Broadband spectrum sensing of radio-frequency (RF) signals and high-resolution radar detection are widely applied in cognitive radio systems,1^–3 intelligent transportation systems,3^,4 and electronic warfare systems.5^–7 With the continuous and rapid development of these systems, the demand for spectrum sensing and radar functions has gradually shifted from relying on two separate systems to the desire for an integrated system that can achieve both functions, which can greatly facilitate the integration and miniaturization of application platforms.

To fulfill this urgent need, joint radar and spectrum sensing systems have already been investigated in the electrical domain. In joint radar and spectrum sensing systems, the two functions are no longer independent but highly integrated in terms of system architecture and resources. They share some parts of system structure and hardware and achieve signal multiplexing or sharing on the signal level. However, when aiming to achieve both high-resolution radar detection and broadband spectrum sensing, the integration of radar and spectrum sensing systems via conventional electronic methods encounters significant challenges. Chiefly, these difficulties stem from the “electronic bottleneck,” which imposes limitations on the operating bandwidth, especially when striving for simultaneous wide tunability and low system complexity. Microwave photonics8^,9 uses photonic technology to generate, transmit, and process the RF signal, which can break through the electronic bottleneck of conventional electronic methods and have the distinguished advantage of low transmission loss, good tunability, wide bandwidth, and immunity to electromagnetic interference. In the past few years, microwave photonic radar,10^–12 spectrum sensing,13 and integrated multifunction RF systems14^,15 have garnered extensive research attention from both academic and industrial communities.

Microwave photonics possesses the capability of enhancing the key performance indicators of individual radar or spectrum sensing systems. With the assistance of microwave photonics, the challenges of generating and receiving ultrawideband wide-range frequency-tunable radar signals are addressed, which greatly improves the flexibility and anti-interference ability of radar systems.16 The classical photonics-assisted radar signal generation methods can be implemented by photonic frequency conversion,17^,18 photonic frequency multiplication,19^,20 optical injection,21 photonic digital-to-analog converters,22 and acoustic-optic frequency-shifting loops.23 For wideband radar signal reception, microwave photonic channelization,24 photonic dechirping,17^–20 and photonic compressed sensing25 can be applied. Microwave photonics can also be applied to spectrum sensing systems, where compared with conventional electronics solutions, its main contribution is the ability to quickly acquire frequency information within an ultrawide frequency measurement range. Although its frequency accuracy may be inferior to electronics solutions, the rapid acquisition of frequency information is crucial in certain application scenarios. To acquire the frequency information of the signal under test (SUT), microwave frequency measurement26^–30 and time–frequency analysis31^–35 can be employed. The basic principle is to map the frequency information of the SUT to other optical or electrical parameters that are easier to measure, which can be implemented via frequency-to-power mapping (FTPM),26^,27 frequency-to-space mapping,27^,28 and frequency-to-time mapping (FTTM).29^–35 In general, the method based on FTPM is only applicable for single-frequency measurement unless it is combined with other methods.27 The method based on FTPM should also be used in conjunction with other methods;27 otherwise, the measurement resolution is poor unless there are a large number of channels at the cost of great complexity.28 The method based on FTTM can achieve multifrequency measurement with suitable system complexity and good measurement accuracy. In addition, this kind of method has good scalability and can be further extended to analyze the two-dimensional time–frequency parameter.31^–35

In recent years, microwave photonic systems that deeply integrate radar and spectrum sensing functions have been studied. It is hoped that with an integrated system, both these RF functions can be achieved simultaneously by sharing part of the signal and the hardware. In these proposed schemes,36^–39 different photonic methods are employed to generate the linearly frequency-modulated (LFM) signals and to implement the dechirping operation for radar function, whereas the SUT is optically converted to an electrical frequency-sweep signal and its frequency is mapped to the time domain using a narrow-bandwidth electrical bandpass filter (EBPF) for spectrum sensing function. In Refs. 36 and 37, tunable frequency-multiplied LFM signal generation, radar echo dechirping, and modulation and measurement of the SUT are realized through optical sideband manipulation. However, the system structure is relatively complex, requiring a complex modulator structure and optical bandpass filter (OBPF) for optical sideband generation and selection, which limits the tunability of the system to a certain extent. In Ref. 38, the two functions do not share the same frequency-sweep source, which not only reduces the interdependence of the two functions but also creates some redundancy in the system. In Ref. 39, optical injection of a semiconductor laser is employed to generate the frequency-sweep source for both radar and spectrum sensing functions, which reduces the cost and complexity of the system. However, due to the limited performance of the frequency-sweep optical signal generated by optical injection, the frequency measurement error and resolution in Ref. 39 are the worst among these schemes. Furthermore, because all these schemes36^–39 realize the FTTM in the electrical domain, in the spectrum sensing function, high-speed photodetectors (PDs) are needed before the EBPF to generate the electrical frequency-sweep signal carrying the frequency information of the SUT, which increases the cost of the system. Moreover, all these proposed schemes are only studied and demonstrated in an experiment, and each of them has realized only a part of the many functions of the joint radar and spectrum sensing system, including ranging, radial velocity measurement, imaging, frequency measurement, and time–frequency analysis. In addition, the operating frequency range and tunability of the systems also need to be further improved.

In this work, to the best of our knowledge, the first microwave photonic prototype capable of simultaneously performing radar detection and spectrum sensing is proposed, designed, built, and investigated, with its radar and spectrum sensing functions being independent and tunable within the ranges of 8 to 40 GHz and 0 to 40 GHz, respectively. An intermediate frequency (IF) LFM signal with a bandwidth of 1 GHz is generated by a direct digital synthesizer (DDS) working in conjunction with an analog electronic circuit. The IF LFM signal is converted to the optical domain via an intensity modulator (IM) and then passes through a fiber Bragg grating (FBG) that suppresses the optical carrier, resulting in the $\pm 2\mathrm{nd}$-order LFM optical sidebands. The $\pm 2\mathrm{nd}$-order LFM optical sidebands beat with each other to generate the frequency-and-bandwidth-quadrupled LFM signal for radar applications, whereas the $-2\mathrm{nd}$-order LFM optical sideband is selected by another FBG and modulated by the SUT. The optical signal carrying the SUT then interacts with a stimulated Brillouin scattering (SBS) gain spectrum to map the frequency of the SUT to the time domain for the spectrum sensing function. By tuning the frequency of the IF LFM signal, the radar function can be operated with a 4-GHz instantaneous bandwidth in a frequency range from 8 to 40 GHz. The spectrum sensing range can be extended from 0 to 40 GHz by adjusting the pump wave frequency, and the instantaneous analysis bandwidth is 2 GHz. The prototype is comprehensively studied and tested, which is capable of achieving a range resolution of 3.75 cm, a range error of less than $\pm 2\mathrm{cm}$, a radial velocity error within $\pm 1\mathrm{cm}/\mathrm{s}$, delivering clear imaging of multiple small targets, and maintaining a frequency measurement error of less than $\pm 7\mathrm{MHz}$ and a frequency resolution of better than 20 MHz.

The system architecture of the microwave photonic prototype for concurrent radar detection and spectrum sensing is depicted in Fig. 1. The prototype directly uses a 220 V AC power supply. We have developed control and data acquisition (DAQ) software using C++, as well as data processing software using MATLAB. Its control, data transmission, and processing are achieved using a computer. According to the functions implemented, the prototype is divided into four layers: Layer 1 is used to generate the optical LFM signal for the radar and spectrum sensing functions; Layer 2 is used for radar signal generation and detection of the radar function, the SUT reception of the spectrum sensing function, and the DAQ function; Layer 3 is used for frequency information identification of the spectrum sensing function; Layer 4 integrates the power supply and command receiving and forwarding function. The rear panel of the prototype provides three interfaces, including a main power switch, a grounding port, and an optical monitoring port for observing the optical spectrum from an erbium-doped fiber amplifier (EDFA1) via an optical switch in Layer 1. The front panel of the prototype has four available interfaces, including one universal serial bus (USB) port, one local area network (LAN) port, and two power switches. The USB port is used to connect the computer and the IF LFM signal generator located in Layer 1, as well as the microwave signal generator (MSG) situated in Layer 3, through a USB data-forwarding board in Layer 4. The LAN port is used to connect the computer and the DAQ board on Layer 2 for data transmission and DAQ control. The two power switches are connected to the sequential power supply board on Layer 4 to sequentially power up all the components that need to be powered in the prototype. All control commands are set and generated using the control and DAQ software and then forwarded to the IF LFM signal generator, the MSG, and the DAQ board. The data acquired by the DAQ board are transmitted back to the computer and stored locally as a file, where they are further processed in the data processing software to get the radar detection and spectrum sensing results. In the process of DAQ, the DAQ board will output a trigger signal to the IF LFM signal generator for synchronization. Therefore, the reference signal used for absolute frequency positioning34 can be removed from the prototype.

In Layer 1 of the prototype, a narrow-linewidth laser diode (LD, Photonteck PHX-C-F-M-C34-13-6-1-0-0) is used to generate a continuous-wave light wave with a power of 13 dBm and a wavelength of 1549.919 nm, and the corresponding frequency is denoted as $f_0$. Thus, the optical signal from the LD can be expressed as $\exp (\mathrm{j}2\pi f_{0}t)$. It is divided into two parts by an optical coupler (OC1). The 10% output of OC1 is modulated by an IF LFM signal at IM1 (Fujitsu FTM7938EZ). The IF LFM signal has a power of around 15 dBm, a period $T$ of 1.4876 ms, a bandwidth of 1 GHz, and an adjustable center frequency $f_{\text{C}}$ from 2.5 to 9.5 GHz, which can be expressed as $$V_{\mathrm{IF}}(t)=V_1\cos (2\pi f_{\mathrm{s}}t+\pi k{t}^2),t\in [0,T),$$(1)where $V_1$ and $k$ are the amplitude and chirp rate of the IF LFM signal and $f_{\text{s}}=f_{\text{C}}-kT/2$ is the start frequency of the IF LFM signal. IM1 is biased and stabilized at the maximum transmission point (MATP) using a modulator bias controller (MBC1, PlugTech MBC-NULL-03). Under the small signal modulation condition, the optical signal from IM1 can be expressed as $$\begin{gathered} E_{\mathrm{IM}1}(t)\propto \frac{1}{2}\exp (\mathrm{j}2\pi f_{0}t)[\exp (\mathrm{j}\pi V_{\mathrm{IF}}(t)/V_{\pi })+\exp (-\mathrm{j}\pi V_{\mathrm{IF}}(t)/V_{\pi })] \\ =\exp (\mathrm{j}2\pi f_{0}t)\cos [m_1\cos (2\pi f_{\mathrm{s}}t+\pi k{t}^2)] \\ \approx \exp (\mathrm{j}2\pi f_{0}t)[J_0(m_1)-J_2(m_1)\exp (-\mathrm{j}4\pi f_{\mathrm{s}}t-\mathrm{j}2\pi k{t}^2) \\ -J_2(m_1)\exp (\mathrm{j}4\pi f_{\mathrm{s}}t+\mathrm{j}2\pi k{t}^2)], \end{gathered}$$(2)where $J_{\mathrm{n}}(·)$ is the $n$’th-order Bessel function of the first kind and $m_1=\pi V_1/V_{\pi }$ is the modulation index. As can be seen, only the optical carrier and $\pm 2\mathrm{nd}$-order optical sidebands are generated at the output of IM1 that is biased at the MATP. The optical signal output from IM1 is injected into FBG1. The temperature of FBG1 is controlled by a temperature controller with a temperature stability of 0.5°C. The transmission spectrum of FBG1 is used as a narrow-bandwidth optical bandstop filter (OBSF) to suppress the optical carrier and leave only the $\pm 2\mathrm{nd}$-order optical sidebands of the IF LFM signal. The $\pm 2\mathrm{nd}$-order optical sidebands of the IF LFM signal from FBG1 can be expressed as $$\begin{gathered} E_{\mathrm{FBG}1}(t)\propto -J_2(m_1)\exp (\mathrm{j}2\pi f_{0}t)[\exp (-\mathrm{j}4\pi f_{\mathrm{s}}t-\mathrm{j}2\pi k{t}^2) \\ +\exp (\mathrm{j}4\pi f_{\mathrm{s}}t+\mathrm{j}2\pi k{t}^2)]. \end{gathered}$$(3)

The obtained $\pm 2\mathrm{nd}$-order optical sidebands are amplified by EDFA1 (Max-Ray EDFA-C-PA-45-SM-M) and serve as the optical LFM signal that is used for the following radar and spectrum sensing functions. In the radar function, the optical LFM signal is used for radar signal generation and as an optical reference signal for radar dechirping. In the spectrum sensing function, the optical LFM signal is used as a frequency-sweep optical signal for further signal modulation and FTTM.

The optical LFM signal from Layer 1 is sent to OC2 via an optical switch. Note that the other output port of the optical switch is connected to the rear panel for monitoring. When power switch 1 is on and power switch 2 is off, the input optical LFM signal is sent to the monitor port on the rear panel. As power switch 2 is on, the optical LFM signal is sent to OC2. The two outputs of OC2 are sent to OC3 and IM2 (Fujitsu FTM7938EZ), respectively. The 10% output of OC3 is injected into PD1 (Coherent XPDV2120R), in which the two $\pm 2\mathrm{nd}$-order optical sidebands of the optical LFM signal beat with each other to generate a frequency-and-bandwidth-quadrupled LFM signal, i.e., the radar signal. The electrical signal from PD1 can be expressed as $$\begin{gathered} i_{\mathrm{PD}1}(t)\propto |E_{\mathrm{FBG}1}(t){|}^2 \\ \propto J_2^2(m_1)[1+\cos (8\pi f_{\mathrm{s}}t+4\pi k{t}^2)]. \end{gathered}$$(4)

Therefore, the frequency-and-bandwidth-quadrupled LFM radar signal can be expressed as $$V_T(t)\propto \cos (8\pi f_{\mathrm{s}}t+4\pi k{t}^2).$$(5)

The bandwidth of the generated radar signal is 4 GHz, whereas its center frequency is adjustable from 10 to 38 GHz. Subsequently, the radar signal is filtered by the EBPF (Talent Microwave TLHF-8G-40G-X), amplified by a low-noise amplifier (LNA1, Talent Microwave TLLA1G40G-40-45) and a power amplifier (PA1, Connphy Microwave CMP-0.1G40G-3020-K), and finally radiated to the free space through a transmitting antenna (Tx). The radar echo from the targets is collected by a receiving antenna (Rx1), amplified by LNA2 (Talent Microwave, TLLA1G40G-40-45), and then applied to the RF port of IM2. The radar echo signal is expressed as $$V_{\mathrm{E}}(t)=V_2\cos [8\pi f_{\mathrm{s}}(t-\tau )+4\pi k(t-\tau {)}^2],$$(6)where $\tau$ and $V_2$ are the time delay and amplitude of the radar echo signal. IM2 is biased and stabilized at the quadrature transmission point using MBC2 (PlugTech MBC-MZM-01) to enable the mixing of the optical reference signal and the radar echo signal. Under the small-signal modulation condition and considering only the first-order optical sidebands, the optical signal from IM2 can be expressed as $$\begin{gathered} E_{\mathrm{IM}2}(t)\propto \frac{1}{2}{E}_{\mathrm{FBG}1}(t)\{\exp [\mathrm{j}\pi /4+\mathrm{j}\pi V_{\mathrm{E}}(t)/V_{\pi }]+\exp [-\mathrm{j}\pi /4-\mathrm{j}\pi V_{\mathrm{E}}(t)/V_{\pi }]\} \\ \approx \frac{\sqrt{2}}{2}{E}_{\mathrm{FBG}1}(t)\{J_0(m_2)-2J_1(m_2)\cos [8\pi f_{\mathrm{s}}(t-\tau )+4\pi k(t-\tau {)}^2]\}, \end{gathered}$$(7)where $m_2=\pi V_2/V_{\pi }$ is the modulation index. After mixing in IM2, the optical signal from IM2 is detected in PD2 (Nortel PP-10G) to realize the radar dechirping. The electrical signal from PD2 is expressed as $$\begin{gathered} i_{\mathrm{PD}2}(t)\propto |E_{\mathrm{IM}2}(t){|}^2 \\ \propto \frac{1}{2}[1+\cos (8\pi f_{\mathrm{s}}t+4\pi k{t}^2)]\times \{J_0^2(m_2)+2J_1^2(m_2) \\ -4J_0(m_2)J_1(m_2)\cos [8\pi f_{\mathrm{s}}(t-\tau )+4\pi k(t-\tau {)}^2] \\ +2J_1^2(m_2)\cos [16\pi f_{\mathrm{s}}(t-\tau )+8\pi k(t-\tau {)}^2]\}. \end{gathered}$$(8)

The low-frequency component from PD2 is the dechirped signal, which can be expressed as $$V_{\mathrm{D}}(t)=\cos (8\pi kt\tau +8\pi f_{\mathrm{s}}\tau -4\pi k{\tau }^2).$$(9)

The frequency of the dechirped signal is $$f_{\mathrm{D}}=4k\tau \mathrm{.}$$(10)The dechirped signal is sampled by the DAQ board and sent to the computer for further processing.

The radar dechirped signals with different durations are utilized to extract the range, radial velocity, and inverse synthetic aperture radar (ISAR) imaging. The target range can be accurately determined by applying a fast Fourier transform (FFT) to a single-period radar dechirped signal via the following equation: $$L=\frac{1}{2}c\tau =\frac{c}{8k}{f}_{\mathrm{D}},$$(11)where $c$ is the velocity of light in a vacuum. Meanwhile, the radial velocity of the target is calculated by measuring the range variation over a 0.5-s interval. For ISAR imaging, the range-Doppler algorithm is employed.17 The ISAR imaging range and the cross-range resolution can be expressed as17$$R_{\mathrm{R}}=\frac{c}{2f_{\mathrm{B}}},$$(12)$$R_{\mathrm{C}}=\frac{c}{8f_{\mathrm{C}}{T}_{\mathrm{i}}\omega },$$(13)where $f_{\mathrm{B}}=4kT$, $T_{\mathrm{i}}$, and $\omega$ are the radar signal bandwidth, the sampling time, and the rotating speed of the imaging targets, respectively.

The optical LFM signal from the 90% output of OC3 is sent to FBG2 via a circulator (CIR1). FBG2 is also thermostatically controlled by another temperature controller. The reflection spectrum of FBG2 functions as an OBPF. The center wavelength and 3-dB bandwidth of FBG2 are 1550.015 nm and 18 GHz. After being reflected by FBG2, the $-2\mathrm{nd}$-order optical sideband with a frequency of $f_0-(2f_{\text{s}}+2kt)$ is selected and then sent to IM3 (Fujitsu FTM7938EZ). The SUT collected by another receiving antenna (Rx2) is amplified by LNA3 (Talent Microwave TLLA1G40G-40-45) and PA2 (Centellax, OA4SMM5) and applied to the RF port of IM3. IM3 is biased and stabilized at the minimum transmission point by MBC3 (PlugTech MBC-NULL-03) to maximize the efficiency of converting the SUT to the optical domain. After modulation, the SUT is converted to a series of frequency-sweep optical sidebands. The output of IM3 is used as a probe wave and sent to a 200-m highly nonlinear fiber (HNLF, YOFC NL1016-B) via an optical isolator.

The 90% output of OC1 is injected into a dual-parallel Mach–Zehnder modulator (DP-MZM, Fujitsu FTM7961EX). A single-tone RF signal with a tunable frequency $f_{\text{t}}$ from 0.1 to 40 GHz is generated by an MSG and applied to the RF ports of the DP-MZM via a 90 deg hybrid coupler (90 deg HYB, Talent Microwave TBG-20400-3k-90) and PA3 (Centellax OA4SMM4). The DP-MZM is biased as a carrier-suppressed single-sideband modulator using MBC4 (PlugTech MBC-IQ-03), shifting the optical carrier applied to the DP-MZM according to the frequency of the RF signal. Then, the optical signal from the DP-MZM is used as a pump wave and reversely sent to the HNLF through EDFA2 (Max-Ray, EDFA-C-PA-45-SM-M), EDFA3 (Max-Ray, EYDFA-C-HP-BA-35-SM-M), and CIR2 to generate an SBS gain. The SBS gain spectrum acts as a narrow-bandwidth optical filter, besides providing a beneficial gain. Because different frequency-sweep optical sidebands generated by different SUT frequencies in the probe wave are filtered by the SBS gain spectrum at different times, optical pulses will be generated at different times in a single sweep period according to the SUT frequency. Therefore, the frequency of the SUT is mapped to the time domain when the optical pulse appears in a sweep period via FTTM. Note that in the FTTM process, by properly designing the relative positions of the optical sidebands and the pump wave, the $+1\mathrm{st}$-order optical sideband of the SUT interacts with the SBS gain spectrum, whereas the $-1\mathrm{st}$-order optical sideband does not. Finally, the optical pulses from port 3 of CIR2 are injected into PD3 (Nortel PP-10G) and converted to low-speed electrical pulses. The electrical pulses from PD3 are sampled by the DAQ board and further processed in the computer for frequency measurement and time–frequency analysis.

The instantaneous frequency analysis bandwidth is 2 GHz, which is determined by the frequency-sweep bandwidth of the optical LFM signal. Nevertheless, by changing the frequency $f_{\text{t}}$ applied to the DP-MZM, i.e., changing the pump wave frequency by $f_{\text{a}}=\pm f_{\text{t}}$ according to the direction of the frequency shifting, the 2-GHz instantaneous frequency analysis bandwidth can be located at any location from 0 to 40 GHz. When the pump frequency is not shifted, i.e., $f_{\text{a}}=0$, the center frequency of the SBS gain is $f_0-f_{\mathrm{SBS}}$, where $f_{\mathrm{SBS}}$ is the Brillouin frequency shift and is 9.4 GHz. When the pump frequency is shifted by $f_{\text{a}}=\pm f_{\text{t}}$, the center frequency of the SBS gain is also shifted to $f_0+f_{\text{a}}-f_{\mathrm{SBS}}$. Generally, the frequency measurement range is from $f_{\text{a}}-f_{\mathrm{SBS}}+2f_{\text{C}}-kT$ to $f_{\text{a}}-f_{\mathrm{SBS}}+2f_{\text{C}}+kT$. Here, $f_{\text{a}}-f_{\mathrm{SBS}}+2f_{\text{C}}-kT$ should be greater than 0. In this work, if the nonlinear medium is not changed and the sweep bandwidth of the IF LFM signal is also kept unchanged, the instantaneous frequency analysis range is from $f_{\text{a}}+2f_{\text{C}}-10.4\mathrm{GHz}$ to $f_{\text{a}}+2f_{\text{C}}-8.4\mathrm{GHz}$.

First, the generation of the optical LFM signal and radar signal is demonstrated. FBG1 has a center wavelength of 1549.919 nm and a 3-dB bandwidth of 10 GHz, and its transmission spectrum is equivalent to an OBSF, as indicated by the black dotted line in Fig. 2. The optical LFM signal from FBG1 is amplified and then monitored by an optical spectrum analyzer (OSA, ANDO AQ6317B) via the monitoring port. Figure 2 shows the optical spectra of the optical LFM signal under different center frequencies of the IF LFM signal from 2.5 to 9.5 GHz. As can be seen, the $\pm 2\mathrm{nd}$-order optical sidebands are dominant with a suppressed optical carrier. The peak power of the LFM sidebands is almost the same because the IF LFM signal power is preset and adjusted at different frequencies according to the response of the modulator. However, the carrier suppression ratios for different IF LFM signals are slightly different, which changes from 9.4 to 19.2 dB. The main reason for the different carrier suppression ratios is that the power of the IF LFM signal is different at different frequencies, resulting in the carrier power of FBG1 input also being different.

The optical LFM signal generated above is detected in the high-speed PD1 to generate the LFM radar signal. The electrical spectra of the LFM radar signal from PA1 are measured by an electrical spectrum analyzer (R&S FSP-40) with a resolution bandwidth of 3 MHz, a video bandwidth of 10 MHz, and a sweep time of 1 s, as shown in Fig. 3. When the center frequency of the IF LFM signal is adjusted from 2.5 to 9.5 GHz with a step of 0.5 GHz, the center frequency of the generated LFM radar signal changes from 10 to 38 GHz with a step of 2 GHz and the bandwidth of the LFM radar signal is quadrupled to 4 GHz.

As can be seen from Fig. 3, as the frequency of the LFM radar signal increases, its power decreases, which is mainly attributed to the frequency response of LNA1, PA1, and PD1. In building the prototype, we do not compensate for the power inconsistencies for two main reasons: (1) it is very difficult to further increase the power of the IF LFM signal at high frequencies and (2) we do not have a programmable automatic gain amplifier at such a large bandwidth. Furthermore, a certain gain unevenness is observed in the electrical spectrum, which is largely introduced by standing waves in the measurement process. In addition, it can be clearly seen that the generated LFM radar signal centered at 10 GHz has lower power from 8 to 9 GHz, which is mainly caused by the 10-GHz bandwidth of FBG1. In addition to filtering out the optical carrier, FBG1 also suppresses part of the optical sidebands to a certain extent when the sidebands are close to the carrier. This issue can be solved by using an FBG with a smaller 3-dB bandwidth instead of FBG1.

Figure 4(a) shows the setup of the ranging experiment using the prototype; a zoomed-in view of the prototype is shown in Fig. 4(b). In the experiment, a corner reflector is placed in front of the prototype as a static target. By changing the distance between the corner reflector and the antenna pair at an interval of 1.8 m, the range of the corner reflector is successively measured when the center frequencies of the LFM radar signal are 14 and 26 GHz, respectively. The ranging results of the corner reflector are demonstrated in Figs. 4(c)–4(e). Due to the photonic dechirping, the echo reflected by the corner reflector is compressed in the frequency domain, with the spectra of the dechirped signal shown in Figs. 4(c) and 4(d), respectively. As can be seen, with the increase of the distance, the peak in the frequency domain also increases. The target range $L$ is determined according to Eq. (11). It is also noted that a fixed peak at 18.24 m always exists, which is mainly caused by the reflection of a column facing the prototype. The range errors are shown in Fig. 4(e), which is no more than $\pm 2\mathrm{cm}$.

To further verify the ranging performance of the prototype, two smaller corner reflectors close to each other are employed as the ranging targets, as depicted in Fig. 5(a). The center frequency of the LFM radar signal is adjusted from 10 to 38 GHz with a step of 2 GHz. The distance between the two corner reflectors is adjusted to the minimum distance that can be fully and easily distinguished when the prototype works at different frequency bands. As shown in Fig. 5(b), the distance between the two smaller corner reflectors is set to between 5.3 and 6 cm over the whole operating bandwidth. The circles represent the measured distance between the two corner reflectors, whereas the solid line represents the actual distance. The errors are $<1.5\mathrm{cm}$. Figures 5(c)–5(f) show the zoomed-in view of the dechirped signal spectra when the center frequencies of the radar LFM signal are 12, 20, 28, and 36 GHz. The corresponding distances between the two corner reflectors are 6, 5.3, 5.5, and 5.3 cm, respectively. It can be seen that two corner reflectors are easily distinguished. It is actually possible to distinguish even closer distances, and only the cases that can be fully and easily distinguished are given in the experiment. According to Eq. (12), the theoretical range resolution is 3.75 cm when the bandwidth of the LFM radar signal is 4 GHz. As shown in Figs. 5(c)–5(f), the full width at half-maximum of the peaks is from 2.9 to 3.5 cm, which is close to the theoretical value.

Then, the radial velocity measurement using the prototype is verified. Figure 6(a) shows the schematic diagram of the radial velocity measurement. A cuboid is positioned on the turntable, maintaining a distance of $R$ from the center of the rotating platform. The turntable rotates in a horizontal plane with an angular velocity of $\omega$, either clockwise or counterclockwise. When the velocity direction of the target is consistent with the radar line of sight, as shown in Fig. 6(a), a measurement of the target’s range at two extremely proximate instances in time is sufficient, from which the radial velocity of the target can be derived based on the rate of change in range. The reason for conducting radial velocity measurement through the aforementioned method is that we lack the equipment that can support the uniform linear motion of a target.

In the experiment, the angular velocity of the turntable is $\omega =0.256\mathrm{rad}/\mathrm{s}$, and the time interval for two range measurements is set to 0.5 s. The time interval cannot be too small to reduce the radial velocity error caused by ranging errors; neither can it be too large, as an overly large interval would result in a significant deviation in the target’s velocity direction between the two range measurements, which would also increase the radial velocity measurement error. Under these circumstances, the radial velocity of the cuboid can be obtained by $(L_2-L_1)/(0.5\mathrm{s})$, whereas the theory value of the radial velocity is obtained by $\omega \times R$. Figure 6(b) shows the radial velocity measurement setup. The cuboid is rotated clockwise or counterclockwise with a rotation radius of 45 cm; the radial velocity of the cuboid is measured and shown in Figs. 6(c) and 6(d). In this case, the theoretical radial velocities of the target when rotated clockwise and counterclockwise are around 11.5 and $-11.5\mathrm{cm}/\mathrm{s}$, respectively. When the rotation radius of the cuboid is changed to 60 cm, the radial velocity measurement results change; they are depicted in Fig. 6(e). As can be seen from Figs. 6(c)–6(e), the measured radial velocity fluctuates around the theoretical value, and the errors are all less than $\pm 1\mathrm{cm}/\mathrm{s}$.

Subsequently, the capability of ISAR imaging of the prototype is demonstrated. Five cylinders made of aluminum with an outer diameter of 3 cm are selected as the targets, and they are placed on the turntable, as shown in Fig. 7(a). The angular velocity of the turntable is also $\omega =0.256\mathrm{rad}/\mathrm{s}$. Along the radar line of sight, the distance between the center of the turntable and the antenna pair is 2.22 m. The accumulation time in ISAR imaging is 1.5 s. Figures 7(b)–7(f) show the ISAR imaging results of the five cylinders. The center frequencies of the radar LFM signal are 12, 18, 24, 30, and 36 GHz. The five targets are easy to distinguish and identify. However, as the frequency increases, due to the lower signal power and greater transmission loss, the target becomes weaker and weaker in the images.

Then, the distance between adjacent cylinders is further reduced to show the ISAR imaging performance. In addition, the center frequency of the LFM radar signal is also changed to show the imaging capability of the prototype at other frequencies. Another three cylinders with an outer diameter of 2.5 cm are placed on the turntable as imaging targets, as shown in Fig. 8(a). Different from that in Fig. 7(a), the distance between adjacent cylinders is changed to 10 cm. Figures 8(b)–8(f) show the ISAR imaging results when the center frequencies of the LFM radar signal are 14, 20, 26, 32, and 38 GHz. As can be seen, the three cylinders can be clearly distinguished. The range resolution of the ISAR imaging is determined by the bandwidth of the radar signal, which is 3.75 cm. The cross-range resolution is determined by the center frequency of the radar signal, the accumulation time, and the target rotation angular velocity. According to Eq. (13), with the increase of the center frequency of the radar signal from 10 to 38 GHz, the cross-range resolution of the ISAR imaging improves from 3.91 to 1.03 cm. The improvement can be clearly observed by comparing Figs. 8(b)–8(f).

The spectrum sensing function of the prototype is then tested. In this test, the center frequency of the IF LFM is set to 4 GHz, and the instantaneous analysis bandwidth is 2 GHz, which is determined by the bandwidth of the $-2\mathrm{nd}$-order optical sideband. Furthermore, the frequency measurement range is adjusted by tuning the pump wave frequency by changing the frequency of the signal output from the MSG. Specifically, as the MSG signal frequency varies from 2.4 to 38.4 GHz in increments of 2 GHz, the corresponding frequency measurement range shifts accordingly, from 0 to 2 GHz initially to 36 to 38 GHz, also in increments of 2 GHz. An arbitrary waveform generator (AWG, Keysight M8195A) is used to generate a single-tone or two-tone RF SUT, whereas a synthesized sweeper (HP 83752B) is used to generate a frequency-sweep SUT with a period of 100 ms. An SUT transmitting antenna is connected to the two signal sources to radiate the SUT to free space, and Rx2 of the prototype collects the SUT in free space, and the frequency information of the SUT is further analyzed by the prototype. It should be noted that because the SUT transmitting antenna can only work in the frequency range from 8 to 40 GHz, the AWG and synthesized sweeper are directly connected to the prototype by RF cables when the SUT frequency is within 8 GHz. In this case, the synthesized sweeper output power is set to $-50\mathrm{dBm}$, and the AWG output power is adjusted to around $-45\mathrm{dBm}$ by attenuators. When the SUT frequency is greater than 8 GHz, the AWG and synthesized sweeper output is connected to the SUT transmitting antenna, and Rx2 is used to receive the SUT from the SUT transmitting antenna 1 m away, as shown in Fig. 9(a). In this case, the AWG and synthesized sweeper output power are set to 4 and 7 dBm, respectively. Furthermore, when Rx2 is used for spectrum sensing, the interference from the radar-transmitting antenna should be isolated to avoid self-interference. In this case, radar and spectrum sensing functions can be performed simultaneously.

Figures 9(b)–9(f) show the frequency measurement results of the single-tone SUT. The error standard deviations of 10 measurements when the SUT frequency changes from 0.3 to 31.9 GHz with a step of 0.4 GHz are shown in Fig. 9(b). It can be seen that the measurement error is $<\pm 7\mathrm{MHz}$. The temporal waveforms of the electrical pulses in a single sweep period corresponding to one measurement at different frequency bands are shown in Figs. 9(c)–9(f). Because the maximum signal frequency that can be generated in our lab does not exceed 32 GHz using the AWG, only measurements up to 31.9 GHz are shown in Fig. 9. It should also be noted that the frequency of the signal can be calculated according to the generated electrical pulses by $2t_{\mathrm{sut}}/T\mathrm{GHz}+f_{\text{a}}+2f_{\text{C}}-10.4\mathrm{GHz}$, where $t_{\mathrm{sut}}$ is the time instant corresponding to the peak value of the pulse within a measurement period $T$. In Fig. 9, the time axis is replaced with the corresponding frequency axis. It can be seen that different SUT frequencies correspond to pulses distributed at different locations. In Fig. 9(c), when the SUT frequency is 0.3 and 0.7 GHz, there are some extra frequency components, which are the harmonics of the signal generated by the AWG. In Fig. 9(f), the waveform amplitude gets smaller when the SUT frequency is larger than 30.3 GHz, which is mainly attributed to the limited bandwidth of the AWG. By accurately adjusting the pump wave frequency, the frequency measurement range can be broadened to span 0 to 40 GHz, provided that the center frequency of the IF LFM signal exceeds 4.2 GHz. Conversely, when the center frequency of the IF LFM signal is 2.5 GHz, the minimum frequency measurement range is limited to 0 to 36.6 GHz.

Figure 10 shows the frequency measurement results of a two-tone test. The two-tone SUT has a fixed frequency at 3 GHz and another frequency tuned from 3.02 to 3.05 GHz with a step of 10 MHz, corresponding to a frequency interval from 20 to 50 MHz. The two-tone SUT can be distinguished even if the frequency interval is only 20 MHz, which means the frequency resolution of the frequency measurement function of the prototype is better than 20 MHz.

As demonstrated in the above experiment, the frequency of a stationary signal can be identified in a single frequency-sweep period, i.e., 1.4876 ms in the prototype. Nevertheless, a nonstationary signal can still be analyzed using the prototype if it exhibits local stationarity over the 1.4876 ms frequency-sweep period. In this case, a time–frequency diagram is commonly used to characterize the SUT by accumulating the results of multiple frequency-sweep periods. Figures 11(a)–11(c) show the time–frequency analysis results of a 100-ms frequency-sweep SUT generated by the synthesized sweeper when the frequency-sweep range of the SUT is 6 to 8 GHz, 10 to 12 GHz, and 16 to 18 GHz, respectively. The corresponding temporal waveforms are shown in Figs. 11(d)–11(f). As can be seen, the linear frequency-sweep characteristics of the SUT are well characterized by the time–frequency diagrams. However, in Fig. 11(b), a noticeable detail is that there is a discontinuity in the time–frequency diagram at 11 GHz, which is attributed to the fact that 11 GHz is a bandwidth point of the synthesized sweeper. A switch event at this point creates a sweep gap when the sweeper sweeps over the bandwidth point.40 It is worth noting that 6.75 GHz is also a bandwidth point, but it is not visible in Fig. 11(a) because the synthesizer switch is deactivated if the sweep range is $<80\%$ of the sweep starting frequency within the 2 to 11 GHz frequency band.40 In addition, the vacancy after the frequency-sweep SUT is the result of the retrace of the synthesized sweeper when a sweep is finished. Moreover, the waveform amplitudes in Figs. 11(d)–11(f) are not uniform but have a similar shape, which is due to the use of the same $-2\mathrm{nd}$-order LFM optical sideband in the test, and the LFM optical sideband is inherently not completely flat. The envelope of the waveform in Figs. 11(d)–11(f) exhibit shapes similar to the radar signal spectrum at a 16 GHz center frequency, as illustrated in Fig. 3, because the same IF LFM signal center frequency is employed. In addition, by further reducing the sweep period $T$ of the IF LFM signal, the time required for a single measurement can be further decreased, enabling faster signal measurement. This will aid the system in measuring signals with more rapid frequency variations. It is worth noting that in the prototype, we do not further reduce the period, primarily due to limitations of the DDS in the IF LFM signal source. Further reduction in the period would significantly degrade the signal quality. Therefore, a high-quality and high-speed electrical frequency-sweep source is crucial for enhancing the performance of this system.

A comparison between the proposed prototype and the previously reported photonics-assisted joint radar and spectrum sensing systems is given in Table 1. As can be seen, the advantages and key significance of this work are as follows: (1) First prototype-level integration: To the best of our knowledge, this is the first microwave photonic prototype that achieves integration of radar and spectrum sensing functions over the large frequency range (8 to 40 GHz for radar and 0 to 40 GHz for spectrum sensing) with high completeness. The two functions can operate concurrently and can be tuned independently within their operating bandwidths. (2) Simultaneous multifunctionality: This is the first microwave photonic system that is reported for simultaneous performing ranging, radial velocity measurement, ISAR imaging, frequency measurement, and time–frequency analysis. (3) Better performance indicators, especially the operating bandwidth of the two functions: Most of the key performance indicators of the prototype are better than the reported photonics-assisted joint radar and spectrum sensing systems; the capability of independently adjustable double functions across a 40-GHz ultrawide operating bandwidth has not been previously reported.

Due to the integration of radar and spectrum sensing functions in the prototype, which shares partial hardware and signals, and given that both radar and spectrum sensing functions may need to operate simultaneously, the potential interference between these two functions during operation as well as external interferences need to be thoroughly considered and mitigated. The potential interferences that the system faces mainly include radar self-interference, interference from radar to spectrum sensing, and interference from free-space signals to radar.

1. 1)When the radar is a frequency-modulated continuous wave radar as employed in the prototype, the self-interferences that leak from the radar transmitting link and antenna into the radar receiving link and antenna need to be considered. In practical applications, radar self-interference can be significantly reduced using the self-interference cancellation technique,41 which can be implemented using a reference signal tapped from the radar transmitter to subtract the radar self-interference in the receiver. Because the focus of this prototype is on the joint radar and spectrum sensing system, and due to constraints related to implementation complexity and cost, we have not incorporated the self-interference cancellation technique in the prototype. In the prototype, we primarily adopt the method of spatial isolation to reduce self-interference by arranging the radar transmitting and receiving links in opposite directions on the layout and separating the radar transmitting and receiving antennas as much as possible on the exterior of the prototype. In addition, the self-interference is further mitigated by attaching some microwave-absorbing materials to the sides of the transmitting and receiving antennas.
2. 2)Because the radar transmitting power is commonly very high, the leakage from the radar transmitting link and antenna into the spectrum sensing receiving link and antenna also needs to be considered, that is the interference from radar to spectrum sensing. This interference is very similar to the radar self-interference discussed above, so the methods mentioned above can also be used to mitigate the interference from radar to spectrum sensing. Furthermore, because the two functions of the prototype can operate independently and be tuned within their respective operating bandwidths, we can effectively mitigate the impact of this interference by ensuring that the radar and spectrum sensing functions operate on nonoverlapping frequency bands as much as possible.
3. 3)Various electromagnetic waves in free space are received by the radar receiving antenna, which may include some active interferences, such as the interference emitted by jammers in electronic warfare, as well as other electromagnetic waves within the same frequency band as the radar. The interference from free-space signals to radar, which is of the same frequency as the radar, when received by the radar receiver, can affect the radar performance. The cognitive radio function of the joint radar and spectrum sensing system can play a crucial role in mitigating this interference. The basic principle involves real-time monitoring of the electromagnetic spectrum and dynamically adjusting the operating frequency band of the radar function based on the surrounding electromagnetic spectrum information. The specific principles will be discussed in detail in the next subsection.

The proposed prototype can be applied to cognitive radio systems, intelligent transportation systems, and electronic warfare systems. Figure 12 shows the schematic diagram of the prototype’s application in cognitive radio scenarios. In cognitive radio systems, the spectrum sensing function of the prototype enables rapid scanning of the electromagnetic spectrum within the range of 0 to 40 GHz. Therefore, the presence of interferences and already utilized frequency bands can be quickly identified and positioned, and the spectrum holes can be acquired. According to spectrum holes found via the spectrum sensing function, the prototype can quickly adjust the radar signal frequency and set it within the spectrum holes, thus avoiding interference with the radar function and impacts from other signals. It is worth noting that since the radar and spectrum sensing functions can operate independently and simultaneously within 40 GHz, even when the radar system is in operation, the spectrum sensing function can continue to scan the entire electromagnetic spectrum continuously. Once any interference is detected within the radar’s operating frequency band, the prototype can once again perform frequency agility on the radar signal, enabling it to operate normally without being disrupted.

In summary, we have proposed, designed, built, and investigated a microwave photonic prototype for concurrent radar detection and spectrum sensing. At the system level, through the microwave photonic design, we achieve true integration by sharing hardware and signals between radar and spectrum sensing. Functionally, radar and spectrum sensing operate concurrently and can be tuned independently within the vast frequency range up to 40 GHz, respectively, which was previously unattainable. In the prototype, the radar signal has an instantaneous bandwidth of 4 GHz and a tunable frequency range from 8 to 40 GHz, whereas the spectrum sensing instantaneous bandwidth is 2 GHz and the spectrum sensing frequency range could be tuned from 0 to 40 GHz. Notably, this represents the first prototype-level integration of microwave photonic radar and spectrum sensing, surpassing previous desktop experiments. In addition, it is the first microwave photonic joint radar and spectrum sensing system capable of simultaneous radar ranging, velocity measurement, imaging, frequency measurement, and time–frequency analysis. Comprehensive experimental results show a ranging error of less than $\pm 2\mathrm{cm}$, a radial velocity error of less than $\pm 1\mathrm{cm}/\mathrm{s}$, clear imaging of multiple small targets, a frequency measurement error of less than $\pm 7\mathrm{MHz}$, and a frequency resolution of better than 20 MHz. The prototype is expected to find applications in cognitive radio systems, intelligent transportation systems, and electronic warfare systems with further miniaturization and integration design.

Taixia Shi is currently a postdoctoral researcher at the School of Communication and Electronic Engineering, East China Normal University. He received his BS degree in physics from the Shanxi Datong University, Datong, China, in 2015, his MS degree in materials science and engineering from the Taiyuan University of Technology, Taiyuan, China, in 2019, and his PhD in communication and information systems from the East China Normal University, Shanghai, China, in 2023. His research interest focuses on microwave photonic signal measurement and photonics-assisted self-interference cancellation.

Dingding Liang is currently pursuing toward his PhD at the School of Communication and Electronic Engineering, East China Normal University, Shanghai, China. He received his BE degree in electronic and information engineering from the Yancheng Teachers University, Yancheng, China, in 2018. His research interest focuses on microwave photonic radar systems.

Lu Wang is currently a postdoctoral fellow at the Beijing Institute of Control Engineering, Beijing, China. She received her PhD in microelectronics and solid-state electronics from the University of Chinese Academy of Sciences, Beijing, China, in 2023. Her research interests include microwave photonic radar and microwave photonic signal measurement.

Lin Li is currently a professor at the Beijing Institute of Control Engineering, Beijing, China. He received his PhD from the University of Chinese Academy of Sciences, Beijing, China, in 2018. His research interests include spacecraft attitude measurement, optical instrument design, optical-mechanical thermal analysis, and spacecraft structural dynamics translation.

Shaogang Guo is currently a professor at the Beijing Institute of Control Engineering, Beijing, China. His research interests include space optoelectronic measurement and sensing, as well as lidar.

Yang Chen is currently a full professor at the School of Communication and Electronic Engineering, East China Normal University, Shanghai, China. He has authored or co-authored more than 70 papers in peer-reviewed journals and more than 20 papers in conference proceedings. His current research interests include microwave photonics, optoelectronic oscillators, radio-over-fiber techniques, and optical communications. He was selected to receive an inaugural IEEE/Optica Journal of Lightwave Technology Outstanding Reviewer Award in 2023 and an IEEE Photonics Journal Outstanding Reviewer Award in 2022. He was listed in the World’s Top 2% Scientists elaborated by Stanford University in 2023 and 2024.

Biographies of the other authors are not available.