Scalable distributed microwave photonic MIMO radar based on a bidirectional ring network

Qiang Sun; Qiang Sun; Jingwen Dong; Chenyu Liu; Chenyu Liu; Jiyao Yang; Jiyao Yang; Xiangpeng Zhang; Xiangpeng Zhang; Wangzhe Li; Wangzhe Li

doi:10.1364/OE.437658

1. Introduction

Distributed multiple-input-multiple-output (MIMO) radars refer to systems which transmit signals at multiple positions and jointly process echoes from multiple receiving antennas [1]. Exploiting the spatial diversity of targets, distributed MIMO radars have advantages in accuracy parameter estimation and have attracted increasing research attention [2–11]. Nevertheless, distributed MIMO radars adopting radio frequency (RF) technology face two issues: 1) Signal transmission capacity and signal coherence of RF links are insufficient for raw data jointly processing [12]; 2) The increase of remote nodes challenges the cost, size, weight, and power (SWaP) of the system. In recent years, microwave photonic technology has been considered as a promising solution to address these problems, taking advantage of its excellent performances in low transmission loss, high frequency, large bandwidth, and flexible multiplexing ability. Adopting microwave photonic technologies, several distributed MIMO radars are proposed, which can be classified into two categories according to signal distribution methods. In one category, optical signals are first sent to a place far from the local node (Ln), where the optical signals are further divided and distributed to different remote nodes [13–16]. In [14], a MIMO radar with a bandwidth of 4 GHz is proposed, in which optical transmitting signals are together sent to a remote place to generate RF signals. Next, the RF signals are distributed to different transmitting nodes (Tns) by cables, which limits the flexibility of the Tns in deployments. In [15,16], an optical transmitting signal and an optical reference signal are grouped together and sent to a remote place sharing a fiber link, resulting in that the Tn and the receiving node (Rn) have to be deployed closely. In this category, the distribution of radars is limited, which makes the distributed MIMO radars difficult to take full advantage of the spatial diversity of targets. In the other category, optical signals are first divided and then separately sent to the corresponding remote node [17–20]. In [18,19], a 2×2 distributed MIMO radar with a bandwidth of 100 MHz is proposed, in which optical signals are divided in the Ln and further sent to remote nodes through their own private fiber link. In this radar, the deployments of Tns and Rns are more flexible. However, there is a common problem in the radar systems of this category: the number of optical fiber links rises with the increase of remote nodes, resulting in limited utilization of transmission capacity and a high complexity of the system. In addition, all the developed methods about signal distribution of photonics-based MIMO radars above are based on a star network [21], leading to a dilemma in which the radars can not take full advantage of the spatial diversity of targets and simultaneously reduce the system complexity.

In this paper, a scalable distributed microwave photonic MIMO radar based on a bidirectional ring network is proposed and experimentally demonstrated. The network is constructed with a fiber ring, in which a Ln, several Tns, and several Rns are connected in series. In the Ln, optical carriers with different wavelengths modulated by an intermediate frequency (IF) signal are injected into the fiber ring and distributed to Tns and Rns along opposite directions. For the reason that the Tns/Rns are wavelength selective, optical signals with corresponding center wavelengths are selected in different Tns/Rns for signal transmitting/echo receiving. Finally, optical signals carrying echoes are sent to the Ln for parallel de-chirp processing and digitized for position and velocity measurements. Benefiting from the pluggable remote nodes and the bidirectional ring network, it is easy to realize a large-scale distributed MIMO radar with flexible scalability. In addition, the bidirectional ring network enables the radar system to take full advantage of the spatial diversity and simultaneously reduce the system complexity. Experimentally, a 2×2 distributed MIMO radar operating in X and Ku bands is established. Two-dimensional position and velocity measurements of a moving target are carried out with errors of better than 8 cm and 0.20 m/s respectively.

2. Principle

2.1 Architecture of the proposed radar system

The proposed distributed microwave photonic MIMO radar is implemented based on a novel bidirectional ring network, as shown in Fig. 1(a). The ring network consists of a fiber ring, a Ln, N Tns and N Rns. Both Tns and Rns are wavelength selective and pluggable. Optical radar signals generated in the Ln are sent to Tns and Rns along opposite directions, acting as optical transmitting signals and optical reference signals respectively. In the ith transmitting node (Tn_i), only the ith optical transmitting signal (the optical transmitting signal with a wavelength of λ_i) is selected and converted to RF signal for signal transmitting. In the ith receiving node (Rn_i), only the ith optical reference signal (the optical reference signal with a wavelength of λ_i) is modulated by echoes and sent back to the Ln for parallel de-chirp processing. After digitization, de-chirped signals are extracted for position and velocity measurements. Something needs to be emphasized is that optical add-drop multiplexers (OADMs) are used to construct wavelength selective and pluggable Tns and Rns. As shown in the inset of Fig. 1(a), the OADM is composed of two back-to-back wavelength-division multiplexers (WDMs). Obviously, optical radar signals can pass through the OADM in any direction, and only the optical signal with a specific wavelength is selected. In addition, redundant OADMs can be connected to the fiber ring. They are easy to be upgraded to remote nodes or degraded back to OADMs depending on the need, thereby realizing the pluggable remote nodes.

Fig. 1. (a) The architecture of the ring network and the OADM. (b) The structure of the Ln. (c) The structure of Tn_i. (d) The structure of Rn_i. Tn_i, the ith transmitting node; Rn_i, the ith receiving node; OADM, optical add-drop multiplexer; OCIR, optical circulator; OS, optical splitter; WDM, wavelength-division multiplexer; OFCG, optical frequency comb generator; MZM, mach-zehnder modulator; EDFA, erbium-doped fiber amplifier; PD, photodetector; LNA, low-noise amplifier; PA, power amplifier; OBPF, optical band-pass filter; EBPF, electical band-pass filter; ELPF, electrical low-pass filter; ADC, analog-to-digital converter; DSP, digital signal processor.

Download Full Size | PDF

As we can see, the measurement can be divided into four steps: signal generation, signal transmitting, signal receiving, and signal processing.

In the signal generation, the optical transmitting signals and optical reference signals with different center wavelengths are generated. As shown in Fig. 1(b), N optical carriers from a laser array are multiplexed by a WDM and fed into an optical frequency comb (OFC) generator. Driven by a sinusoidal signal, the generator produces N OFCs with the same free spectrum range (FSR), and each OFC has multiple comb teeth. The output of the optical frequency comb generator can be expressed as:

(1)$${E_{OFCs}} \propto \sum\limits_{i = 1}^N {\exp [j2\pi {f_{ci}}t]} \times \sum\limits_{m = 0}^{M - 1} {\exp [j(2\pi m{f_0}t)]}$$

where f_ci is the frequency of ith optical carrier, f₀ is the FSR of the OFCs, M is the number of comb teeth of each OFC. Here, N OFCs with M comb teeth are obtained and each comb tooth can be regarded as an optical carrier. The OFCs are modulated by an IF symmetrical triangular linearly frequency modulated (STLFM) signal via an MZM to generate optical radar signals. The STLFM signal S_t(t) can be expressed as:

(2)$${S_t}(t) = \cos [2\pi ({f_e}t + \frac{B}{2}t + {( - 1)^p} \times \frac{1}{2}k{t^2})]\;\;\;p = \left\{ \begin{array}{l} 0\textrm{ } - \frac{T}{2} < t < 0\\ 1\textrm{ }0 < t < \frac{T}{2} \end{array} \right.$$

The p = 0 and p = 1 correspond to the up-chirp segment and down-chirp segment of the STLFM signal respectively. The f_e, k, B = kT/2 and T are the center frequency, chirp rate, bandwidth and duration of the STLFM signal. The MZM is biased at the null point to suppress optical carriers and even order sidebands. The optical radar signals from the MZM can be expressed as:

(3)$$\begin{array}{l} {E_T}(t) = {E_{OFCs}} \times \{ \exp [j({\beta _0}{S_t}(t) + \frac{\pi }{2})] + \exp [ - j({\beta _0}{S_t}(t) + \frac{\pi }{2})]\} \\ \textrm{ } \propto \sum\limits_{i = 1}^N {\exp [j2\pi {f_{ci}}t]} \times \sum\limits_{m = 0}^{M - 1} {\exp [j(2\pi \cdot m{f_0}t)]} \times \cos [2\pi ({f_e}t + \frac{B}{2}t\textrm{ + }{( - 1)^p} \times \frac{1}{2}k{t^2})] \end{array}$$

where β₀ is the modulation index of the MZM. Here, the high order sidebands are ignored under the small-signal approximation. For convenience, the ith optical radar signal E_Ti(t) is exploited for further mathematic derivation, which can be expressed as:

(4)$${E_{Ti}}(t) = \exp [j2\pi {f_{ci}}t] \times \sum\limits_{m = 0}^{M - 1} {\exp [j(2\pi \cdot m{f_0}t)]} \times \cos [2\pi ({f_e}t + \frac{B}{2}t\textrm{ + }{( - 1)^p} \times \frac{1}{2}k{t^2})]$$

Comparing Eq. (3) and Eq. (4), the E_Ti(t) is 1/N of E_T(t), and the other (N-1)/N of E_T(t) is similar to the E_Ti(t). After amplified by an erbium-doped amplifier (EDFA), one part of E_Ti(t) is sent to Tn_i with a time delay of τ_txi = τ₁+τ₂+…+τ_2i-1 and the other part is sent to Rn_i with a time delay of τ_rxi= τ_2i+1+… +τ_2n+1.

In the signal transmitting, RF STLFM signals with different center frequencies are generated and emitted to targets. As shown in Fig. 1(c), in the Tn_i, the ith optical transmitting signal is selected by an OADM and sent to a PD via an OCIR. In the PD, by heterodyning the ±1 order sidebands, 2M-1 frequency-doubled RF STLFM signals are generated. This means that 2M-1 RF signals can be transmitted simultaneously by one transmitting node if needed. To avoid spectrum overlapping of RF signals, the FSR of OFCs, the center frequency and bandwidth of the IF STLFM signal should be carefully set. The RF STLFM signals are filtered by an electrical band-pass filter (EBPF), and only the desired transmitting signal is reserved. The desired transmitting signal is expressed as Eq. (5):

(5)$$s{t_i}(t) \propto \left\{ \begin{array}{l} \cos [2\pi ({h_i}{f_0}(t - {\tau_{txi}}) + (2{f_e} + B)(t - {\tau_{txi}}) + {( - 1)^p} \times k{(t - {\tau_{txi}})^2})]\textrm{ }{h_i}\textrm{ = }0\\ \cos [2\pi ({h_i}{f_0}(t - {\tau_{txi}}) \pm (2{f_e} + B)(t - {\tau_{txi}}) \pm {( - 1)^p} \times k{(t - {\tau_{txi}})^2})]\textrm{ }{h_i}\textrm{ = 1,2}\ldots \textrm{(}M\textrm{ - 1)} \end{array} \right.$$

As shown in Fig. 2, by beating the −1 order sideband of f_ci and +1 order sideband of f_ci+h_if₀ (h_i=0, 1…, M−1), the transmitting signal with a center frequency of h_if₀+2f_e is generated. By beating the +1 order sideband of f_ci and −1 order sideband of f_ci+h_if₀ (h_i=1, 2…, M−1), the transmitting signal with a center frequency of h_if₀−2f_e is generated.

Fig. 2. (a) Optical spectrum of ith optical transmitting signal before the PD. (b) Electrical spectrum of the output of the PD in Tn_i.

Download Full Size | PDF

As shown in Fig. 2(b), the magnitude of an RF signal with a lower center frequency is usually larger than that with a higher center frequency because the RF signal with a lower center frequency is obtained by heterodyning more ±1 order sidebands. After proper RF gain controlling, the desired RF STLFM signal is emitted to targets.

In the signal receiving, echoes with different center frequencies are received by modulating the corresponding optical reference signals. Concretely, in the Rn_i, the echo with a center frequency of h_if₀±2f_e (h_i=1, 2…, M−1) or 2f_e is collected by the receiving antenna. As shown in Fig. 1(d), the ith optical reference signal is selected by an OADM, amplified by the EDFA and routed into an MZM biased at the quadrature point. Via the MZM, the ith optical reference signal is modulated by the echo. The output signal of the MZM is given by

(6)$$\begin{aligned}{} {E_{MZM}} &\propto {E_{Ti}}(t - {\tau _{rxi}}) \times \{ {J_0}({\beta _i}) - 2{J_1}({\beta _i})\cos [2\pi ({h_i}{f_0}(t - {\tau _{txi}} - {\tau _{echo\_i}})\\ &\textrm{ } \pm (2{f_e} + B)(t - {\tau _{txi}} - {\tau _{echo\_i}}) \pm {( - 1)^p} \times k{(t - {\tau _{txi}} - {\tau _{echo\_i}})^2})]\} \end{aligned}$$

where β_i is the modulation index of the MZM. An optical band-pass filter (OBPF) is used to select the desired sidebands. The output of the OBPF can be expressed as

(7)$${E_{Ri}}(t) = \exp [j2\pi {f_{ci}}(t - {\tau _{rxi}})] \times \{ \frac{{{J_0}({\beta _i})}}{2}{E_1} - \frac{{{J_1}({\beta _i})}}{2}{E_2}\}$$

when the center frequency of the echo is h_if₀+2f_e (h_i=0, 1…, M−1), the spectrum components of E₁ and E₂ are illustrated in Fig. 3(b), expressed as

(8)$$\begin{array}{l} {E_1} = \exp [ - j2\pi (({f_e} + \frac{B}{2})(t - {\tau _{rxi}}) + {( - 1)^p} \times \frac{1}{2}k{(t - {\tau _{rxi}})^2})]\\ {E_2} = \exp [j2\pi ({h_i}{f_0}(t - {\tau _{rxi}}) + ({f_e} + \frac{B}{2})(t - {\tau _{rxi}}) + {( - 1)^p} \times \frac{1}{2}k{(t - {\tau _{rxi}})^2}\\ - {h_i}{f_0}(t - {\tau _{txi}} - {\tau _{echo\_i}}) - (2{f_e} + B)(t - {\tau _{txi}} - {\tau _{echo\_i}}) - {( - 1)^p} \times k{(t - {\tau _{txi}} - {\tau _{echo\_i}})^2})] \end{array}$$

The E₂ is generated by modulating the +1 sideband of f_ci+h_if₀ with the echo whose center frequency is h_if₀+2f_e. When the center frequency of the echo is h_if₀−2f_e (h_i=1, 2…, M−1), the spectrum components of E₁ and E₂ are expressed as

(9)$$\begin{array}{l} {E_1} = \exp [j2\pi (({f_e} + \frac{B}{2})(t - {\tau _{rxi}}) + {( - 1)^p} \times \frac{1}{2}k{(t - {\tau _{rxi}})^2})]\\ {E_2} = \exp [j2\pi (h{f_0}(t - {\tau _{rxi}}) - ({f_e} + \frac{B}{2})(t - {\tau _{rxi}}) - {( - 1)^p} \times \frac{1}{2}k{(t - {\tau _{rxi}})^2}\\ - {h_i}{f_0}(t - {\tau _{txi}} - {\tau _{echo\_i}}) + (2{f_e} + B)(t - {\tau _{txi}} - {\tau _{echo\_i}}) + {( - 1)^p} \times k{(t - {\tau _{txi}} - {\tau _{echo\_i}})^2})] \end{array}$$

Fig. 3. (a) The spectrum before MZM in Rn_i. (b) The spectrum after OBPF in Rn_i.

Download Full Size | PDF

Here, the E₂ is generated by modulating the -1 sideband of f_ci+h_if₀ with the echo whose center frequency is h_if₀-2f_e. The difference between Eq. (8) and Eq. (9) results from different center frequencies of OBPFs, which are set according to the signal generation approaches (generated using the −1 order sideband of f_ci or the +1 order sideband of f_ci). Something need to be emphasized is that the OADM also can select optical reference signals with other wavelengths (λ_i1, λ_i2, λ_i3, …) to receive multiple echoes. Passing by the other Tns (Tn_i, Tn_i_-1…, Tn₁) and Rns (Rn_i_-1, Rn_i_-2…, Rn₁), the ith modulated optical reference signal is sent to the signal processing unit with a time delay of τ_{rxi_rc}= τ₁+τ₂ … +τ_i.

In the signal processing, all modulated optical reference signals are sent to a low-speed PD through an OCIR, as shown in Fig. 1(b). Due to the different center wavelengths of modulated optical reference signals from different Rns, parallel de-chirp processing for all echoes is implemented. The obtained de-chirped signals can be expressed as

(10)$$\begin{aligned} Sr(t) = \sum\limits_{i = 1}^N &{S{r_i}(t)} \\ \textrm{ } = \sum\limits_{i = 1}^N &\{ \cos [2\pi (({h_i}{f_0} \pm (2{f_e} + B))({\tau _{txi}} + {\tau _{echo\_i}} - {\tau _{rxi}})\\ \textrm{ } &\pm {( - 1)^p} \times 2kt({\tau _{txi}} + {\tau _{echo\_i}} - {\tau _{rxi}})\\ \textrm{ } &\pm {( - 1)^p} \times k({({\tau _{rxi}} + {\tau _{rxi\_rc}})^2} - {({\tau _{txi}} + {\tau _{echo\_i}} + {\tau _{rxi\_rc}})^2}))]\} \end{aligned}$$

To remove the high-frequency interference, de-chirped signals are filtered by an electronic low-pass filter (ELBP) with a cut-off frequency of B.

In the proposed radar system, Rayleigh backscattering is caused due to the same wavelength of optical transmitting signals and optical reference signals. However, the ratio of Rayleigh backscattering power to the counter-propagating signal power is more than 30 dB at a fiber length of 20 km [22]. Given that the relatively short fiber length and the high pulse compression gain, the influence of Rayleigh backscattering is acceptable in the proposed system. When the fiber length is further extended, the influence of Rayleigh backscattering can be reduced by shifting the frequencies of optical reference signals.

2.2 Two-dimensional position and velocity measurements method based on the proposed radar system

The two-dimensional position and velocity measurements can be achieved exploiting the following two steps generally: 1) Calculate the distances and radial velocities of the target; 2) Calculate the two-dimensional position and velocity of the target based on the relative positions of the target and phase centers of antennas.

Firstly, calculate the distances and radial velocities exploiting the de-chirped signals. Define φ_ri(t) to be the phase of Sr_i(t). Given τ_{echo_i} = 2(R_{0_i}−v_it)/c (R_{0_i} is the initial distance, v_i is the radial velocity of the target), the φ_ri(t) can be expressed as

(11)$$\begin{aligned} {\varphi _{ri}}(t) = 2\pi &(({h_i}{f_0} \pm (2{f_e} + B))(2({{R_{0\_i}} - {v_i}t} )/c)\\ \textrm{ } &\pm {( - 1)^p} \times 2kt({\tau _{txi}} + 2({{R_{0\_i}} - {v_i}t} )/c - {\tau _{rxi}})\\ \textrm{ } &\pm {( - 1)^p} \times k( - {({\tau _{txi}} + 2({{R_{0\_i}} - {v_i}t} )/c + {\tau _{rxi\_rc}})^2}))\\ &\textrm{ + }2\pi (({h_i}{f_0} \pm (2{f_e} + B))({\tau _{txi}} - {\tau _{rxi}})\\ \textrm{ } &\pm {( - 1)^p} \times k{({\tau _{rxi}} + {\tau _{rxi\_rc}})^2}) \end{aligned}$$

Only the first three items on the right side of Eq. (11) are related to time t and are not zero after derivation. The derivation of the Eq. (11) can be expressed as

(12)$$\begin{aligned} {f_i}(t) = \frac{{d{\varphi _{ri}}(t)}}{{2\pi dt}} =&{\pm} {( - 1)^p} \times 2k(\frac{{2{R_{0\_i}} - 4{v_i}t}}{c} + {\tau _{txi}} - {\tau _{rxi}}) \pm {( - 1)^p} \times \frac{{4{v_i}k}}{c} ({\tau _{txi}} + {\tau _{rxi\_rc}} + 2({R_{0\_i}} - {v_i}t)/c)\\ &\textrm{ } - \frac{{2{v_i}}}{c}({h_i}{f_0} \pm (2{f_e} + B)) \end{aligned}$$

Due to the target motion, the frequencies of de-chirped signals generated by beating the up-chirp segments (p=0) or the down-chirp segments (p=1) are slightly different, which can be expressed as

(13)$$\begin{aligned} {f_{i\_u}}(t) =&{\pm} 2k(\frac{{2{R_{0\_i}} - 4{v_i}t}}{c} + {\tau _{txi}} - {\tau _{rxi}}) \pm \frac{{4{v_i}k}}{c}({\tau _{txi}} + {\tau _{rxi\_rc}} + 2({R_{0\_i}} - {v_i}t)/c)\\ &\textrm{ } - \frac{{2{v_i}}}{c}({h_i}{f_0} \pm (2{f_e} + B)) \end{aligned}$$

(14)$$\begin{aligned} {f_{i\_d}}(t) =&{\pm} 2k(\frac{{2{R_{0\_i}} - 4{v_i}t}}{c} + {\tau _{txi}} - {\tau _{rxi}}) \pm \frac{{4{v_i}k}}{c}({\tau _{txi}} + {\tau _{rxi\_rc}} + 2({R_{0\_i}} - {v_i}t)/c)\\ &\textrm{ + }\frac{{2{v_i}}}{c}({h_i}{f_0} \pm (2{f_e} + B)) \end{aligned}$$

Considering that there are only positive frequencies in the real world, f_{i_d}(t) has been inverted.

Specifically, for a fixed target, Eq. (13) and Eq. (14) can be expressed as

(15)$${f_i}(t) = 2k(\frac{{2{R_{0\_i}}}}{c} + {\tau _{txi}} - {\tau _{rxi}})$$

Assuming the current detection range of radar is from R_near to R_far, the f_i should fall in [2k(2R_near/c+τ_txi-τ_rxi), 2k(2R_far/c+τ_txi-τ_rxi)). Similarly, f_i+1 should belong to [2k(2R_near/c+τ_txi+1-τ_rxi+1), 2k(2R_far/c+τ_txi+1-τ_rxi+1)). If 2k(2R_near/c+τ_txi+1-τ_rxi+1) > 2k(2R_far/c+τ_txi-τ_rxi) is satisfied, the collision between f_i and f_i+1 is avoided. For a moving target, the influence of the Doppler shift can be eliminated by adding a guardband between the adjacent frequency ranges. In this way, the frequencies of de-chirped signals will appear in different frequency ranges, which means that the de-chirped signal corresponding to a specific Rn can be distinguished and extracted without interference. When the target is out of the current detection range, the adjustment of τ_txi is needed to ensure the f_i fall in the origin [2k(2R_near/c+τ_txi-τ_rxi), 2k(2R_far/c+τ_txi-τ_rxi)) and avoid possible confusions.

The second term on the right side of Eq. (13) and Eq. (14) can be omitted under the small-signal approximation. Subtracting Eq. (13) from Eq. (14), v_i can be obtained as

(16)$$\textrm{ }{v_i} = \frac{{[{f_{i\_d}}(t) - {f_{i\_u}}(t)]c}}{{4[{h_i}{f_0} \pm (2{f_e} + B)]}}\textrm{ }$$

Adding Eq. (13) and Eq. (14), R_{0_i} can be obtained as

(17)$${R_{0\_i}} = [\frac{{{f_{i\_u}}(t)\textrm{ + }{f_{i\_d}}(t)}}{{ \pm 4k}} - {\tau _{txi}} + {\tau _{rxi}}]\frac{c}{2} + 2{v_i}t$$

Secondly, calculate the two-dimensional position and velocity of the target based on the relative positions of the target and phase centers of antennas. Under the far-field condition, the geometric model of the measurements can be illustrated as Fig. 4. For clarity, only the Tn₁-Rn₁ and Tn₂-Rn₂ are drawn. The red dot and green dot are the phase centers of the pair of Tn₁-Rn₁ and Tn₂-Rn₂. The black dot represents the target, which is at the intersection of two circles. The centers of the circles are at the phase centers. The radiuses are the distances obtained by Tn₁-Rn₁ and Tn₂-Rn₂. Combining prior information like antenna pointing, the position of the target can be acquired based on time of arrivals (TOA)[23]. At the same time, the relative positions of the target and phase centers of antennas are obtained. Finally, the magnitude v and direction θ of the velocity can be calculated through the Eq. (18):

(18)$$[{v\cos \theta \textrm{ }v\sin \theta } ]= [{{v_1}\textrm{ }{v_2}} ]\times \left[ \begin{array}{l} \cos {\theta_1}\textrm{ } - \sin {\theta_1}\\ \cos {\theta_2}\textrm{ }\sin {\theta_2} \end{array} \right]$$

θ is the angle between the direction of the velocity and x-axis. v₁ and v₂ are the radial velocities obtained by Tn₁-Rn₁ and Tn₂-Rn₂ respectively. θ₁ is the angle between v₁ and x-axis. θ₂ is the angle between v₂ and x-axis. After two steps above, the two-dimensional position and velocity measurements are achieved based on the proposed distributed microwave photonic MIMO radar system.

Fig. 4. Geometric model of two-dimensional position and velocity measurements

Download Full Size | PDF

3. Experiment and results

Proof-of-concept experiments are carried out. A 2×2 distributed microwave photonic MIMO radar is constructed by a Ln, a fiber ring, two Tns and two Rns. The lengths of the fibers corresponding τ₁, τ₂, …, τ_2n-1 (n = 2) are set as 2 km, 1 km, 200 m, 40 m, 20 m respectively to avoid spectrum overlapping of de-chirped signals.

In the signal generation unit, two optical carriers generated by two lasers (RIO orion and TeraXion PS-TNL) at 1549.32 nm and 1550.12 nm are multiplexed by a WDM and fed into an MZM which operates as an OFC generator. Via the MZM biased at the null point, the two carriers are modulated by a sinusoidal signal of 7 GHz from a signal source (E8267D). After the modulation, two OFCs are generated, and each of them has more than two teeth. An IF STLFM signal with 1.8-GHz center frequency, 0.6-GHz bandwidth, 1100-us pulse repetition interval and 1000-us duration is generated by an arbitrary waveform generator (AWG) (Tektronix, AWG70001A). The STLFM signal modulates the OFCs via another MZM biased at the null point. The optical spectra of the generated optical radar signals are shown in Fig. 5(a), measured by an optical spectrum analyzer (Yokogawa, AQ6370D). After proper amplification, optical radar signals are split into optical transmitting signals and optical reference signals by an optical splitter. Through the fiber ring, the former are anticlockwise sent to Tn₁ and Tn₂ sequentially and detected by a PD in each node. The generated radar signals in Tn₁ and Tn₂ cover 9.8-11 GHz and 17-18.2 GHz respectively. The electrical spectra of two radar signals measured by a spectrum analyzer (Keysight, N9030A) are shown in Fig. 5(b) and Fig. 5(c). At the same time, the optical reference signals are sent to Rns clockwise through the fiber ring. In each Rn, the echo reflected from the target modulates the corresponding optical reference signal in an MZM biased at the quadrature point. The modulated optical reference signal is filtered by a tunable OBPF (Yenista, XTM-50) and sent back to the signal processing unit clockwise. The transition spectra of OBPFs and the optical spectra of the modulated optical reference signals before and after optical filtering are shown in Fig. 6. In the signal processing unit, all modulated optical reference signals are sent to a PD for parallel de-chirp processing. The de-chirped signals are filtered by an ELPF and digitized by an oscilloscope (Keysight, DSO-X 92004A) for further processed off-line.

Fig. 5. (a) The optical spectra of the optical signals. (b) The electrical spectrum of radar signals in Tn₁. (c) The electrical spectrum of radar signals in Tn₂.

Download Full Size | PDF

Fig. 6. The transition spectra of OBPFs (red lines) and the optical spectra of the modulated optical reference signals before (light blue lines) and after (dark blue lines) optical filtering in Tn₁ (a) and Tn₂ (b).

Download Full Size | PDF

Experiments of the position and velocity measurements are carried out to verify the feasibility of the proposed distributed microwave photonic MIMO radar. The relative positions of Tns, Rns and a target are shown in Fig. 7 accompanied with the photo. The Tn₁ and Rn₁ are located at (0 m, 5.6 m), (0 m, 5.9 m). The Tn₂ and Rn₂ are located at (4.85 m, 0 m), (5.15 m, 0 m). The target is located in a rectangle area of 2 m×3 m. Limited by experimental condition, only the signal path Tn₁-target-Rn₁ and Tn₂-target-Rn₂ is demonstrated in the experiment. To simultaneously reflect radar signals from Tn₁ and Tn₂ to Rn₁ and Rn₂ respectively, the target is composed of two trihedral corner reflectors placed closely on a remote-controlled car.

Fig. 7. The relative position of antennas and the target.

Download Full Size | PDF

Firstly, the position measurement performance of a static target is verified. Before the measurement, a back-to-back test is carried out to calibrate the frequencies (2k(τ_txi-τ_rxi)) of de-chirped signals. Results show that the frequencies obtained by Tn₁-Rn₁ and Tn₂-Rn₂ are 20.91 MHz and 40.32 MHz respectively. Then, the measurement of a static target is performed. The spectrum of the de-chirped signal is shown in Fig. 8(a). Obviously, there are two peaks around 20.996 MHz and 40.414 MHz obtained by Tn₁-Rn₁ and Tn₂-Rn₂ respectively. The frequency deviations between the measured values and calibration values can be used to calculate R_{0_1} and R_{0_2}. Furthermore, the position of the target is obtained based on TOA. To investigate the accuracy of the proposed radar system, measurements are performed at different positions. The results and corresponding actual positions are shown in Fig. 8(b). The location is successfully achieved, and the measurement error is lower than 8 cm.

Fig. 8. (a) The spectrum of the de-chirped signal. (b) The measured positions and the actual positions of six measurements.

Download Full Size | PDF

Secondly, position and velocity measurements of a moving target are carried out. The target on the remote-control car moves along a straight line (Fig. 7, the blue line) at a velocity of around 1.6 m/s. The blurred cars represent the movement route of the target. The distances R_{0_i} and the radial velocities v_i are obtained by Tn₁-Rn₁ and Tn₂-Rn₂. Furthermore, the distances (blue diamonds) and velocities (red squares) in x direction and y direction are calculated, which are shown in Fig. 9(a) and Fig. 9(b) respectively. The actual distances (blue dotted line) and velocities (red dotted line) in x direction and y direction are also shown in Fig. 9(a) and Fig. 9(b) respectively. Obviously, the measured distances are consistent with the actual values. The measured velocities fluctuate around actual velocities, which may be caused by the vibration of the target during the movement of the car. Based on distances R_{0_i} and radial velocities v_i, the magnitude and direction of the velocity vector can be calculated through Eq. (18). As shown in Fig. 10, red diamonds and brown arrows represent the magnitudes and directions of the velocities, respectively. Blue dots, which are the projection of red diamonds on the x-y plane, represent the measured positions of the target. The red diamonds fluctuate around 1.6 m/s, with a maximum deviation of less than 0.20 m/s. The brown arrows steadily point to the target moving direction. The majority of blue dots fall onto the actual track of the target (blue line). The measurement results above are consistent with the state of the target motion, verifying the feasibility of the proposed distributed MIMO radar.

Fig. 9. The measured distances, measured velocities, actual distances and actual velocities of moving target in x (a), y (b) direction.

Download Full Size | PDF

Fig. 10. The measured magnitudes and directions of the velocities, measured positions, and actual track of moving target.

Download Full Size | PDF

4. Conclusion

A scalable distributed microwave photonic MIMO radar based on a bidirectional ring network is proposed, employing wavelength-division multiplexing, frequency-division multiplexing, and time-division multiplexing technologies. A 2×2 distributed microwave photonic MIMO radar operating in X and Ku bands is established. The position and velocity measurement results are consistent with reality. The measurement errors of position and velocity are better than 8 cm and 0.20 m/s respectively. Based on the bidirectional ring network and pluggable remote nodes, the proposed system can take full advantage of the spatial diversity and simultaneously reduce the system complexity, providing a promising solution for large-scale distributed MIMO radar with flexible scalability.

Funding

National Key Research and Development Program of China (2018YFA0701900, 2018YFA0701901); National Natural Science Foundation of China (61690191, 61701476).

Disclosures

The authors declare no conflicts of interest.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

References

1. A. M. Haimovich, R. S. Blum, and L. J. Cimini, “MIMO radar with widely separated antennas,” IEEE Signal Process. Mag. 25(1), 116–129 (2008). [CrossRef]

2. E. Fishler, A. Haimovich, R. S. Blum, L. J. Cimini, D. Chizhik, and R. A. Valenzuela, “Spatial diversity in radars—Models and detection performance,” IEEE Trans. Signal Process. 54(3), 823–838 (2006). [CrossRef]

3. Q. He, N. H. Lehmann, R. S. Blum, and A. M. Haimovich, “MIMO radar moving target detection in homogeneous clutter,” IEEE Trans. Aerosp. Electron. Syst. 46(3), 1290–1301 (2010). [CrossRef]

4. Q. He, R. S. Blum, H. Godrich, and A. M. Haimovich, “Target velocity estimation and antenna placement for MIMO radar with widely separated antennas,” IEEE J. Sel. Top. Signal Process. 4(1), 79–100 (2010). [CrossRef]

5. N. H. Lehmann, A. M. Haimovich, R. S. Blum, and L. Cimini, “High resolution capabilities of MIMO radar,” in 2006 Fortieth Asilomar Conference on Signals, Systems and Computers, (IEEE, 2006), 25–30.

6. E. Fishler, A. Haimovich, R. Blum, D. Chizhik, L. Cimini, and R. Valenzuela, “MIMO radar: An idea whose time has come,” in Proceedings of the 2004 IEEE Radar Conference (IEEE Cat. No. 04CH37509), (IEEE, 2004), 71-78.

7. N. H. Lehmann, E. Fishler, A. M. Haimovich, R. S. Blum, D. Chizhik, L. J. Cimini, and R. A. Valenzuela, “Evaluation of transmit diversity in MIMO-radar direction finding,” IEEE Trans. Signal Process 55(5), 2215–2225 (2007). [CrossRef]

8. H. Godrich, A. M. Haimovich, and R. S. Blum, “Target localization accuracy gain in MIMO radar-based systems,” IEEE Trans. Inf. Theory 56(6), 2783–2803 (2010). [CrossRef]

9. X. Song, P. Willett, and S. Zhou, “Jammer detection and estimation with MIMO radar,” in 2012 Conference Record of the Forty Sixth Asilomar Conference on Signals, Systems and Computers (ASILOMAR), (IEEE, 2012), 1312–1316.

10. D. E. Hack, L. K. Patton, B. Himed, and M. A. Saville, “Detection in passive MIMO radar networks,” IEEE Trans. Signal Process. 62(11), 2999–3012 (2014). [CrossRef]

11. M. Dianat, M. R. Taban, J. Dianat, and V. Sedighi, “Target localization using least squares estimation for MIMO radars with widely separated antennas,” IEEE Trans. Aerosp. Electron. Syst. 49(4), 2730–2741 (2013). [CrossRef]

12. G. Serafino, F. Scotti, L. Lembo, B. Hussain, C. Porzi, A. Malacarne, S. Maresca, D. Onori, P. Ghelfi, and A. Bogoni, “Toward a new generation of radar systems based on microwave photonic technologies,” J. Lightwave Technol. 37(2), 643–650 (2019). [CrossRef]

13. J. Fu, F. Zhang, D. Zhu, and S. Pan, “Fiber-distributed ultra-wideband radar network based on wavelength reusing transceivers,” Opt. Express 26(14), 18457–18469 (2018). [CrossRef]

14. F. Zhang, B. Gao, and S. Pan, “Photonics-based MIMO radar with high-resolution and fast detection capability,” Opt. Express 26(13), 17529–17540 (2018). [CrossRef]

15. F. Scotti, S. Maresca, L. Lembo, G. Serafino, A. Bogoni, and P. Ghelfi, “Widely Distributed Photonics-Based Dual-Band MIMO Radar for Harbour Surveillance,” IEEE Photonics Technol. Lett. 32(17), 1081–1084 (2020). [CrossRef]

16. F. Scotti, S. Maresca, L. Lembo, G. Serafino, A. Bogoni, and P. Ghelfi, “Dual-Band Radar System with Multiple Distributed Sensors for Coherent MIMO,” in 2020 21st International Radar Symposium (IRS), (IEEE, 2020), 347–350.

17. M. Zhang, Y. Ji, Y. Zhang, Y. Wu, H. Xu, and W. Xu, “Remote radar based on chaos generation and radio over fiber,” IEEE Photonics J. 6, 1–12 (2014). [CrossRef]

18. L. Lembo, S. Maresca, G. Serafino, F. Scotti, F. Amato, P. Ghelfi, and A. Bogoni, “In-field demonstration of a photonic coherent MIMO distributed radar network,” in 2019 IEEE Radar Conference (RadarConf), (IEEE, 2019), 1–6.

19. S. Maresca, G. Serafino, F. Scotti, F. Amato, L. Lembo, A. Bogoni, and P. Ghelfi, “Photonics for Coherent MIMO Radar: an Experimental Multi-Target Surveillance Scenario,” in 2019 20th International Radar Symposium (IRS), (IEEE, 2019), 1–6.

20. M. Maresca, D. S. Jacome, L. Lembo, F. Scotti, G. Serafino, A. Malacarne, C. Rockstuhl, P. Ghelfi, and A. Bogoni, “Photonics-enabled 2Tx/2Rx coherent MIMO radar system experiment with enhanced cross range resolution,” in Optical Fiber Communication Conference, (Optical Society of America, 2020), Th2A. 41.

21. R. Randhawa and J. Sohal, “Comparison of optical network topologies for wavelength division multiplexed transport networks,” Optik 121(12), 1096–1110 (2010). [CrossRef]

22. S.-K. Liaw, S.-L. Tzeng, and Y.-J. Hung, “Rayleigh backscattering induced power penalty on bidirectional wavelength-reuse fiber systems,” Opt. Commun. 188(1-4), 63–67 (2001). [CrossRef]

23. T. Yao, D. Zhu, D. Ben, and S. Pan, “Distributed MIMO chaotic radar based on wavelength-division multiplexing technology,” Opt. Lett. 40(8), 1631–1634 (2015). [CrossRef]

Scalable distributed microwave photonic MIMO radar based on a bidirectional ring network

Abstract

1. Introduction

2. Principle

2.1 Architecture of the proposed radar system

2.2 Two-dimensional position and velocity measurements method based on the proposed radar system

3. Experiment and results

4. Conclusion

Funding

Disclosures

Data availability

References

Data availability

Cited By

Figures (10)

Equations (18)

Optics Express