Radar system using multi-tonal frequency modulation and related methods

Abstract

A multi-tonal frequency-modulated continuous-wave (FMCW) radar system for enhanced motion detection and object classification. The system includes a voltage-controlled oscillator (VCO) that varies rad

Description

TECHNICAL HELD

[0001] The present disclosure relates to radar signal processing and detection, specifically an enhanced Frequency-Modulated Continuous Wave (FMCW) radar system using multi-tonal frequency modulation for improved motion detection and classification. BACKGROUND

[0002] Radar systems have long been utilized for detecting and tracking objects by emitting radio waves and analysing the reflected signals. Frequency-Modulated Continuous Wave (FMCW) radar is a popular type of radar system that uses frequency modulation to measure the velocity and range of objects. Traditional FMCW radars typically employ a single-frequency sweep, which can be effective in certain scenarios but faces significant limitations in complex environments. These limitations include poor signal-to-noise ratio (SNR) in cluttered settings, susceptibility to multi-path interference, and restricted classification capabilities, especially for weak Doppler signals.

[0003] In real-world applications, such as unmanned aerial vehicle (UAV) detection, surface velocity radar for hydrology, and security surveillance, these limitations become more pronounced. For instance, small UAVs often produce weak Doppler returns, making them difficult to detect and classify amidst background noise. Similarly, in hydrology, turbulent water flow results in a broad Doppler spread, complicating the differentiation of structured flow. In security and industrial monitoring, distinguishing human movement from vehicles or background clutter poses a challenge. These scenarios highlight the need for improved radar systems that can enhance Doppler signal clarity, improve object classification accuracy, and increase resilience to environmental noise. SUMMARY

[0004] The multi-tonal frequency-modulated continuous-wave (FMCW) radar system may include a voltage-controlled oscillator (VCO) that can vary the radar transmission frequency and a signal processing unit that can extract Doppler signals at different frequencies. The VCO may 1 be driven by multiple simultaneous sine waves or may exploit nonlinearities to generate additional harmonics, creating multiple distinct Doppler frequency bands. The signal processing unit can extract the Doppler signals separately at these different frequencies to increase the signal-to-noise ratio (SNR) and reduce interference. An artificial intelligence (Al)-based classification system may process the extracted Doppler signals for improved object classification and velocity discrimination.

[0005] The system may include multiple simultaneous sine waves, which can comprise at least two sine waves with frequencies selected from a group consisting of 5 kHz, 11 kHz, and 19 kHz.

[0006] The signal processing unit may also be configured to demodulate a received radar signal to obtain in-phase (I) and quadrature (Q) components at an intermediate frequency (IF).

[0007] The I and Q components may be digitized using an analogue-to-digital converter (ADC). In some embodiments, the signal processing unit partitions the digitized I and Q components into overlapping time frames for time-frequency analysis. Each frame may be multiplied by a window function, such as a Hann window, to taper the edges and minimize spectral leakage. For each frame, the signal processing unit may compute a complex fast Fourier transform (CFFT) to generate a Doppler spectrum that captures both positive and negative Doppler frequencies. These spectra are then assembled over time to form a Doppler spectrogram.

[0008] The multi-tonal frequency modulation may produce a series of spectral sidebands with amplitudes governed by Bessel functions. The signal processing unit may be configured to isolate and exploit specific FM sideband orders from the Doppler return signal by applying digital filtering or demodulation, thereby yielding multiple parallel Doppler channels corresponding to distinct FM modulation orders. The Al-based classification system may comprise at least some examples of a convolutional neural network (CNN) and a transformer model.

[0009] The system may further include an unmanned aerial vehicle (UAV) detection subsystem that can differentiate UAV types using multi-tonal Doppler spectrograms.

[0010] The UAV types may comprise at least some examples of quadcopters and fixed-wing drones.

[0011] The system may also include a surface velocity radar subsystem that can enhance river flow velocity measurements by distinguishing structured flow from turbulent flow using the multiple distinct Doppler frequency bands.

[0012] A security and surveillance subsystem may be included in the system, which can detect and differentiate human motion, vehicle movement, and intrusions using the Al-based classification system.

[0013] The security and surveillance subsystem may further differentiate between walking and running, and between humans and vehicles.

[0014] A method for improved motion detection and classification using the multi-tonal FMCW radar system may involve driving a VCO with multiple simultaneous sine waves or exploiting nonlinearities to generate additional harmonics, thereby creating multiple distinct Doppler frequency bands. The method can include extracting Doppler signals separately at different frequencies to increase SNR and reduce interference, and processing the extracted Doppler signals using an Al-based classification system for improved object classification and velocity discrimination.

[0015] The method may further involve demodulating a received radar signal to obtain I and Q components at an intermediate frequency IF and digitizing the I and Q components using an ADC.

[0016] The method may also include at least some examples of differentiating UAV types using multi-tonal Doppler spectrograms, enhancing river flow velocity measurements by distinguishing structured flow from turbulent flow using the multiple distinct Doppler frequency bands, and detecting and differentiating human motion, vehicle movement, and intrusions using the AI-based classification system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] FIG. 1 illustrates components of a multi-tonal FMCW radar system, including signal processing and Al-based classification subsystems.

[0018] FIG. 2 illustrates enhancing object classification and velocity discrimination using AI and Doppler frequency bands.

[0019] FIG. 3 is a flowchart that details the signal processing steps for generating the spectrogram. DETAILED DESCRIPTION

[0020] The disclosure provides a multi-tonal frequency-modulated continuous-wave (FMCW) radar system designed to enhance motion detection and classification. The system may include a voltage-controlled oscillator (VCO) that can be driven by multiple simultaneous sine waves or may exploit nonlinearities to generate additional harmonics, creating multiple distinct Doppler frequency bands. This configuration can enhance Doppler signal processing by increasing the signal-to-noise ratio (SNR) and reducing interference. A signal processing unit may extract Doppler signals separately at different frequencies, further improving radar system performance. An artificial intelligence-based classification system can process the extracted Doppler signals, enhancing object classification accuracy and velocity discrimination. The system may be applied in various domains, such as unmanned aerial vehicle (UAV) detection, where it can use multi-tonal Doppler spectrograms for real-time classification, and surface velocity radar for hydrology, where it can distinguish structured flow from turbulent flow. Additionally, the system can be utilized in security and surveillance applications to detect and differentiate human motion, vehicle movement, and intrusions, adapting to different motion-sensing applications.

[0021] FIG. 1 illustrates a signal processing unit 102, which may be integral to a multi-tonal frequency-modulated continuous-wave (FMCW) radar system 100. The signal processing unit 102 may process Doppler signals to extract information at different frequencies. This unit may be configured to extract Doppler signals separately at the different frequencies, which may increase the signal-to-noise ratio (SNR) and reduce interference. The extraction of Doppler signals may be achieved by driving a voltage-controlled oscillator (VCO) 106 with multiple simultaneous sine waves or by exploiting nonlinearities to generate additional harmonics. This process may create multiple distinct Doppler frequency bands, which may enhance the clarity of the Doppler signals. The signal processing unit 102 may also be involved in demodulating a received radar signal to obtain in-phase (I) and quadrature (Q) components at an intermediate frequency (IF). These components may be digitized using an analogue-to-digital converter (ADC), which may further aid in the processing of the Doppler signals. The signal processing unitl02 may be involved in demodulating a received radar signal to obtain in-phase (I) and quadrature (Q) components at an intermediate frequency (IF). These components may be digitised using an analogue-to-digital converter (ADC), which may further aid in the processing of the Doppler signals. The processed signals may then be used for improved object classification and velocity discrimination, potentially utilizing an artificial intelligence-based classification system 104. The Al-based system may process the extracted Doppler signals, leveraging the richer spectral data to improve classification accuracy. The signal processing unit 102 may thus play a role in enhancing the overall performance of the radar system by ensuring that the Doppler signals are processed efficiently and accurately.

[0022] The signal processing unit 102 may further isolate and exploit specific FM sideband orders in the Doppler return signal. Multi-tone frequency modulation inherently produces a series of spectral sidebands spaced by the modulation frequencies, with amplitudes governed by Bessel functions. The radar return thus contains multiple Doppler-shifted components (e.g., first-order, second-order) corresponding to different Bessel notations. By applying appropriate digital filtering or demodulation, the processor can extract Doppler signals associated with each modulation tone or sideband order. For example, if the VCO 106 is driven with a 5 kHz tone, the Doppler spectrum will exhibit energy at frequency offsets of ±5 kHz (first-order sidebands), ±10 kHz (second-order), etc., relative to the carrier. The signal processing unit 102 can separate these components by mixing the received I / Q stream with a reference at the modulation frequency of interest or by selecting spectral bands around the expected offsets. This yields multiple parallel Doppler channels, each corresponding to a distinct FM modulation order. This separation leverages the known FM spectral structure and ensures that motion-induced frequency shifts can be observed in several independent bands for improved detection reliability. The ability to extract Doppler signals from either side of each sideband order increases the total signal-to-noise ratio, removes the effect of data loss for low-Doppler targets caused by low-frequency filtering at baseband, and provides a richer feature set for classification.

[0023] In one embodiment, the signal processing unit 102 partitions the digitized I / Q data into overlapping time frames for time-frequency analysis. For example, the processor may segment the signal into frames of 0.5 seconds with 50% overlap between consecutive frames. Each frame is multiplied by a window function, such as a Hann window, to taper the edges and minimize spectral leakage. This framing and windowing prepares the data for computing a Doppler spectrogram. For each frame, the signal processing unit 102 computes a complex fast Fourier transform (CFFT) to generate a Doppler spectrum. Because the input to the CFFT is the complex baseband signal (I + jQ), the resulting spectrum inherently captures both positive and negative Doppler frequencies, corresponding to motion toward or away from the radar. The spectrogram is formed by assembling these spectra over time, producing a time-frequency representation of the Doppler shifts. In a multi-tonal scenario, a moving target produces multiple streaks or ridges in the spectrogram, representing the Doppler returns in the carrier and the various sideband orders.

[0024] The AI-Based Classification System, identified as component 104, may serve a role in the multi-tonal frequency-modulated continuous-wave (FMCW) radar system by classifying objects and enhancing detection accuracy through the application of artificial intelligence. This system may utilize advanced AI techniques to process the extracted Doppler signals, which are obtained separately at different frequencies. The AI-Based Classification System may be configured to improve object classification and velocity discrimination by leveraging the richer spectral data provided by the multi-tonal Doppler signals. The system may enhance classification accuracy by utilizing convolutional neural networks (CNN) and transformer models, which may be particularly effective in differentiating between various object types, such as distinguishing human motion from vehicle movement or identifying different types of unmanned aerial vehicles (UAVs).

[0025] The AI-Based Classification System may be integrated with the signal processing unit, which processes Doppler signals to extract information at different frequencies, thereby increasing the signal-to-noise ratio (SNR) and reducing interference. This integration may allow the AI system to utilize the enhanced Doppler signal clarity to perform more accurate classifications. The system may also be capable of adapting to different motion-sensing applications, such as security and surveillance, by detecting and differentiating human motion, vehicle movement, and intrusions. The AI-Based Classification System may further improve radar system performance by processing the Doppler signals in a manner that enhances the overall classification capability of the radar system.

[0026] In the context of the radar system, the AI-Based Classification System may be driven by multiple simultaneous sine waves or may exploit nonlinearities to generate additional harmonics, creating multiple distinct Doppler frequency bands. This approach may facilitate the extraction of Doppler signals separately at different frequencies, which may be crucial for improving object classification and velocity discrimination. The AI-Based Classification System may thus play a pivotal role in the radar system by providing higher classification accuracy and increasing the robustness of the system to environmental noise, thereby enhancing the overall effectiveness of the radar system in various applications.

[0027] The Voltage-Controlled Oscillator (VCO) component 106 may be integral to the multi-tonal frequency-modulated continuous-wave (FMCW) radar system. This component may generate multiple frequency bands essential for Doppler signal processing. The VCO may be driven by multiple simultaneous sine waves or may exploit nonlinearities to generate additional harmonics, thereby creating multiple distinct Doppler frequency bands. This process may enhance the signal-to-noise ratio (SNR) and reduce interference, which may be crucial for improved object classification and velocity discrimination. The VCO's ability to generate additional harmonics may be pivotal in creating these distinct frequency bands, which may be extracted separately to enhance the clarity of Doppler signals. The multi-tonal VCO drive generation may involve nonlinear excitation, where the VCO is driven by multiple sinusoidal waveforms, potentially introducing additional harmonics and creating distinct Doppler bands. This approach may significantly improve the radar system's performance by enhancing Doppler signal processing and increasing classification accuracy. The VCO's role in driving the radar transmission frequency with multiple sine waves may be fundamental in creating the necessary frequency bands for effective Doppler signal extraction and processing. This capability may be essential for the radar system's adaptability to various motion-sensing applications, including UAV detection, surface velocity radar for hydrology, and security surveillance. The VCO component 106 may thus be an element in the radar system, enabling enhanced signal processing and improved classification accuracy through its multi-tonal frequency modulation capabilities.

[0028] The UAV Detection Subsystem, identified as component 108, may be integral to the multi-tonal FMCW radar system, facilitating the detection and classification of unmanned aerial 7 vehicles (UAVs). This subsystem may utilize multi-tonal Doppler spectrograms to differentiate UAV types, potentially employing three independent frequency-modulated signals and machine learning-based spectral analysis. The UAV Detection Subsystem may be designed to classify UAVs in real-time, leveraging CNN and Transformer models to distinguish between various UAV types, such as quadcopters and fixed-wing drones. The subsystem may be driven by multiple simultaneous sine waves or may exploit nonlinearities to generate additional harmonics, thereby creating multiple distinct Doppler frequency bands. These frequency bands may enhance the signal-to-noise ratio (SNR) and reduce interference, which may be crucial for improving object classification and velocity discrimination. The UAV Detection Subsystem may process the extracted Doppler signals separately at different frequencies, which may further enhance the classification accuracy by providing richer spectral data. This approach may allow the system to detect and classify UAVs more effectively, even in cluttered environments where traditional single-tone FMCW radars may struggle. The UAV Detection Subsystem may be a component in applications requiring precise UAV detection and classification, contributing to the overall performance and adaptability of the radar system in various motion-sensing applications.

[0029] The Surface Velocity Radar Subsystem, identified as component 110, may enhance river flow velocity measurements by distinguishing between structured and turbulent flow types. This subsystem may utilize multi-tonal frequency-modulated continuous-wave (FMCW) signals to improve the accuracy of fluid flow measurements. The Doppler signals may be extracted separately at different frequencies, which may increase the signal-to-noise ratio (SNR) and reduce interference, thereby enhancing the clarity of the Doppler signal. The subsystem may be driven by multiple simultaneous sine waves or may exploit nonlinearities to generate additional harmonics, creating multiple distinct Doppler frequency bands. These bands may be processed to improve object classification and velocity discrimination. The Doppler signal processing may involve demodulating the received radar signal to obtain in-phase (I) and quadrature (Q) components at an intermediate frequency (IF), which may then be digitized using an analog-to-digital converter (ADC). The processed Doppler signals may be used to detect and classify the motion of targets within the radar's field of view. The system may leverage richer spectral data to improve classification accuracy, thereby enhancing the performance of the radar system in hydrology applications. The Surface Velocity Radar Subsystem may thus provide a solution for distinguishing structured flow from turbulent flow, contributing to more precise river flow velocity measurements.

[0030] The Security and Surveillance Subsystem, identified as component 112, may be integral to the multi-tonal frequency-modulated continuous-wave (FMCW) radar system. This subsystem may detect and differentiate human motion, vehicle movement, and intrusions, potentially enhancing security and surveillance capabilities. The subsystem may utilize an Al-based classification system to improve detection accuracy, which may be achieved by processing Doppler signals extracted at different frequencies. The Al-based classification system may be configured to differentiate between walking and running, as well as between humans and vehicles, thereby adapting to various motion-sensing applications. The subsystem may leverage the enhanced Doppler signal processing capabilities of the radar system, which may be driven by multiple simultaneous sine waves or may exploit nonlinearities to generate additional harmonics. This approach may create multiple distinct Doppler frequency bands, which may increase the signal-to-noise ratio (SNR) and reduce interference, thereby improving object classification and velocity discrimination. The subsystem may also benefit from the adaptability of the radar system to different motion-sensing applications, such as industrial monitoring and UAV surveillance, by utilizing the Al-based classification to detect and classify motion effectively. The integration of these technologies may allow the Security and Surveillance Subsystem to provide robust and reliable motion detection and classification, enhancing the overall performance of the radar system in security and surveillance contexts.

[0031] FIG. 2 is a flowchart illustrating a method in step 200 for driving a voltage-controlled oscillator (VCO) with multiple simultaneous sine waves or exploiting nonlinearities to generate additional harmonics, thereby creating multiple distinct Doppler frequency bands, according to an embodiment. At step 200, the VCO may be driven by multiple simultaneous sine waves, which can create multiple distinct Doppler frequency bands. Alternatively, the nonlinearities of the VCO may be exploited to generate additional harmonics, contributing to the creation of these distinct frequency bands. This process may enhance the radar system's ability to detect and classify motion by providing a richer set of Doppler signals. The generation of multiple distinct Doppler frequency bands may increase the signal-to-noise ratio (SNR) and reduce interference, thereby improving the clarity of the Doppler signals. The use of multiple simultaneous sine waves or the exploitation of nonlinearities may allow for the creation of additional harmonics, which can further enhance the radar system's performance in motion detection and classification. The VCO's ability to be driven by these multiple sine waves or to exploit nonlinearities may be crucial in achieving the desired Doppler frequency bands, which can significantly improve the radar system's object classification and velocity discrimination capabilities.

[0032] At step 202, the signal processing unit may be employed to extract Doppler signals separately at different frequencies. This extraction can potentially enhance the signal-to-noise ratio (SNR) and reduce interference, which may be crucial for the effective functioning of the radar system. The process may involve the demodulation of a received radar signal to obtain in-phase (I) and quadrature (Q) components at an intermediate frequency (IF). These components may then be digitized using an analog-to-digital converter (ADC). The separation of Doppler signals at various frequencies may allow for a clearer distinction between different motion signatures, which can be particularly beneficial in cluttered environments where traditional single-frequency systems might struggle. The enhanced SNR achieved through this method may improve the clarity of Doppler signals, thereby facilitating more accurate motion detection and classification. This step may serve as a foundational process in the radar system, enabling subsequent steps to leverage the improved data quality for further analysis and classification.

[0033] In the context of the multi-tonal frequency-modulated continuous-wave (FMCW) radar system, step 204 may involve the processing of extracted Doppler signals using an artificial intelligence (Al)-based classification system. This step may be crucial for improving object classification and velocity discrimination. The Al-based classification system may be configured to handle the Doppler signals that have been extracted separately at different frequencies, potentially enhancing the signal-to-noise ratio (SNR) and reducing interference. The Al-based classification system may utilize richer spectral data to improve classification accuracy, which may be particularly beneficial in distinguishing between various types of motion or objects.

[0034] The processing of Doppler signals may involve several sub-processes. For instance, the system may differentiate unmanned aerial vehicle (UAV) types using multi-tonal Doppler spectrograms. This may be achieved through the use of spectrograms and models for classification, potentially allowing for real-time UAV classification. The system may also enhance river flow velocity measurements by distinguishing structured flow from turbulent flow using the multiple distinct Doppler frequency bands. This may improve fluid flow measurement accuracy and enhance hydrology applications. Additionally, the system may detect and differentiate human motion, vehicle movement, and intrusions using the Al-based classification system. This capability may be applied to various motion-sensing applications, improving security and surveillance by adapting to different scenarios.

[0034] The Al-based classification system may be an integral part of the radar system, leveraging advanced machine learning techniques such as convolutional neural networks (CNN) and transformer models to process the Doppler signals. This approach may allow the system to classify objects and motions with higher accuracy, even in challenging environments with significant background noise or clutter. The use of multiple simultaneous sine waves or the exploitation of nonlinearities to generate additional harmonics may create multiple distinct Doppler frequency bands, which may be essential for the Al-based classification system to function effectively. Overall, step 204 may represent a component of the radar system, enabling enhanced motion detection and classification capabilities.

[0035] FIG. 3 is a flowchart illustrating a detailed signal processing pipeline for generating a Doppler spectrogram, according to an embodiment. This process may be carried out by the signal processing unit 102. The process begins at step 300, where the received radar signal is demodulated to obtain its in-phase (I) and quadrature (Q) components at an intermediate frequency. At step 302, these analogue I and Q components are digitised by an analogue-to-digital converter (ADC), creating a digital I / Q data stream. At step 304, the signal processing unit partitions this data stream into a plurality of overlapping time frames. This segmentation is foundational for conducting time-frequency analysis, allowing the system to observe how the Doppler signature evolves over short time intervals. Next, at step 306, each time frame is multiplied by a window function, such as a Hann window. This step is crucial for minimizing spectral leakage by tapering the edges of each frame, which improves the accuracy of the subsequent frequency analysis. At step 308, the signal processing unit computes a complex fast Fourier transform (CFFT) on each windowed frame. Because the input is the complex I+jQ signal, the resulting Doppler spectrum inherently captures both positive and negative Doppler frequencies, corresponding to motion toward and away from the radar, respectively. The process continues at step 310, where the sequence of Doppler spectra generated from each frame is assembled over time. This assembly forms a time-frequency Doppler spectrogram, which provides a rich visualization of the target's motion signatures. Finally, at step 312, the completed Doppler spectrogram is output for further processing by the Al-based classification system 104, which leverages this enhanced data for improved object classification and velocity discrimination.

Claims

1. A multi-tonal frequency-modulated continuous-wave (FMCW) radar system for improved motion detection and classification, the system comprising:- a voltage-controlled oscillator (VCO) configured to vary a radar transmission frequency, and - a signal processing unit configured to extract Doppler signals at different frequencies, characterized in that:- the VCO is driven by multiple simultaneous sine waves or exploits nonlinearities to generate additional harmonics, thereby creating multiple distinct Doppler frequency bands; and- the signal processing unit is configured to extract the Doppler signals separately at the different frequencies to increase signal-to-noise ratio (SNR) and reduce interference.

2. The system according to claim 1, wherein an artificial intelligence (Al)-based classification system is configured to process the extracted Doppler signals for improved object classification and velocity discrimination.

3. The system according to either claim 1 or claim 2, wherein the multiple simultaneous sine waves comprise at least two sine waves with frequencies selected from a group consisting of 5 kHz, 11 kHz, and 19 kHz.

4. The system according to any one of the preceding claims, wherein the signal processing unit is further configured to demodulate a received radar signal to obtain in-phase (I) and quadrature (Q) components at an intermediate frequency (IF).

5. The system according to claim 4, wherein the I and Q components are digitized using an analogue-to-digital converter (ADC).

6. The system according to claim 5, wherein the signal processing unit is configured to partition the digitised I and Q components into a plurality of overlapping time frames for time-frequency analysis.

7. The system according to claim 6, wherein each time frame is multiplied by a window functionto taper edges and minimize spectral leakage.

8. The system according to claim 7, wherein the window function is a Hann window.

9. The system according to any one of claims 6 to 8, wherein the signal processing unit is configured to compute a complex fast Fourier transform (CFFT) for each time frame to generate a Doppler spectrum, wherein the Doppler spectrum captures both positive and negative Doppler frequencies.

10. The system according to claim 9, wherein the signal processing unit is configured to assemble the Doppler spectra over time to form a Doppler spectrogram representing a timefrequency representation of Doppler shifts.

11. The system according to claim 1, wherein the distinct Doppler frequency bands comprise a series of spectral sidebands having amplitudes governed by Bessel functions, and wherein the signal processing unit is configured to extract Doppler signals associated with a specific sideband order.

12. The system according to any one of claim 2 and a combination of claim 2 with any of the other preceding claims, wherein the Al-based classification system comprises at least one of a convolutional neural network (CNN) and a transformer model.

13. The system according to claim 1, further comprising an unmanned aerial vehicle (UAV) detection subsystem configured to differentiate UAV types using multi-tonal Doppler spectrograms.

14. The system according to claim 13, wherein the UAV types comprise at least one of quadcopters and fixed-wing drones.

15. The system according to any one of the preceding claims, further comprising a surface velocity radar subsystem configured to enhance river flow velocity measurements by distinguishing structured flow from turbulent flow using the multiple distinct Doppler frequencybands.

16. The system according to any one of the preceding claims, further comprising a security and surveillance subsystem configured to detect and differentiate human motion, vehicle movement, and intrusions using the Al-based classification system.

17. The system according to claim 16, wherein the security and surveillance subsystem is further configured to differentiate between walking and running, and between humans and vehicles.

18. A method for improved motion detection and classification using a multi-tonal frequency-modulated continuous-wave (FMCW) radar system, the method comprising:- driving a voltage-controlled oscillator (VCO) with multiple simultaneous sine waves or exploiting nonlinearities to generate additional harmonics, thereby creating multiple distinct Doppler frequency bands, and- extracting Doppler signals separately at different frequencies to increase signal-to-noise ratio (SNR) and reduce interference.

19. The method according to claim 18, further comprising the step of processing the extracted Doppler signals using an artificial intelligence (Al)-based classification system for improved object classification and velocity discrimination.

20. The method according to either claim 18 or claim 19, further comprising demodulating a received radar signal to obtain in-phase (I) and quadrature (Q) components at an intermediate frequency (IF) and digitizing the I and Q components using an analogue-to-digital converter (ADC).

21. The method according to claim 20, further comprising:partitioning the digitised I and Q components into a plurality of overlapping time frames;applying a window function to each time frame; andcomputing a complex fast Fourier transform (CFFT) for each time frame to generate a Doppler spectrum.

22. The method according to any one of claims 18 to 21, further comprising at least one of:- differentiating unmanned aerial vehicle (UAV) types using multi-tonal Doppler spectrograms, - enhancing river flow velocity measurements by distinguishing structured flow from turbulent flow using the multiple distinct Doppler frequency bands, and- detecting and differentiating human motion, vehicle movement, and intrusions using the AI-based classification system.T +44(0)30 0300 2000A

Images