency system. The detection results are shown in Fig. 4(c). In addition, the team made experimental advancements in the single-channel multistatic networking system^[144], as shown in Fig. 4(d).

Figure  4.  Detection results of VHF/UHF band passive radars

DownLoad: Full-Size Img PowerPoint

(3) Passive radar system based on LTE signal. As the fourth-generation digital mobile communication standard, LTE has many characteristics, such as wide distribution, large redundancy, and easy networking. Therefore, LTE has become a widely concerned IO^[145,146]. The team of Wuhan University comprehensively analyzed the LTE signal structure and effectively suppressed its inter-frame blur band^[147]. On this basis, a highly integrated and miniaturized multichannel passive radar system based on AD9361 was independently designed and developed. The system was used to conduct the first ground and low-altitude target detection experiment based on the LTE signal (1867.5 MHz FDD-LTE) in China^[148,149]. Fig. 5 shows that the system realizes the continuous monitoring of ground vehicles and UAVs. The radar detection results are consistent with the real target information, which confirms the accuracy of the results and shows the application potential of LTE passive radar.

Figure  5.  Detection results of LTE-based passive radar

DownLoad: Full-Size Img PowerPoint

5. Development Trends of Passive Radar

At present, the theoretical research of passive radar has been fully verified, many breakthroughs have been made in key technologies of passive radar, and the system feasibility of passive radar has been widely recognized. However, a large gap still exists for the existing passive radar to meet the needs of practical applications. The main bottleneck problems of the existing passive radar can be summarized as follows.

(1) The environmental adaptability of passive radar is insufficiently strong. Passive radar is a type of electromagnetic environment-dependent radar. Its architecture, processing scheme, and performance are determined by the space-time coverage and signal characteristics of the IO. When passive radar is deployed in different places, determining the location, frequency, mode, power, and other parameters of the IOs in advance is necessary, and the hardware and software configurations should be customized correspondingly.

(2) The robustness of passive radar is insufficiently high. The existing passive radar is usually based on a single (or single class) IO. In addition to the waveform characteristics, the performance of passive radar is also affected by the system conditions, station settings, and surrounding environments. The technical index measured under a certain IO in a certain place may change elsewhere, resulting in the unstable performance of the passive radar.

(3) The intelligence of passive radar is relatively low. The performance of passive radar is closely related to the signal waveform, electromagnetic environment, physical channel, and system configuration scheme. The existing processing methods often require human judgment and intervention to approach the best working point. The specific system configuration scheme remains unchanged for a long time after its identification. Dynamic adjustment mechanisms and intelligent processing means based on on-site state evaluation feedback are also lacking.

The traditional bistatic architecture of single transceiver pairs should be broken through to solve the bottleneck problems of the passive radar in environmental adaptability, index robustness, and intelligent processing. Based on the electromagnetic environment recognition, optimal system configuration, and intelligent information processing, building passive radar networks with multiple IOs may be a promising solution to break through the above bottlenecks. Fig. 6 shows that MIPAR has strong plasticity in technical system and application mode, which can substantially expand the information acquisition dimension.

Figure  6.  Schematic of multi-illuminator passive radar

DownLoad: Full-Size Img PowerPoint

The overall development trend of passive radar networks can be summarized in the following three directions.

(1) Integration of multiple IOs. With the rise of digital broadcasting, digital television, and digital communication network in the world, the optional IOs of passive radar have expanded from analog modulation signals, such as FM broadcasting and TV sound, to digital signals, including digital audio broadcasting (DAB/CDR), digital television (DVB/DTMB), satellite, mobile communication (LTE), and wireless local area network (WiFi). This expansion provides favorable conditions for the integration of multiple IOs.

The system configuration, detection power, and detection accuracy of passive radars are affected by the waveform characteristics, transmission power, and coverage of the IOs. Therefore, passive radar systems using a single IO are generally suitable for specific application scenarios. For example, passive radar based on FM broadcast signals is used for long-distance and large-scale air target monitoring but cannot be competent for the task of “low, slow, and small” target monitoring in key areas. On the contrary, the miniaturized passive radar based on digital TV signals can play an important role in the construction of a low-altitude surveillance network. However, addressing the requirements of long-range early warning of civil and fighter aircraft is difficult. The integration of multiple IOs can help effectively expand the application scenarios, improve the target detection performance, and reduce the dependence on the environment. First, the passive radar must be endowed with the capability of radiation source screening through the design idea of software and modular integrated systems. Then, the multidimensional information provided by multiple transceiver pairs should be effectively integrated and utilized. The integration of multiple IOs with different frequency bands (HF/VHF/UHF), systems (analog/digital), and coverage modes (multifrequency network/ SFN) is complementary. For example, the TwinVis system developed by Hensoldt company of Germany simultaneously uses FM, DAB, and DVB-T signals as IOs for target detection, which is a typical integration attempt of multiple IOs^[106,150].

The integration of multiple IOs will be an important measure for passive radar systems to meet the needs of high-precision (large signal bandwidth), long-distance (large signal transmission power), large-range (wide signal coverage), and multiscene (ground, sea, and air target detection) detection. The development of multi-illuminator integrated systems and multi-illuminator information mining and fusion will be the focus of the next research.

(2) Network configuration. The location method based on the azimuth and distance is adopted in passive radars with single transceiver pairs. The positioning error of this method increases with the oblique distance. Thus, the detection accuracy within the radar coverage cannot be fully guaranteed. Simultaneously, with the change in target attitude and the occlusion of buildings and mountains, ensuring the continuous detection of targets is difficult. By contrast, the passive radar network with multiple transceiver pairs can markedly improve the positioning accuracy of targets through the TDOA positioning method. Moreover, the detection blind area can be reduced by the reasonable placement of radar receivers. Therefore, networked detection is an important and attractive development direction in passive radar. Such detection can provide comprehensive and detailed detection results by maximizing the spatial, frequency, and polarization diversities in the process of multistation information fusion.

Compared with passive radars with single transceiver pairs, networked passive radars have advantages in detection accuracy, stability, and reliability. Fig. 7 shows the detection results of multi-illuminator passive radars^[142]. The tracking results after fusion are significantly improved compared with those of single passive radars in continuity, accuracy, and number of false tracks. Many research institutions and universities perform investigations on multistation detection^[151,152], multistation positioning^[153-155], multistation tracking, and information fusion^{[105,156-159]}. The detection performance of networked passive radar will be largely affected by the placement of radar receivers and the selection of transmitters.^[160,161]. However, studies on these issues are relatively lacking.

Figure  7.  Detection results of the multi-illuminator passive radar

DownLoad: Full-Size Img PowerPoint

Overall, investigations on the networked passive radar are currently in the development stage. The core problems, such as the optimal configuration of transceiver stations, joint positioning of multiple transmitters and receivers, data association, and large-scale network partition handover, still need to be studied and broken through.

(3) Intelligent information processing. Passive radars use noncooperative illuminators whose transmitting stations, radiation parameters, and signal waveforms are not designed for radar target detection. Therefore, these radars are highly dependent on the environment, and the signal processing process is substantially more complex than that of active radar. Under these conditions, intelligent information processing, which can autonomously sense the detection environment and make adaptive adjustments, will markedly improve the detection effectiveness of passive radars. Intelligent information processing can be applied throughout the passive radar acquisition and processing chain, from IO selection to target identification. Intelligent search for the best possible IO, intelligent signal processing^[162], and intelligent target classification and identification^[163-165] will all be part of the processing. The process employs a “smart” brain to compensate for the shortcomings of the passive radar system, transforming the radar into an intelligent system capable of “thinking and learning” and adapting to complex detection environments.

Intelligent signal processing has a wide range of potential applications in passive radar processing. Considering reference signal reconstruction, a combination of electromagnetic environment perception and channel-independent modeling will help improve channel estimation and equalization performance, resulting in a clean reference signal. An intelligent sensing module for clutter characteristics (comprising components such as clutter type determination, clutter distribution modeling, and clutter parameter estimation) can be established to provide useful a priori information on clutter. Simultaneously, the combination of time-, frequency-, spatial-, and subcarrier-domain clutter suppression methods can be independently optimized, which will help significantly improve clutter suppression performance. Considering target detection, refined modeling and feature learning of residual clutter or complex clutter areas can be performed to extract personalized information regarding clutter and targets, which may be helpful for improving the detection probability of weak targets and reducing false alarms. Fig. 8 depicts the conventional CFAR and the multilayer perceptron detection results^[166], demonstrating the superior performance of the intelligent detection algorithm. Considering target tracking, self-calibration of track errors can be accomplished by comparing and learning the radar track and the corresponding reference information (such as ADS-B, GPS, and AIS information). Simultaneously, self-assessment of the target movement situation can help ensure target tracking continuity and improve track quality. Considering target state monitoring and target classification, intelligent information mining and classification algorithms based on physical features, such as phase, SNR, micro-Doppler, and track morphology, are expected to deviate from traditional image recognition-based processing frameworks. Simultaneously, the increasingly rich dimensions of information acquisition lay the groundwork for accurate target classification, target motion prediction, and other promising applications. Intelligent signal processing has considerable potential but is still in its early stages and requires additional in-depth research. Furthermore, in addition to using intelligent algorithms to improve the performance of a single processing module, the system is required to adjust processing parameters intelligently, select optimal processing algorithms, and the most efficient processing flow based on factors such as the detection environment and the type of targets.

Figure  8.  Comparison between the detection results of the intelligent and constant false alarm rate detectors

DownLoad: Full-Size Img PowerPoint

6. Conclusion

Currently, the research on single signal processing technologies, such as reference signal purification, clutter suppression, and coherent detection of passive radar, has been relatively in-depth, and its system feasibility has been widely recognized. The research on the passive radar is in the stage of prosperity, with many different ideas and schemes. However, with the deepening of research, the shortcomings of passive radar detection performance limited by external IOs have become increasingly prominent, which objectively urges the germination of ideas of “networking detection” and “intelligent information processing.” In information processing, the intelligent processing idea driven by machine learning is first explored. In addition to the track information, comprehensive and fine features of the targets are extracted for target classification and recognition. Therefore, the development of passive radar presents the trend of networking, intelligence, and multi-illuminator integration.

The combination of networking and intelligence in MIPAR forms a closed-loop process, which increases the autonomy of the original pure passive working mode of passive radars and qualitatively improves the overall performance of the passive radar. Therefore, in-depth research on the theory and method of intelligent detection in passive radar networks is of considerable importance to break through the bottlenecks in the development of passive radar technology, solve the deep application problem of passive radar, and promote the development of new system radar technology.