Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

ICTON 2020 Th.A1.

Towards an Integrated Photonics-Based Radar


Paolo Ghelfi1, Filippo Scotti1, Fabio Falconi2, Claudio Porzi2, Suzanne Melo2, and Antonella Bogoni1,2
1
PNTLab, CNIT, via Moruzzi 1, Pisa, Italy
2
TECIP Institute, Scuola Superiore Sant’Anna, via Moruzzi 1, Pisa, Italy
Tel: +39 050 882237, e-mail: paolo.ghelfi@cnit.it
ABSTRACT
The paper reviews the recent activities towards the implementation of a photonics-based radar as a packaged
integrated photonic device. Recent experimental results are presented, obtained with a radar implemented as
a photonic integrated circuit in a Silicon-on-Insulator platform. The potentials and open issues of using photonic
integration and its related packaging for flexible radar systems are also discussed.
Keywords: microwave photonics, radar, photonic integration.

1. INTRODUCTION
In the last few years, photonics has been proposed as a powerful tool for the development of microwave systems,
and in particular of radar systems. In fact, the wide bandwidth of photonics has proved to enable an
unprecedented frequency flexibility in radars, allowing the generation and detection of microwave signals across
extremely large frequency ranges, from few GHz up to 40 GHz and beyond [1,2], thus promising the
implementation of a new generation of general-purpose, frequency-flexible and software-defined radar
transceivers.
Besides the improved flexibility, photonics can potentially bring to radars a reduction of size, weight and
power consumption (SWaP), thanks to the much shorter wavelengths used. This trend becomes largely more
evident if integrated photonic technologies are taken into account, which can enable the implementation of very
small surveillance systems to be used in demanding applications (e.g., on drones or other unmanned vehicles).
So far, only few examples of integrated photonics-based RF transceivers have been presented [3, 4].
In this paper, we present our recent activities towards the implementation of an integrated photonics-based
radar, reporting experimental results on a radar photonic integrated circuit (PIC) implemented in the Silicon-on-
Insulator (SOI) platform. The potentials and open issues of the implementation approach based on photonic
integrated technologies are also discussed.

2. ARCHITECTURE OF THE INTEGRATED PHOTONICS-BASED RADAR


We have designed our integrated photonics-based radar considering the SOI technological platform, because of
its capability to integrate passive components, electro-optical modulators and photodiodes with well-established
maturity. Figure 1 reports the architecture of the designed PIC, which has been realized through a multi-project
wafer service at IMEC.

DAC

TX GC
MZM MMI

MMI PD RF Clock x2
GC
Laser CK

PD ADC

MZM
RX
PIC
Radar Target Emulator 1km
LNA fiber

VA RF filter PD

Figure 1. The architecture of the integrated-photonics radar. The figure also reports the schemes for the clock
generation and for the radar target emulation.

978-1-7281-8423-4/20/$31.00 ©2020 IEEE 1

Authorized licensed use limited to: Middlesex University. Downloaded on November 04,2020 at 07:46:18 UTC from IEEE Xplore. Restrictions apply.
ICTON 2020 Th.A1.5

Since the SOI platform cannot implement light sources, an external laser is considered to be launched into the
PIC through a vertical grating coupler (GC). The laser is split by a 1x2 multimode interference splitter (MMI) to
feed both the transmitter (TX) and the receiver (RX) sections of the radar. The two sections are almost identical,
each composed of a Mach Zehnder modulator (MZM) followed by a Si-Ge photodiode (PD). In the case of the
TX, an MMI before the PD allows to monitor the optical signal out of the PIC through a GC for debugging
purposes.
In both the TX and RX, one arm of the MZM is driven by an RF clock signal that aims at generating an optical
comb inside the MZM. In order to do so, a reference clock (CK) at frequency fCK is split into two replicas, with
one of them frequency multiplied by 2, and then recombined together. The RF clock signal thus obtained is
composed of two tones at fCK and 2fCK, and it is amplified and split to drive one arm of the MZM in both the TX
and RX sections. Thanks to the high modulation voltage of the RF clock signal, several optical modulation
sidebands are generated inside the MZMs at an optical detuning of ±fCK, ±2fCK, ±3fCK, ±4fCK, etc. from the laser.
In the TX MZM, the second arm is driven by a signal centered at intermediate frequency (IF) fIF, generated by
a digital-to-analog converter (DAC). At the TX PD, the optical components generated by the RF comb beat with
the optical sidebands generated by the DAC, implementing the RF up-conversion of the IF signal to several
multiples of the clock frequency, at fRF = n·fCK ± fIF. After the TX PD, specific RF front ends can filter and
amplify the selected up-converted radar signal. More than one signal at IF can be applied simultaneously, so that
multiple RF signals can be generated at the same time. Similarly, the received radar echo at fRF (or multiple
echoes at different RF frequencies) drives the second arm of the RX MZM, and the RX PD implements the
down-conversion to the corresponding fIF, which can be acquired by an analog-to-digital converter (ADC).

3. CHARACTERIZATION OF THE RADAR PIC


In the following characterization experiments, we use an external tunable laser source with 15dBm output power.
The optical signal is first sent to a polarization controller (PC) to enter the chip with TE mode and then it is
coupled to the PIC through the input GC, the loss of which has been measured in 3 dB. All the electrical signals
driving the MZMs are applied to the PIC through bias-tees to allow proper biasing of the p-n junctions. The
reference CK signal is provided by a synthesizer at fCK = 3.5 GHz.
Figure 2a reports the RF spectra produced at the TX PD when the DAC is set to generate a plain radar pulse
with a duration of 2 —s every 40 —s, centered at fIF = 100 MHz. As can be seen, clear replicas of the IF signal are
generated at 3.4 GHz (S band), 6.9 GHz (C band), 10.4 GHZ (X band), and 14.1 GHz (Ku band), confirming the
capability of the PIC to up-convert the radar signal to several output frequencies generated as a multiplication of
the CK signal. The RX section implements the complementary function with similar performance.
Figure 2b shows the two-tones analysis of the RX section for the S and X bands. The graph shows the values
of the first harmonic and of the third-order intermodulation distortion. The level of the thermal noise of the PD is
also reported. The RF-to-IF conversion loss in X band turns out to be 59 dB, with a spurious-free dynamic range
(SFDR) of 99dB·Hz2/3. For the S band instead, the conversion loss is 55 dB, and the SFDR results to be
100dB·Hz2/3. The two analysed bands present similar results, confirming again the frequency flexibility of the
radar PIC. However, the measured values for the conversion losses are rather low: we will discuss this point in
the next paragraph.
The availability of the GC after the TX MZM gives us the possibility to test the PIC as a radar system, by
exploiting a fiber-based radar target emulator. The optical signal is extracted from the TX section of the PIC at
the GC, delayed in an optical fiber spool of 1km, and sent to an external PD (with performance similar to the
integrated one). After RF-filtering the radar signal at the desired frequency, a cascade of low-noise amplifiers

-60 -60
S Band C Band
Potenza[dBm]

[dBm]
[dBm]

[dBm]

-80 -80
Potenza

-100 -100
Power

Power

-120 -120

-140 -140
3.394 3.396 3.398 3.4 3.402 3.404 3.406 6.894 6.896 6.898 6.9 6.902 6.904 6.906
Frequenza [GHz]
Frequency [GHz] Frequenza [GHz]
Frequency [GHz]

-60 -60
X Band ku Band
Potenza [dBm]

[dBm]
Potenza[dBm]

-80
[dBm]

-80

-100 -100
Power

Power

-120 -120

-140 -140
10.394 10.396 10.398 10.4 10.402 10.404 10.406 14.094 14.096 14.098 14.1 14.102 14.104 14.106
Frequenza [GHz] Frequenza [GHz]
Frequency [GHz] Frequency [GHz]

a) b)
Figure 2: a) RF signal generation in the S, C, X and Ku bands; b) System SFDR.

Authorized licensed use limited to: Middlesex University. Downloaded on November 04,2020 at 07:46:18 UTC from IEEE Xplore. Restrictions apply.
ICTON 2020 Th.A1.5

(LNAs) boosts the signal, and a variable attenuator (VA) allows varying the RF power at the input of the RX
section in the PIC. With this setup, we characterize the system in terms of sensitivity, i.e., we measure the signal-
to-noise ratio (SNR) of the cross-correlation of the received echo signal, calculated after the down-conversion
and digitization in an ADC, as reported in Fig. 3a. The test is carried out using 2 —s-long linearly chirped pulses
with a bandwidth of 100 MHz and integrating the echoes for 20 ms. Considering a minimum SNR of 10dB, the
minimum detectable echo power is -52 dBm and -38 dBm for the S and X bands respectively. It is to be noticed
that the echo power values are those directly measured at the PIC input, without considering any receiving
electrical front-end, so the sensitivity should strongly improve with the usual amplification stages of a radar
system.
Figure 3b shows an example of detection in the X band, performed with the setup described above. In the
graph, the orange curve is the cross-correlation of the detected signal after the propagation in the fiber spool,
while the blue curve is the case when adding a 2m-long fiber patchcord. As can be seen, the system correctly
detects the two targets as separated by 1.5 m, which is the free-space equivalent range distance of the 2 m-fiber
roundtrip delay. The radar range resolution can be clearly observed in the inset: in our case, the 1.5 m resolution
given by the 100 MHz bandwidth of the DAC is confirmed.

4. COMMENTS AND CONCLUSIONS


The scheme proposed above clearly shows the potential of implementing a frequency-flexible, miniaturized
radar system exploiting integrated photonics. However, the conversion loss is rather high, significantly higher
than that of standard RF mixers (which is usually as low as 10 dB). This also reduces the dynamic range, which
is one of the most important parameter in a radar system, thus it dampens the effective use of such technology in
real field applications.
There are several factors affecting the conversion loss of a microwave-photonics system [5]. When dealing in
particular with integrated microwave photonics, one of the main factors is the optical loss of the PIC. As a rule
of thumb, every dB of optical loss becomes 2 dB of RF system loss, and the scheme presented here is no
exception. As described above, in its current design the PIC has one MMI between the MZM and the PD at the
TX section, that is not strictly necessary. Simply removing this MMI could allow improving the conversion loss
by 6 dB. Similarly, any optical loss reduction lowers the conversion loss, and this applies to any integrated
device in the system: GCs, MZMs, splitters, waveguides.
As it is evident, the optical power of the laser source is another parameter that strongly affects the amount of
RF or IF power obtainable at the PDs output. The higher the laser power, the lower the conversion loss (and still
the rule 1 dB-in-optics = 2 dB-in-RF holds). The limit here is given by the two-photon absorption (TPA), the
nonlinear attenuation due to the light-matter interaction when the optical signal is tightly confined, as in the SOI
platform. In the experiment above, we have been using a laser at 15 dBm, already close to the measured TPA
threshold of 17 dBm. Anyway, some “tricks” could be used to increase the acceptable optical power, as placing
the first MMI closer to the input GC so that the length of waveguide where TPA can take place (before the
optical power is split at thus reduced) is as short as possible.
Another very important parameter is the Vpi of the MZMs, i.e., the voltage required to shift the phase of the
optical signal by pi. In practice, the Vpi indicates the effectiveness in loading an RF signal on an optical carrier,
and if the Vpi halves, the conversion loss reduces 6 dB. In our PIC we have used the MZM design from the IMEC
library, which shows a Vpi of about 10 V, but some room for improvement is there, e.g., increasing the MZM
length to improve the effectiveness of the phase shift in the p-n junction. On the other hand, longer MZMs also
show higher insertion loss, and reduced electro-optical bandwidth. Therefore, a trade-off must be sought.
A relevant comment is related to the integration platform technology. As we have said above, the SOI platform

a) b)
Figure 3: a) System sensitivity in S and X bands; b) Target detection with the radar PIC.

Authorized licensed use limited to: Middlesex University. Downloaded on November 04,2020 at 07:46:18 UTC from IEEE Xplore. Restrictions apply.
ICTON 2020 Th.A1.5

does not have gain elements, but recent advancements are providing the capability to hybrid integrate
semiconductor optical amplifiers (SOAs) in Indium Phosphide on top of the SOI. Therefore, this technology can
potentially allow the implementation of more complex schemes (e.g., including filters and better comb
generators, and even the laser). Still these technologies are not well established nor easily available, but they are
improving very fast, opening several interesting possibilities for the next future.
Finally, a fundamental step for moving the radar PIC closer to a real field application is the packaging. It is
evident that a PIC needs to be protected in a package in order to ensure its compliance to any class of
environment tougher than the lab, but this operation implies facing and solving specific problems. In the case of
the radar PIC, the package shall minimize the losses of the RF and optical interfaces, as well as guarantee the
frequency flexibility of the PIC. Moreover, the packaging operation shall reduce or contain the RF crosstalk
between the TX and RX sections of the PIC. In fact, the two sections are confined in a very small area of few
mm squared, where RF contacts distribute signals as high as few V. In our characterization, we measure
a crosstalk of -80 dB between the TX and RX sections, but this value could be not enough since it is lower than
the dynamic range. Anyway, in our system experiment such a level of crosstalk is not a problem because it is not
high enough to saturate the RX amplification stage, and the false echo generated by the crosstalk can be easily
filtered out digitally to avoid false target detections.
Another practical consideration related to packaging is about the use of optical fibers. In particular, it could be
very attractive not to have any optical fiber outside of the package, in order to avoid any risk of damage and to
simplify the installation of the PIC on an electronic board. This goal could be actually reached if the laser source
is internal to the package: either with hybrid co-packaging (i.e., a bare laser aligned to the SOI PIC), or with
a technological platform that includes the gain medium.
In conclusion, the radar PIC presented here shows the potentials of using integrated microwave photonics
in the field of radar systems, demonstrating a large frequency flexibility with interesting performance. Still
several technical details have to be faced, but several possible solutions can be foreseen. Therefore, this PIC
works as a fundamental exercise towards the implementation of an integrated photonics-based radar system.

ACKNOWLEDGEMENTS
This work has been funded by the Italian Ministry of Defence under the project MULTIRADAR-PIC, contract
number 1992.

REFERENCES
[1] P. Ghelfi et al., “A fully photonics-based coherent radar system,” Nature, vol. 507, no. 7492, pp. 341-345,
Mar. 2014.
[2] F. Scotti et al., “In-field experiments of the first photonics-based software-defined coherent radar,” Journal
of Lightwave Technology, vol. 32, no. 20, pp. 3365-3372, Oct. 2014.
[3] M. Chen, H. Yu, B. Yang, Y. Li, H. Chen, and S. Xie, “A silicon integrated microwave-photonic
transceiver,” in Proc. 2017 Optical Fiber Communications Conference and Exhibition (OFC), Mar. 2017,
pp. 1-3.
[4] S. Li, Z. Cui, X. Ye, J. Feng, Y. Yang, Z. He, R. Cong, D. Zhu, F. Zhang, S. Pan, “Chip-based photonic
radar for high-resolution imaging,” https://arxiv.org/ftp/arxiv/papers/1905/1905.12802.pdf
[5] V. J. Urick, K. J. Williams, and J. D. McKinney, “Fundamentals of microwave photonics,” John Wiley &
Sons, 2015.

Authorized licensed use limited to: Middlesex University. Downloaded on November 04,2020 at 07:46:18 UTC from IEEE Xplore. Restrictions apply.

You might also like