Microwave Journal
www.microwavejournal.com/articles/35410-holographic-radar-development

Holographic Radar Development

February 12, 2021

Holographic radar system architectures are less complex than conventional surveillance array radar architectures, yet holographic radars offer capabilities exceeding those of the most sophisticated active electronically scanned array radars. Holographic radars can therefore challenge conventional radars in both cost and performance.

Holographic radar is not constrained by the sampling nature of scanning radar and so (loosely) has the capability to provide information with higher dimensionality. Fundamentally, staring radars meet the requirements of the Electromagnetic Uniqueness Theorem and provide an analytical solution to all electromagnetic interactions within their coverage. Conversely, scanning radars heavily under-sample so they cannot meet the same criteria. Some further benefits of staring radar are 1) the capability to provide very high Doppler frequency resolution by observation over dwell times unconstrained by the scan pattern or surveillance constraints and 2) contiguous measurement with high update rates, so fleeting, rapidly varying or complex returns can be sensed.

These characteristics enable high-resolution clutter rejection (i.e., detection in the presence of complex background environments of complex target returns) and optimized recognition and tracking. The efficacy of staring operation has been demonstrated with both modeled and real targets and clutter, on both large and small scales and in different applications.

As a basic example, consider detection of a complex maneuvering target. A scanning radar with a 2-degree azimuth beamwidth might detect and report its position every second, with a Doppler frequency resolution of 200 Hz. A holographic radar, however, can divide its radar return into 2 Hz wide components, for retaining and reporting a fine-grain Doppler signature, and fully sample its varying position and velocity, to rapidly determine changes in trajectory for effective tracking and threat assessment. As such, holographic radars provide both highly flexible radar coverage and utility, offering a solution to provide ubiquitous radar sensing - the ultimate in flexibility.

HOLOGRAPHIC RADAR ARCHITECTURE

Figure 1

Figure 1 Holographic radar antenna patterns showing a broad transmit beam (red) and mesh of narrow receive beams (blue).

Fundamentally, the architecture of a holographic radar comprises a transmitter that constantly transmits over the entire field of view and an array receiver with a persistent mesh of narrow beams that fill the field of view, negating the need to scan to provide accurate angular measurements (see Figure 1). As a design consequence, the transmitter and the receiver units do not share a common aperture or common RF modules, since a broad transmit pattern is not efficiently produced by the type of multibeam array antenna required by the receiver. Such separation of the transmit and receive circuitry reduces their respective complexities (see Figure 2). The transmit antenna is usually a simple fixed planar array or a reflector antenna, and the receiver array is significantly simplified, since it does not contain high-power RF circuitry or complex transmit-receive switching elements.

Although staring radars have simplified RF, IF and control hardware and contain no moving parts, the consequence is increased processing complexity. The number of simultaneously processed beams is significantly increased. For example, a scanning radar typically processes only one beam at a time, while a staring radar processes many beams simultaneously. Also, the quantity of data processed in a single dwell is generally higher, to provide similar sensitivity to an equivalent scanning radar - since a scanning radar contains a high gain antenna on both transmit and receive - and to realize the benefits of a staring radar, such as high-resolution Doppler processing. Note, for example, the processing load of a fast Fourier transform, which is commonly used in radar signal processors, has an order greater than one with increasing length.

f2.jpg

Figure 2 Holographic radar system architecture.

Staring radars tend to be smart, adaptable and defined by software. Furthermore, due to sensor persistence, staring radars are ideally suited to operate with multiple coincident modes in multiple, software-defined processing chains and/or cognitive processes.

f3.jpg

Figure 3 Hardware design approach enables multiple applications.

MODULAR ETHOS

The radar hardware is designed to be re-used for multiple applications and units that can be reconfigured (see Figure 3). This design ethos results in numerous atypical radar design features, including:

  • Antenna elements spaced at λ/2, allowing scanning and adaptive processing over up to ±90 degrees from boresight.
  • Independently digitized receiver elements to maximize flexibility, allowing the creation of small to large arrays.
  • Signals digitized at IF to minimize calibration problems, with digital down-conversion local to the element, in addition to other rudimentary pulse-level, raw data radar-processing functions.
  • Use of a single low phase noise oscillator reference distributed to all transmit and receive modules, each deriving its own RF frequencies and clocks. Each receiver module comprises just 2 x 2 elements, for high flexibility.
  • Use of distributed simple timing signals, processed locally within each transmit and receive module for detailed timing.
  • Transmit power generated in a single or small number of solid-state power amplifiers and distributed to a small number of transmitter elements, appropriate to broad-beam transmission.
  • Near linear mode transmit power amplifier operation, with digitally controlled pulse-shaping technology to manage spectral masks and sidelobes.
  • A calibration network for monitoring the transmitted waveform, so the smallest variations in performance can be tracked and accounted for.
  • A flexible and expandable processing architecture comprising GPUs and general purpose server-class processors.


HOLOGRAPHIC RADAR DEVELOPMENTS

Figure 4

Figure 4 Radar developed for long-range detection, tracking and recognition of drones.

The original concept for holographic radar was as a C-Band demonstrator for a defense application. Successful proof of that concept led to the development of a long-range holographic radar for the civilian market, to address windfarm mitigation in air traffic management systems. Given existing air traffic control radar standards and long-range requirements for this application, L-Band was selected for its frequency of operation. From this early work, the Theia radar family1 was developed to address a range of applications, from land-based windfarm mitigation to longer range systems covering off-shore windfarms. With operational systems in continuous use for over five years, Theia is a good example of the expandable modular architecture in practice.

This modular architecture formed the foundation for developing a new concept radar with a conformal antenna, to investigate the efficient implementation of a staring array comprising 12 interdependent planar arrays for 360-degree azimuthal coverage. A demonstrator radar system, developed to explore this concept, contained four planar arrays, each with 128 receiver elements and sufficient transmit power to achieve 54 nautical mile detection of passenger aircraft. This four array demonstrator, nicknamed the Quad, was the basis for developing further ubiquitous radar sensing concepts and applications, including detecting drones. The initial investigations indicated the technology can support flight safety requirements: long-range detection, tracking and recognition of drones. A new product, named Gamekeeper, was rapidly developed to meet the needs of this emerging market, marking the coming of age of holographic radar (see Figure 4).

COUNTER DRONE RADAR

Radar is the only sensor capable of providing the essential sensing requirements for a drone surveillance and counter-UAV system: long-range, all weather performance; non-cooperative/autonomous drone detection and recognition; and consistent accuracy, coverage and performance. Market requirements for such a system are very demanding. They include: a high power-aperture product capable of detecting very low radar cross section (RCS) objects up to 7.5 km range; high dynamic range for detecting very low RCS objects in the presence of high clutter; high elevation accuracy for effectively discriminating air targets from ground objects; and high confidence tracking and recognition, to reduce the burden on operators given “confuser” targets in environments of interest. Normally, these requirements can only be met by large, high cost military radar systems, such as weapons locating radars.

Aveillant has addressed these market needs with a system more compatible with the civilian market. An initial operational capability for early adopters was developed for operational deployments in 100 days, by reconfiguring standard transmit and receive modules and commercial off-the-shelf (COTS) components used across Aveillant’s product line (see Figure 5). Gamekeeper radar’s capability has continuously evolved since its first operational deployment, with performance, stability and operational improvements released regularly, often multiple times per year, predominantly through software upgrades. These upgrades use a combination of adaptive agile development methodologies and the design features of holographic radar that make it a software-defined radar. This enables rapid technology insertion; in some cases, low technology readiness level techniques and technologies have been matured and implemented in the Gamekeeper radar to provide both step changes and continual improvements in capabilities.

f5.jpg

Figure 5 Radar family based on the reconfigurable design approach.

TECHNOLOGY ENABLERS

Key enablers for this adaptivity include advanced pulsed waveform control and state-of-the-art processors. Advanced pulsed waveform control is essential, given that the electromagnetic spectrum is congested, particularly in areas around airports and urban areas. Thus, radar transmissions, particularly those providing new capabilities, are highly controlled. This objective is more easily achieved using a staring radar architecture, versus other array antenna architectures, as the staring radar contains just one independent, easily accessible transmit channel. Transmitter and waveform control properties include:

  • Linear, solid-state power amplifiers. At L-Band, COTS units using LDMOS transistors are readily available. They provide several kilowatts output power and greater than 50 dB gain with gain flatness less than 1 dB, which simplifies interfacing.
  • Programmable and adaptable, high dynamic range pulse-shaping circuits. Implementations include circuits based on calibrated vector modulators or combinations of controllable phase shifters and attenuators. Using FPGAs, these circuits adjust transmitted RF amplitudes to accuracies less than 1 dB and phase to within 2 degrees approximately every 10 ns.
  • Calibration networks, with all receivers continuously evaluating transmitted waveforms. These passive networks split a coupled version of the transmitted waveform and route it to the input of each receiver element during designated calibrated periods. The signal is analyzed real-time to ensure consistent performance. Passive splitter networks and cable distribution networks have high stability in both phase and amplitude over their lifetimes and operating temperature ranges.
  • Full synchronization and adaptive control of transmitted waveforms and receiver modes. Programmable waveform generators and processing in each receiver element module (using FPGA and Zynq SoCs) enable rapid adaptivity. All processors are synchronized to an accurate timing datum generated at a single source. Adaptivity includes changing the transmitted waveforms and associated matched filters and can include receiver processing modes to adapt to changing requirements or environments.
  • High-order filtering of transmitted and received signals, ensuring low interference and susceptibility to adjacent spectrum users. While accurately controlling the transmitted waveforms ensures transmissions are accurately constrained in frequency, to ensure no unintentional transmission, the transmitted RF signals and all local oscillator frequencies require high-order filtering, typically using surface acoustic wave filters. Also, due to high use of the electromagnetic spectrum, particularly at L-Band, to operate in the presence of other users of adjacent frequencies, multiple stages of filtering are necessary to remove interference. High-order receiver filtering is required before the low noise amplifier, to prevent saturation, and at multiple stages within the receiver, to minimize the likelihood of interference affecting performance. Using this design methodology, signals at just 40 MHz separation from the holographic radar’s center frequency are attenuated by 120 dB.
  • Low gain receiver antennas and IF digitization, enabling adaptive interference monitoring and cancellation. A low gain receiver antenna reduces the risk that in-band and adjacent frequency systems will saturate the front-end of the receiver. Antenna gain is provided by digital beamformers which can be configured to place deep nulls in the directions of interference sources. If moving, this is performed adaptively. If digitization is performed using digitizers operating at sample rates many times the radar’s bandwidth, which enables perfect quadrature down-conversion (i.e., no image frequencies) and the ability to sense returns over much greater bandwidths. Digital decimating filters are typically employed in digital down-conversion; adaptive digital filters can be used to minimize sources of interference at any adjacent frequencies within the digitized bandwidth.

The use of state-of-the-art processors has enabled significant developments unbounded by the limitations of processing performance. Key aspects of the processing architecture include:

  • Processors with supported, clear development roadmaps, enabling future upgradability. Typically, these are supplied by the market leaders, such as Intel for general purpose processing, NVIDIA for GPUs and Xilinx for FPGA and SoCs.
  • Expandable processing architectures compatible across the product line. Initially, range processing is performed at the element level, which is then followed by beamforming processing. Algorithms are configured so that different bands of ranges or angles can be processed in isolation by different processors, enabling division of both processing resources and data. This capability is facilitated using a combination of high bandwidth network switches and lossless data compression at key data communication interfaces.
  • Use of FPGAs and Xilinx Zynq SoCs to provide critical timing functionality and high rate element level processing. For example, these devices perform down-conversion of signals in approximately 100 MHz sample rate data streams with sub-ns timing; they also provide flexibility to meet the needs of multiple applications.
  • Use of GPUs for element, antenna array and raw array signal processing. GPUs are ideal for common signal processing tasks at high processing rates, so they are effective performing the high data rate radar signal processing for beamforming, Fourier transforms and detection.
  • Server-class processors for conditional processing, general purpose processing and interfacing to other systems. These processors are used for operations such as tracking and recognition.

Examples of further development include:

Figure 6

Figure 6 An OCXO enables low phase noise and improves detection of small targets in high clutter.

  • Innovations to reduce the radar’s overall phase noise, to enable detection of smaller targets in the presence of high clutter. Ovenized crystal oscillators (OCXO) provide cost-effective, low phase noise: the latest devices have provided a 100 MHz reference frequency to the radar with -160 dBc/Hz phase noise at 1 kHz offset (see Figure 6). Further improvement is achieved by developing filters in frequency multiplication circuits to improve the phase noise at higher frequency offsets.
  • Numerous improvements in tracking and target recognition, to use the benefits of operating in a staring radar configuration. As an example, machine learning is used to determine the difference between a bird and a drone, judging their respective track kinematics and the properties of their micro-Doppler signatures.
  • Adaptive beamforming, to improve the sensitivity in the presence of high static and dynamic clutter. Holographic radars are used in high clutter environments, such as airports and urban areas. Given the natural sidelobes of planar antenna arrays are high compared to a parabolic antenna, static and dynamic clutter (e.g., road vehicles) can obfuscate targets across a radar’s entire field of view. Adaptive techniques have been developed to identify regions of high clutter and generate beam patterns with nulls in these regions, to reduce the impact of clutter across the entire field of view, significantly improving detection sensitivity in affected environments.
  • Techniques to enable multiple radars to coexist using the same frequency channel without interference. This development uses waveform and processing schemes and implements synchronization and frequency locking circuitry between radars to provide a common pulse timing reference and identical frequency reference.

THE FUTURE

Like the challenges offered by the proliferation of windfarms and drones in the medium- to long-term future, concepts of modular staring radar architectures will be adaptable to emerging sensor needs. In the shorter term, the development of variants with extended coverage, higher power and higher sensitivity against drones are planned. System developments will improve discrimination and overall installed performance, which are essential to reducing the operator burden of security systems. Development plans include implementing concepts such as deep learning, adaptive cognitive techniques and frequency and polarization diversity.

Holographic radar is integral to the development of non-cooperative sensing concepts for unmanned traffic management systems, which will enable drones to operate safely in the lower airspace environment, unlocking their potential. Initial concept development and experimentation has begun as part of the National Beyond visual line of sight Experimentation Corridor initiative. This will grow to meet future needs for ubiquitous sensing, broad area and localized generic sensing for applications ranging from security to environmental monitoring. This need to improve coverage while maintaining accuracy is unlikely to be met by increased radar power and sensitivity, rather by sensor proliferation. In the near future, multi-site and multi-static holographic radar (polygraphic) operation is inevitable.

References

1. “THEIA 16A Windfarm Tolerant Radar,” AVEILLANT Ltd., Web: www.aveillant.com/products/theia16a/.

2. “GAMEKEEPER 16U Counter-UAS Radar,” AVEILLANT Ltd., Web: www.aveillant.com/products/gamekeeper/.