Open Access
Add to BibSonomy
Abstract
A group-delay-unit-based integrated silicon photonic integrated circuit (PIC) is employed as a reconfigurable analog radio frequency decoder, which provides a real-time temporal and spectral analysis of any arbitrary multi-tone signal in the micro- and mm-wave range. The circuit is based on cascaded Mach-Zehnder interferometer embedded silicon microring resonators as variable delay units. The temporal decoding of the multi-tone input signal is demonstrated by tuning the signal with respect to the ring resonator delay and resonance. A one-to-one conformal time-to-frequency mapping provides real-time spectral decoding of the signal under test without additional digital signal processing. The idea is validated by several experimental results with single-tone and two-tone input signals in a compact, low-power, silicon PIC. The proposed real-time temporal analog frequency decoder may be very intriguing for high-speed, low-latency wireless applications, such as autonomous driving and 6G.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Real-time spectral sensing is the process of continuously monitoring the radio spectrum to detect and identify available white spaces (not currently used radio bands at a distinct position), which is crucial for the efficient utilization and allocation of accessible communication bands [1]. Such a real-time signal analysis in the temporal and spectral domains might be especially important for Radar, Lidar, and various medical, 5G, and upcoming 6G related applications [2,3]. To ensure reliable, uninterrupted, and low-latency service, a dynamic real-time spectrum sensing scheme requires real-time scanning and detection over a wide frequency range. For low-frequency bands, this may be possible by using electronic-based high-speed digital signal processing (DSP). However, especially for higher carrier frequencies and bandwidths, this leads to substantial computational and economic penalties. Additionally, electronic DSPs have a limited operating range and suffer from electromagnetic interference (EMI). Photonic-based spectrum sensing, instead, can offer a wide bandwidth, high measurement precision, and accuracy, immunity to EMI, and low-latency performance [4].
Therefore, various photonic-assisted spectrum sensing systems have been extensively investigated in the last decade [4–14]. Especially, spectrum sensing based on frequency-to-time mapping (FTTM) [4,6–10] can outperform approaches based on an amplitude comparison function (ACF) [11,12] or frequency-to-power mapping [13,14]. The performance of spectrum sensing systems can be compared by various parameters such as resolution, accuracy, and measurement range, whereas different parameters are more or less important for various applications [15,16]. Each application has unique requirements that determine the emphasis on specific performance parameters. For instance, a wide measurement range and high accuracy are essential in spectrum monitoring and management. Cognitive radar systems with critical parameters of interest include resolution, accuracy, and autonomous driving, and electronic warfare systems require real-time implementation of frequency measurement systems.
Stimulated Brillouin scattering-based spectrum sensing systems can provide a very large measurement range and sub-MHz accuracy [7,8], but the resolution is limited by the bandwidth of the Brillouin gain spectrum. However, when combined with heterodyne detection, an attometer resolution can be achieved [9]. But, specific chip platforms are required for integration, leading to an enhanced complexity. For an integrated solution without special platforms, a two-step filtering with an integrated ring resonator and heterodyne detection can be used. This significantly improves the measurement accuracy to $\pm$ 0.4 MHz for a high measurement range [4]. However, the increased data acquisition time limits the real-time analysis for time-varying broadband signals. A fully integrated reconfigurable photonic spectrum sensing system that can provide a real-time signal analysis with low system latency and complexity, which can also operate over a wide measurement range from radio to millimeter waves, would be of great interest.
In this paper, we present a fully integrated, power-efficient compact PIC for real-time temporal and spectral analysis of micro and mm-wave signals. The PIC is based on cascaded tunable delay blocks implemented by Mach-Zehnder interferometer (MZI) with embedded ring resonators. Separate integrated thermal heaters provide proper adjustment of the delay and resonance. Once the chip is properly adjusted, the measurement can be carried out in real-time without thermal tuning.
The basic concept was first proposed with simulation results in our previous work [17]. Here, we present the fabricated PIC on a standard 220 nm silicon-on-insulator (SOI) platform with a footprint of 2 mm$^2$ and show experimental results for several single-tone, as well as multi-tone input signals up to 26 GHz. As the proposed PIC can resolve (or decode) all input frequency components in real-time without any sophisticated signal processing, it can be described as a ’photonic frequency decoder - PIC (PFD-PIC)’. To the best of our knowledge, this is the first practical on-chip demonstration of real-time temporal frequency decoding in the mm-wave domain.
2. Photonic frequency decoder (PFD)
2.1 Concept
The schematic concept of the photonic radio wave frequency decoder is shown in Fig. 1. A signal under test (SUT) consisting of three unknown frequency components ($f_1$, $f_2$, and $f_3$), as shown in the frequency and time domains in Fig. 1(a, b), is convoluted in time by a Gaussian-shaped pulse (Fig. 1(c)) to avoid any time-domain overlapping between the delayed signals. This multiplication between the signal and the pulse is considered as a coded signal and was the primary motivation behind labeling the system as a Decoder. For this time restriction, a pulse source like a mode-locked laser (MLL) [18] or a continuous wave laser diode together with an electro-optic modulator [19] can be used. The SUT is then converted into the optical domain by an electrical-to-optical conversion with a modulator (E/O), Fig. 1(d) [20]. The output of the modulator in the frequency domain is the carrier ($f_c$) modulated with the unknown signal components as upper sideband (USB) and lower sideband (LSB). Using an optical bandpass filter (OBPF), the carrier together with the USB or LSB can be selected before the signal is fed into the PFD-PIC. For the USB, the higher frequency components will experience a smaller delay compared to the lower frequencies, as shown in Fig. 1(e), while for the LSB it is the other way round, as depicted in Fig. 1(g). The width of the pulse envelope restricting the SUT is defined by the lower frequency component ($f_1$), such that the convoluted output of the modulator (E/O) should have at least two full cycles of $f_1$ (as shown in Fig. 1(f) and (h)), so that it can be clearly detected. The repetition rate of the pulse envelope is determined by the maximum overall group delay of the discrete frequency component. To avoid any temporal overlap, the repetition rate of the modulating pulse must exceed the maximum time delay.
Fig. 1. Schematic illustration of the proposed photonic radio wave frequency decoder concept. The signal under test (SUT) with the unknown frequency components $f_1$, $f_2$, and $f_3$ is shown in (a), and the corresponding time-domain trace is shown in (b). First, the SUT is restricted in time by a pulse modulation (c). The input signal to the photonic frequency decoder - PIC (PFD-PIC) is shown in green (d). By an optical band pass filter (OBPF), the upper (USB) or lower sidebands (LSB) of the modulation are filtered out and thus, the negative or positive chirp of the PFD-PIC can be selected. Dependent on the selection, each signal frequency in the SUT experiences a different group delay (USB: (e), (f)) and LSB: (g), (h)). PG: pulse generator; E/O: electrical to optical conversion; PD: photodiode; OSC: oscilloscope.
The proposed PFD-PIC consists of four cascaded MZI embedded race-track type ring resonator structures (group-delay-units, GDU) as shown in Fig. 2. The cascaded structures provide discrete group delays to the input frequencies and lead to a real-time temporal separation of the frequencies in the SUT, as depicted in Fig. 1. The group delay engineering for each individual component is determined by the number of these sub-units. By using $N$ number of delay units, it becomes feasible to create a discrete delay profile for at most $N$ frequency components, resulting in a continuous variation of the overall group delay with respect to frequency. The optical output is converted to the electrical domain by a photodiode (PD), and the temporally resolved signal can be visualized by an oscilloscope (OSC) in the experiment, or by a power-time measurement system, without any additional signal processing. In either case (USB or LSB), each unknown signal component can be mapped to a specific group delay for a given PFD-PIC.
Fig. 2. Layout of the PFD-PIC fabricated on a 220 nm SOI platform (top), with four cascaded group delay units (GDU) consisting of a Mach-Zehnder interferometer (MZI) and a ring resonator. The DC pads are used for tuning the phase shifters of the MZI and the ring, to obtain a specific group delay and to overlap the resonances. (a) Microscopic image of the bonded PFD-PIC (red box) on a printed circuit board, and (b) zoomed-in view of a single GDU.
2.2 PFD-PIC design and tuning
The photonic chip is designed with four race-track type silicon ring resonators connected in series. All the rings are coupled to an MZI structure with their main through port. To tune the resonances and delays, two integrated metal heaters are connected with each ring-MZI structure. The MZI heater is used for delay tuning, whereas the integrated ring heater serves for resonance tuning. The detailed tuning procedure for such a system was discussed in [17]. The photonic chip was fabricated on a standard 220 nm SOI silicon photonics technology platform [21]. The thermal heaters have a length of 250 $\mu$m and require around 40 mW of DC power for a full free spectral range (FSR) tuning. The power consumption and accuracy of the integrated heaters can be optimized by adopting special design approaches such as air trenches, deep trenches or by implementing undercut trenches [22,23].
The four GDUs R-1, R-2, R-3, and R-4 are shown in Fig. 2. For independent tuning, each ring and MZI has its own DC power supply pads for the heaters. The wire-bonded PFD-PIC on a printed circuit board with its DC control pads and a microscopic image of one of the group delay units are presented in Fig. 2(a) and (b), respectively. Figure 3 presents the optical spectrum at the output of the PFD-PIC, measured with an optical spectrum analyzer (OSA) and a broadband noise source as input. In Fig. 3(a), R-1 and R-2 are in resonance, whereas R-3 and R-4 are out of resonance. By tuning the heaters (ON), all three resonances coincide, as shown in Fig. 3(b). Therefore, a much higher extinction ratio (ER) of 11 dB with an FSR of 49 GHz can be achieved. The achievable delay is determined by the steepness of the resonance or the Q-factor of the resonators. Therefore, the delay can be adjusted by tuning the phase shifters of the MZI while keeping the fixed free spectral range [17]. It should be noted that, in the presented concept, the frequency-to-time mapping function is established by predefining the tuning of individual resonances relative to the unknown signal components, resulting in real-time discrete delays for different components. However, in the fabricated device, the delay caused by individual resonances alone was not sufficient to simultaneously discriminate between multiple components in the temporal domain. As a result, the four resonances were aligned to achieve a coherent delay that was high enough to observe the delay in the proof-of-concept results presented in the subsequent section. The thermal crosstalk between the rings can cause an undesired resonance shift for the cascaded structure. However, this can be mitigated by implementing deep air trenches isolation, as discussed in [23–25]. Furthermore, direct delay tuning can be achieved by monitoring group delay in the optical domain using an optical vector network analyzer (OVNA). Such a direct delay tuning provides additional flexibility to the proposed PFD-PIC-based time domain frequency decoding system.
Fig. 3. Optical output spectrum of the PFD-PIC for a broadband noise source as input. (a) The resonances of R-1 and R-2 are the same but different from R-3 and R-4. (b) After thermal tuning, all resonances are aligned at a single frequency with a higher extinction ratio (ER) of 11 dB and almost the same free spectral range (FSR) of 49 GHz.
3. Experiment and results
The proof-of-concept experimental setup for characterizing the PFD-PIC is shown in Fig. 4. A tunable fiber laser (FL, NKT Photonics) served as the optical source to characterize the PFD-PIC. The FL has provided a wider range of tunability, as an alternative, a laser diode can be used. The output of the FL was modulated with a Gaussian pulse by a Mach-Zehnder modulator (MZM-1). The pulse with a full-width at a half maximum duration of 200 ps was generated by an arbitrary waveform generator (AWG, Tektronix 70001A). To resolve the two frequencies in the time domain, the duration of the pulse should be in the range of the achievable delay, and it should be long enough to contain at least two complete periods of the lowest frequency in the SUT. The pulses were multiplied with the signal under test (SUT), generated with a signal generator (SG, Agilent Technologies E8257D), by another MZM-2. A waveshaper (WS, Finisar 1000s) was incorporated to filter out the upper or lower sideband (USB, or LSB), according to the desired chirp selection. This filtered signal was coupled to the PFD-PIC chip. The output was split for a 1% optical spectrum monitor signal, and the residual 99% was converted to the electrical domain by a photodiode (PD, Finisar XPDC2120R). The time domain signal was measured and analyzed by an electrical oscilloscope (OSC, Agilent 86100C). To compensate for losses of the chip and the modulators, erbium-doped fiber amplifiers (EDFA, LiComm OFA-TCH) were used, followed by bandpass filters (BPF) to suppress the spontaneous emission noise (ASE). Additionally, polarization controllers (PC) were incorporated to align the polarization of the optical field for the modulators and the chip.
Fig. 4. Schematic illustration of the experimental setup for characterizing the PFD-PIC. FL: Fiber laser, PC: polarization controller, MZM: Mach-Zehnder modulator, AWG: arbitrary waveform generator, EDFA: erbium-doped fiber amplifier, BPF: bandpass filter, SUT: signal under test, SG: signal generator, WS: wave shaper, PD: photodiode, OSC: oscilloscope, OSA: optical spectrum analyzer.
To test the PFD-PIC, a single-tone radio frequency (RF) signal of 10 GHz was used. After modulation with the SUT, the USB was filtered out by the WS. The resulting input signal to the PFD-PIC is shown in black in Fig. 5(a). For comparison, the resonance spectrum of the chip is depicted with a blue dotted line in Fig. 5(a). As can be seen, the USB is not in the resonance region. Therefore, this out-of-resonance signal can be used as a reference in the time domain, as depicted with the black trace in Fig. 5(b). To delay the signal, the laser wavelength was tuned so that the USB corresponds to the resonance. The red trace in Fig. 5(a) shows the output spectrum after tuning. Due to the resonance, the USB experiences a time delay of 51 ps, as depicted by the blue trace in Fig. 5(b). At the same time, the signal amplitude is reduced due to the resonance attenuation. The predefined delay can be mapped to a 10 GHz signal by a frequency-to-time mapping.
Fig. 5. (a) Input optical spectrum of the SUT before (black) and output spectrum after (red) aligning the USB to the chip resonance (shown in the grey box). (b) The corresponding reference for a non-alignment (black) time limited by a Gaussian pulse, and the 51 ps delayed signal after alignment (blue). This predefined delay can be mapped to a 10 GHz signal by frequency-to-time mapping.
In further experiments, two single-tone RF signals of 20 GHz and 26 GHz were tested separately to demonstrate the measurement range of the chip. First, the 20 GHz and 26 GHz reference signals were measured as described above, and the results are shown by the black traces in Fig. 6(a) and (b), respectively. After aligning to the resonance, the 20 GHz component experienced a total delay of 49 ps (blue trace in Fig. 6(a)), and the delay of the 26 GHz frequency was 46 ps (orange in Fig. 6(b)). The delay can be varied by tuning the thermal phase shifters in the MZI arms. Since for the delay, a ring resonance has been used, and the signals were attenuated in both cases. The quality of the system accuracy in terms of spectral alignment and the group delay could be enhanced by adapting high-quality integrated phase shifters [22,23].
Fig. 6. The reference signal at 20 GHz (black), and the delayed signal after the PFD-PIC for 20 GHz (blue, (a)) and for 26 GHz (orange, (b)).
Finally, the chip’s performance was also tested by a two-tone signal. An electrical combiner (Mini-Circuits) was used, to sum up a 20 and 26 GHz frequency before modulation. The frequency separation was tested by tuning the laser, so that one of the SUT frequencies was aligned close to the resonance. The filtered input signal spectrum to the chip is shown in Fig. 7(a) (black), and the corresponding time domain reference signal is depicted in Fig. 7(b) (black). First, the higher frequency component (26 GHz) was aligned close to the ring resonance as shown in Fig. 7(a) (red). Due to the resonance, the 26 GHz frequency experiences a higher delay than the 20 GHz component, which is still away from the resonance, as demonstrated in 7(d). As shown, the two frequency components can be clearly discriminated in the time domain after propagation through the chip without any post-processing. Note that the alignment of the laser to the resonance is not necessary for the discrimination. To discriminate the unknown frequencies, the resonance alignment can be performed before the measurement, allowing real-time measurement. Additionally, instead of tuning the laser, an RF offset frequency for the modulation may be used to tune the frequencies in and out of the resonance. To show the flexibility of the measurement, in the next experiment, the 20 GHz frequency was tuned close to the ring resonance (Fig. 8(a)). The output signal and the spectrum after the chip are shown in Fig. 8(b). As expected, the 20 GHz component is delayed (and attenuated) and can be differentiated from the 26 GHz component in real-time.
Fig. 7. (a) Input spectrum of the two-tone signal away from resonance (black), and after tuning the laser wavelength to the resonance (red). (b) Corresponding time-domain reference signal multiplied with a Gaussian-shaped pulse. (c) Spectrum of the delayed output when the 26 GHz component is aligned close to the resonance. (d) Signal after the chip. The 26 GHz component is highlighted by the orange box and the non-delayed 20 GHz frequency is highlighted in blue.
Fig. 8. (a) Output signal after the chip, when the 20 GHz frequency is tuned close to the resonance. (b) The spectrum of the output signal.
4. Discussion
We have presented the first proof-of-concept experimental results for a multi-tone frequency measurement by delaying one frequency of a SUT against the other in a PFD-PIC. The overall group delay of 50 ps exhibited by the first generation of the fabricated PFD-PIC was sufficient to delay a single component and observe real-time temporal discrimination. However, as described in the simulation study [17], the number of sub-units ($N$ ring-assisted MZI) determines the ability to discriminate multi-tone signal components ($N$) in real-time by defining a discrete delay for each individual frequency component of the SUT. This can be achieved by adjusting the group delay of the device resonances independently. By applying suitable discrete delays ($t_1$, $t_2$, $t_3$) to each of the three signal components ($f_1$, $f_2$, $f_3$), it becomes feasible to locate the temporal envelope at each instant in real-time, as described in the concept Fig. 1. If the frequencies are known in advance, the frequency-to-time mapping can be established for each individual signal component within the resolution and measurement range by a pre-adjustment of the ring resonance to the frequency to measure. This would result in a higher number of frequency tones that can be measured in real-time.
The ring resonance has to be tuned to the frequency to be measured by thermal tuning. Alternatively, the laser wavelength can be tuned (which was done in the experiments), or the sidebands are tuned to the resonance by an additional RF offset. The last option in particular may offer the highest tuning speeds. In this case, however, the measurement time depends on the tuning.
The measurement range of our method depends on the FSR and, therefore, on the ring diameter. Here we have shown test signals up to 26 GHz, but since the FSR is 49 GHz, the presented PFD-PIC allows the measurement up to the millimeter wave range. Higher frequencies up to the THz range may become possible for smaller rings. The PFD group delay profile defines the resolution of the proposed method. To enhance the measurement resolution, a higher number of sub-units can be used. Additionally, drop ports for the ring structures and heterodyne detection [4] with electrical filtering can be implemented to prevent any undesired mixing products in the time domain. A silicon nitride (Si$_3$N$_4$) platform would offer low propagation losses and ultra-high Q-factor for the resonators [26], resulting in larger aggregated delay and finer temporal mapping of weaker signal components. One of the challenging factors in this segmented network might be on-chip losses, which could be addressed by employing silicon-compatible on-chip amplifiers [27]. With these new structures and platforms, it may be feasible to achieve a frequency resolution of a few hundred MHz over a sensing range of several GHz [17]. However, for the fabricated PFD-PIC, the achievable resolution was determined to be 1 GHz based on the resonance response shown in Fig. 3(b). The maximum group delay obtainable with the fabricated structure was 50 ps which corresponds to the achievable time discrimination of multi-tone signals at 20 and 26 GHz.
The accuracy of the measurements depends on the precise alignment of the ring resonances to obtain an accurate frequency-to-time mapping. Thermal crosstalk has a significant impact on densely packed structures, leading to measurement errors that can easily be avoided by deep trenches to provide thermal isolation for each delay unit of the PFD-PIC [23–25].
Table 1 shows a comparison of various photonic-assisted frequency analysis systems presented in the literature. For the presented method, the measurement range is not limited by a loop delay [6], the Brillouin frequency shift [7], or a high dispersion in a long fiber [28]. Instead, it is only restricted by the free-spectral range of the cascaded rings in the PFD-PIC.
Table 1. Comparison of different photonics-assisted frequency analysis systems
For a real-world application of the proposed frequency analysis system, it should have an economic power consumption and a small footprint. In principle, the proposed photonic processing block is passive and does not add any pervasive electrical power load if implemented in the conventional microwave photonic system. However, the DC power consumed for the full FSR tuning of the four resonators results in a power consumption of around 160 mW. The footprint of the chip is 2 mm$^2$, including tuning pads. Methods based on a frequency shifting loop [6] or stimulated Brillouin scattering are challenging to integrate because of the required specific loop delay and scattering medium, respectively. A fully integrated solution with a ring resonator has been shown in [29]. However, it had a total footprint of 63 mm$^2$, and the measurement range was limited to 3.5 GHz. FTTM-based photonic systems can provide an integrated solution for multi-tone signal measurement with better accuracy and resolution. Two-step filtering provides best-in-class measurement accuracy of 0.4 MHz but is limited to a measurement range of 10 GHz [4]. Additionally, the measurement requires offline signal processing, leading to a relatively long measurement time and additional system complexity. Compared to all the above-mentioned photonic processing blocks for signal measurement, the proposed PFD-PIC on a silicon-compatible platform is eligible for mass production with a reconfigurable and small form factor, which is suitable for applications with space constraints and could be important for radar warning receivers, secure communication, and wireless networks with 6G and beyond.
5. Conclusion
We have demonstrated a fully reconfigurable compact photonic frequency decoder- PIC (PFD-PIC), which enables real-time spectral and temporal analysis of signals up to the millimeter wave range. Several single-tone frequencies i.e., 10 GHz, 20 GHz, and 26 GHz have been delayed and measured in real-time. Furthermore, a two-tone signal consisting of 20 and 26 GHz was separated in real-time. The proposed PFD-PIC is fully adaptive as the lower or higher frequency components could be delayed and resolved directly in the time domain without requiring any additional signal processing. The presented concept has a very low system complexity, power consumption, and footprint and may therefore be viable for next-generation low-latency, high-speed, wireless communication systems.
Funding
Deutsche Forschungsgemeinschaft (322402243, 403154102, 403579441, 424607946, 424608109, 424608191, 424608271, 454954953, 491066027); Open Access Publication Funds of Technische Universität Braunschweig.
Acknowledgments
The authors would like to thank Janosch Meier, Younus Mandalawi, and Marleen Serak from TU Braunschweig and Arijit Misra from Institute of Integrated Photonics, RWTH Aachen University for the constructive discussions, and Judith Felgner from Physikalisch-Technische Bundesanstalt (PTB) Braunschweig for assistance with the bonding of the chip on PCB.
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. M. B. Weiss, K. Werbach, D. C. Sicker, and C. E. C. Bastidas, “On the application of blockchains to spectrum management,” IEEE Transactions on Cognitive Communications and Networking 5(2), 193–205 (2019). [CrossRef]
2. M. A. Richards, Fundamentals of Radar Signal Processing (McGraw-Hill Education, 2014).
3. J. Capmany and D. Novak, “Microwave photonics combines two worlds,” Nat. Photonics 1(6), 319–330 (2007). [CrossRef]
4. K. Singh, S. Preußler, A. Misra, L. Zhou, and T. Schneider, “Photonic Microwave Frequency Measurement With High Accuracy and Sub-MHz Resolution,” J. Lightwave Technol. 40(9), 2748–2753 (2022). [CrossRef]
5. J. E. Kadum, S. Preußler, R. Das, Y. Mandalawi, and T. Schneider, “Temporal Disentanglement of Wireless Signal Carriers Based on Quasi-Light-Storage,” J. Lightwave Technol. 40(20), 6762–6768 (2022). [CrossRef]
6. T. A. Nguyen, E. H. Chan, and R. A. Minasian, “Photonic multiple frequency measurement using a frequency shifting recirculating delay line structure,” J. Lightwave Technol. 32(20), 3831–3838 (2014). [CrossRef]
7. J. Liu, T. Shi, and Y. Chen, “High-Accuracy Multiple Microwave Frequency Measurement With Two-Step Accuracy Improvement Based on Stimulated Brillouin Scattering and Frequency-to-Time Mapping,” J. Lightwave Technol. 39(7), 2023–2032 (2020). [CrossRef]
8. J. Kadum, R. Das, A. Misra, and T. Schneider, “Brillouin-scattering-induced transparency enabled reconfigurable sensing of RF signals,” Photon. Res. 9(8), 1486–1492 (2021). [CrossRef]
9. S. Preussler and T. Schneider, “Attometer resolution spectral analysis based on polarization pulling assisted Brillouin scattering merged with heterodyne detection,” Opt. Express 23(20), 26879–26887 (2015). [CrossRef]
10. S. R. Konatham, R. Maram, L. R. Cortés, J. H. Chang, L. Rusch, S. LaRochelle, H. G. de Chatellus, and J. Aza na, “Real-time gap-free dynamic waveform spectral analysis with nanosecond resolutions through analog signal processing,” Nat. Commun. 11(1), 3309 (2020). [CrossRef]
11. B. Zhang, X. Wang, and S. Pan, “Photonics-based instantaneous multi-parameter measurement of a linear frequency modulation microwave signal,” J. Lightwave Technol. 36(13), 2589–2596 (2018). [CrossRef]
12. H. Jiang, D. Marpaung, M. Pagani, K. Vu, D. Choi, S. J. Madden, L. Yan, and B. J. Eggleton, “Wide-range, high-precision multiple microwave frequency measurement using a chip-based photonic Brillouin filter,” Optica 3(1), 30–34 (2016). [CrossRef]
13. M. Pagani, B. Morrison, Y. Zhang, A. Casas-Bedoya, T. Aalto, M. Harjanne, M. Kapulainen, B. J. Eggleton, and D. Marpaung, “Low-error and broadband microwave frequency measurement in a silicon chip,” Optica 2(8), 751–756 (2015). [CrossRef]
14. H. Emami, M. Ashourian, and M. Ebnali-Heidari, “Dynamically Reconfigurable All Optical Frequency Measurement System,” J. Lightwave Technol. 32(24), 4796–4802 (2014). [CrossRef]
15. S. Pan and J. Yao, “Photonics-based broadband microwave measurement,” J. Lightwave Technol. 35(16), 3498–3513 (2017). [CrossRef]
16. D. Marpaung, J. Yao, and J. Capmany, “Integrated microwave photonics,” Nat. Photonics 13(2), 80–90 (2019). [CrossRef]
17. R. Das and T. Schneider, “Integrated group delay units for real-time reconfigurable spectrum sensing of mm-wave signals,” Opt. Lett. 45(17), 4778–4781 (2020). [CrossRef]
18. D. E. Spence, P. N. Kean, and W. Sibbett, “60-fsec pulse generation from a self-mode-locked Ti:sapphire laser,” Opt. Lett. 16(1), 42–44 (1991). [CrossRef]
19. J. Li, S. Fu, K. Xu, J. Wu, J. Lin, M. Tang, and P. Shum, “Photonic ultrawideband monocycle pulse generation using a single electro-optic modulator,” Opt. Lett. 33(3), 288–290 (2008). [CrossRef]
20. J. Meier, K. Singh, A. Misra, S. Preußler, J. C. Scheytt, and T. Schneider, “High-Bandwidth Arbitrary Signal Detection Using Low-Speed Electronics,” IEEE Photonics J. 14(2), 1–7 (2022). [CrossRef]
21. A. Rahim, T. Spuesens, R. Baets, and W. Bogaerts, “Open-access silicon photonics: Current status and emerging initiatives,” Proc. IEEE 106(12), 2313–2330 (2018). [CrossRef]
22. S. Liu, J. Feng, Y. Tian, H. Zhao, L. Jin, B. Ouyang, J. Zhu, and J. Guo, “Thermo-optic phase shifters based on silicon-on-insulator platform: state-of-the-art and a review,” Front Optoelectron. 15(1), 9 (2022). [CrossRef]
23. S. De, R. Das, R. K. Varshney, and T. Schneider, “CMOS-Compatible Photonic Phase Shifters With Extremely Low Thermal Crosstalk Performance,” J. Lightwave Technol. 39(7), 2113–2122 (2021). [CrossRef]
24. S. De, R. Das, K. Singh, Y. Mandalawi, T. Kleine-Ostmann, and T. Schneider, “Thermal crosstalk mitigation in a dual-drive Mach-Zehnder modulator,” International Conference on Numerical Simulation of Optoelectronic Devices (2022), pp. 27–28.
25. S. De, R. Das, R. K. Varshney, and T. Schneider, “Design and simulation of thermo-optic phase shifters with low thermal crosstalk for dense photonic integration,” IEEE Access 8, 141632–141640 (2020). [CrossRef]
26. D. J. Blumenthal, R. Heideman, D. Geuzebroek, A. Leinse, and C. Roeloffzen, “Silicon nitride in silicon photonics,” Proc. IEEE 106(12), 2209–2231 (2018). [CrossRef]
27. J. Riemensberger, N. Kuznetsov, J. Liu, J. He, R. N. Wang, and T. J. Kippenberg, “A photonic integrated continuous-travelling-wave parametric amplifier,” Nature 612(7938), 56–61 (2022). [CrossRef]
28. M. Attygalle and D. B. Hunter, “Improved photonic technique for broadband radio-frequency measurement,” IEEE Photonics Technol. Lett. 21(4), 206–208 (2008). [CrossRef]
29. D. Marpaung, “On-chip photonic-assisted instantaneous microwave frequency measurement system,” IEEE Photonics Technol. Lett. 25(9), 837–840 (2013). [CrossRef]
30. Z. Li, B. Yang, H. Chi, X. Zhang, S. Zheng, and X. Jin, “Photonic instantaneous measurement of microwave frequency using fiber Bragg grating,” Opt. Commun. 283(3), 396–399 (2010). [CrossRef]
