Open Access
Add to BibSonomy
Abstract
Recent developments in microwave photonic filters (MPFs) offer superior properties for radio frequency (RF) signal processing, such as large instantaneous bandwidth, high resolution and multifunctional shapes. However, it is quite challenging to realize two or more characteristics simultaneously to meet the diverse needs in complex electromagnetic environment. In this paper, we propose a reconfigurable RF signal spectral processor with both large instantaneous bandwidth and high resolution. In the proposed spectral processor, sufficient taps supplied by an optical frequency comb (OFC) offer a large instantaneous bandwidth to process broadband RF signals. Flexible tap coefficients can be obtained by manipulating an optical spectral shaper (OSS), which provides excellent reconfigurability. This tap-by-tap manipulation is realized with a high resolution of hundreds of megahertz, allowing precise shape configuration of the response. In the experiment, we demonstrate a flat-top response with a wide bandwidth of 7.1 GHz. Reconfigurable features such as tunable bandwidth, adjustable center frequency and diverse shapes are also shown. In particular, the measured frequency resolution of 96.5 MHz demonstrates the ability for precise configuration.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Reconfigurable spectral processors have been the research focus and hotpot in radio frequency (RF) signal processing in recent years. The photon-assisted scheme represented by microwave photonic filters (MPFs) have attracted great attention for excellent performances such as low loss, strong immunity to electromagnetic interference, rapid tunability and reconfigurability [1]. MPFs can be realized in an incoherent scheme with a delay line structure, or in a coherent scheme that converts the optical filter response to the RF response [2]. In an incoherent scheme, multiple taps are provided by a laser array [3,4], a spectral-sliced broadband optical source (BOS) [5–7] or an optical frequency comb (OFC) [8–10]. These taps are modulated with the input RF signal by electro-optical modulators (EOMs) and then propagate through a dispersion element to obtain different time delays. In this way, adjustable time delays and flexible tap coefficients provide the MPF with great reconfigurability. In a coherent scheme, the RF response is mapped directly from the optical filter response. In usual, coherent MPFs can be achieved by removing one of the sidebands of the phase-modulated signal through optical filters formed by stimulated Brillouin scattering (SBS) [11–13], fiber Bragg grating (FBG) [14,15], microring resonators (MRs) [16,17], etc. This technique is known as phase modulation to intensity modulation (PM-IM) conversion. However, these coherent MPFs are usually designed to perform a specific function, such as a single-passband filter or a single-notch filter. Recently, the growing demands in wireless system and satellite communication have motivated researchers to design multifunctional filters [18–20]. Various responses such as multiple passbands, switchable response and other practical shapes have been topics of interest. With a cascaded optical interferometric filter, a multi-passband response was demonstrated in [18]. In [19], multifunctional shapes were achieved in a reconfigurable MPF using a phase-shifted FBG. By implementing a single mode Fabry-Perot laser diode (SMFP-LD) [20] or an intensity-consistent single-stage-adjustable cascaded microring (ICSSA-CM) [21], switchable response MPFs were realized. However, since the filter response was determined by optical filters or internal rings, it is quite difficult to configure the filter response with arbitrary shapes.
Considering that the OFC can provide a large number of taps and flexible tap coefficients, the comb-based scheme becomes a potential solution for reconfigurable MPFs. Generally, flexible tap coefficients can be obtained by shaping the OFC line by line. Based on these adjustable tap coefficients, the comb-based MPF has the ability to configure the filter shape arbitrarily. In [22], the comb-based MPF performed switchable responses with Gaussian shape and Sinc shape. By utilizing dual combs, the response shape of the proposed MPF in [23] was shown with Sinc shape, Gaussian shape and flat-top shape. In [24], benefit from the large taps provided by an integrated Kerr microcomb, the MPF was demonstrated with several reconfigurable shapes as well as a wide bandwidth up to 4.6 GHz. In the above comb-based schemes, the MPF response bandwidth was typically several gigahertz. However, these MPFs were difficult to realize accurate shape configuration in their broadband response due to the poor frequency resolution. In our previous work, a megahertz-resolution programmable microwave spectral shaper was demonstrated with “bandwidth scaling” technique [25]. Nevertheless, limited by insufficient taps and in-band interferences, the MPF did not provide broadband instantaneous bandwidth so far.
In this paper, we present a large-bandwidth RF signal spectral processor based on an OFC. Rely on a narrowband Fabry-Perot (FP) filter, the broadband input RF signal is sliced into numerous narrow slices in optical spectrum. Benefit from the large number of comb lines, plenty of slices can be obtained and reshaped by an optical spectral shaper (OSS), giving the spectral processor arbitrary shape response and reconfigurable phase response. In the experiment, we demonstrate the processor with scalable bandwidth, adjustable center frequency and reconfigurable shapes. Bandwidth expansion is also realized through the image interference suppression technique. Furthermore, the high resolution shown in the spectral processor greatly supports precise manipulation on the reconfigurable responses.
2. Principle
The principle of the reconfigurable comb-based spectral processor is shown in Fig. 1. Specifically, the spectral processor has five sections: OFC generation, RF signal multicasting, optical slicing, tap coefficients weighting, and RF signal recovery.
Fig. 1. The principle of the spectral processor. OFC: optical frequency comb; LO Branch: local oscillator Branch.
As shown in Fig. 1, an OFC is divided into a signal branch and a local oscillator (LO) branch, respectively. In the signal branch, a suppressed carrier single sideband modulation (SC-SSB) modulator multicasts the input RF signal on all comb lines. In Fig. 1(a), a trapezoid representing a spectral replica of the input signal is replicated and then distributed on each comb line. To facilitate the distinction, we group a comb line and the corresponding spectral replica into one channel. After the signal multicasting, the spectral slicing is implemented by an optical filter with periodic narrow passbands. Assume that the free spectral ranges (FSRs) of the comb and the optical filter are ωFSR-comb and ωFSR-FPF, respectively. If ωFSR-comb and ωFSR-FPF differ slightly with a difference of ΔωFSR, the slicing will proceed on spectral replicas channel by channel with a frequency step of ΔωFSR. The principle of the slicing is demonstrated in Fig. 1(b). When the first transmission peak of the optical filter is aligned with the spectral replica in 1-st channel, a spectrum slice will be cut out from the spectral replica. In the next channel, the slicing occurs at a different position of the input signal spectral replica due to the FSR difference. As a result, different input signal spectrum slices are obtained with a frequency step of ΔωFSR in different channels, similar to the Vernier effect.
In the LO branch, the LO comb is weighted line by line to a preset specific profile that can be directly mapped to the RF response profile. As shown in Fig. 1(c) and Fig. 1(d), the LO comb is shaped into a wave profile in weighting section. Subsequently, the weighted LO comb and the spectrum slices in the signal branch are coherent detected in recovery section to reconstruct the input RF signal. As a result, the input signal is reconstructed and shaped into the specific wave shape we programmed.
The spectrum of the generated OFC can be expressed as
Since multicast section is working at SC-SSB mode, the input RF signal s(t) is multicast with a single upper sideband on the comb lines. The optical signal after modulation is expressed as
In slicing section, the optical filter with periodic passbands, such as an FP filter, is usually used to obtain multiple narrowband slices. Assuming that the FSR of the FP filter is ωFSR-FPF, the optical signal after the slicing in the signal branch is given by
In the LO branch, the amplitude of comb is weighted with flexible coefficients and the phase of each comb lines can also be manipulated by the OSS. The expression of the weighted comb is given by
In the recovery section, the input RF signal is recovered and the spectral processor frequency response H(ω) can be given by
From Eq. (6), the frequency response of the spectral processor is the superposition of N frequency-shifted transmission passband Y(ω). And the superposition of all transmission passbands can be independently controlled by optical tap coefficient An. Independent phase control of each passband can also be accomplished by φn. Note that the frequency step of the superposition is ΔωFSR, and when the transmission bandwidth of Y(ω) is similar to the frequency step, each tap can realize spectral control with a bandwidth of ΔωFSR. When ΔωFSR and the transmission bandwidth of Y(ω) are close but not equal, only the flatness of the response passband is slightly affected. In this way, tap-by-tap response configuration with a frequency resolution of ΔωFSR can be realized. Since the tap coefficients can be manipulated one by one in the optical domain, flexible tap coefficients with both amplitude and phase can be applied in H(ω), which brings excellent reconfigurability to the processor. Consequently, in addition to the reconfigurable shape, the response of the spectral processor can be programmed with tunable center frequency and scalable bandwidth. Also, the spectral processor is able to form reconfigurable phase responses through tap-by-tap phase manipulation. Based on this tap-by-tap phase configuration, reconfigurable phase responses such as delay amplification [25] and dispersion amplification [26] have been realized.
In practical scenarios, besides the reconfigurable response, the instantaneous bandwidth expansion of the spectral processor is also important. On the one hand, according to Eq. (6), the sum of N narrowband passbands constitutes an instantaneous bandwidth of N·ΔωFSR. If ΔωFSR is increased for bandwidth expansion, the frequency step will become larger and the response resolution will reduce. Therefore, we choose to increase tap amount N by employing an OFC with sufficient comb lines. Therefore, the bandwidth is expanded while maintaining a high frequency resolution. On the other hand, when the instantaneous bandwidth is close the spacing of the comb lines, the image interference that limits the bandwidth expansion appears. To visualize the impact of the image interference, Fig. 1(e) shows an example in the 2-nd channel. If the spectral slice of 2-nd channel beats with the comb line of 2-nd channel (the orange comb line), a desired target slice can be obtained. But when the slice beats with the comb line of 3-rd channel (the yellow comb line), the generated image interference will overlap with other desired slices in other channel, which compromises the signal integrity. As a consequence, this kind of image interference will hinder the expansion of the instantaneous bandwidth. In addition, components generated between desired slices and the comb lines of higher channels are outside the operating frequency range and do not affect the in-band response of the spectral processor. In other conventional comb-based scheme, the instantaneous bandwidth is often set within ΔωFSR/2 to avoid spectrum aliasing. Nevertheless, the image interferences can be effectively suppressed by our implementation of the interference suppression technique using the balanced Hartley structure [27]. As a result, the instantaneous bandwidth of the spectral processor can be expanded to ΔωFSR.
3. Experiment
The experimental setup is shown in Fig. 2. A continuous-wave (CW) laser is applied to a modulator array consisting of a Mach-Zehnder modulator (MZM, EOspace, AZ-DV5-40-PFV-SFV-LV-VL) and 5 phase modulators (PMs, EOspace, PM-0S5-20-PFA-PFA-UV). Since the modulator array is driven by a sinusoidal signal with a frequency of 10 GHz, the FSR of the generated OFC is 10 GHz. It is worth noting that the cascaded multiple phase modulators greatly increase the number of flat comb lines. Then, the OFC is split into two branches. In the signal branch, the input RF signal modulates the OFC through an in-phase/quadrature (I/Q) modulator (EOspace, IQ-ODVS-35-PFA-PFA-LB) operating at the SC-SSB mode. After the multicasting, the spectral replicas on comb lines are sliced by an FP filter with an FSR of 10.1 GHz and a bandwidth of 94.7 MHz to obtain narrowband slices. In this arrangement, the frequency step determined by ΔωFSR is set to 100 MHz. In the LO branch, an OSS (Finisar, waveshaper 1000s) is used to weight the comb lines with different tap coefficients. Since no channel division device such as the WDM is applied, all channels in each branches undergo unified signal processing. So there are no channel-to-channel alignment issues in the spectral processor. In order to reduce the delay difference between the upper and lower branches, an optical adjustable delay line is implemented in the LO branch. The weighted LO comb is delivered to the coherent detection with the narrowband slices. In the coherent detection, an optical hybrid, an electrical hybrid and two balanced photodetectors (BPDs, Finisar, BPDV2150OR-VF-FA) are implemented to form a Hartley structure for signal recovery and the interference suppression. A vector network analyzer (VNA, Agilent PNA-X N5244A) is implemented to analyze the response of the spectral processor.
Fig. 2. The experimental setup of the spectral processor. CW Laser: continuous-wave laser; PM: phase modulator; MZM: Mach-Zehnder modulator; I/Q Modulator: in-phase/quadrature modulator; FPF: Fabry-Perot filter; OSS: optical spectral shaper; BPD: balanced photodetector; VNA: vector network analyzer.
In the proposed spectral processor, the performance of the comb plays a significant role in expanding the instantaneous bandwidth. Since the instantaneous bandwidth is determined by N·ΔωFSR, a large tap number N is required to achieve a wide instantaneous bandwidth. As shown in the inset of Fig. 3(a), flat comb lines can be obtained by the cascaded modulators method. As shown in Fig. 3, the comb can provide 75 continuous comb lines within the power fluctuation range of 3 dB. In this way, a flat-top response with a large bandwidth of 7.1 GHz covering 1.2 GHz to 8.3 GHz is realized, as shown in Fig. 3(b). It can be seen from the broadband flat-top response that the in-band fluctuation is less than 5 dB and the out-of-band rejection ratio exceeds 30 dB.
Fig. 3. (a) Flat comb with a large number of taps. (b) The large-bandwidth response of the spectral processor.
Although theoretically a large bandwidth can be realized with a large number of taps, the presence of in-band interferences limits the instantaneous bandwidth expansion. In the SC-SSB modulation implemented by the I/Q modulator, the residual unwanted sideband (-1st sideband) and carrier will be converted into in-band interferences such as image interferences in the coherent detection, affecting the output of the desired signal. Although a SC-SSB modulation sideband rejection ratio exceeding 32 dB can be achieved in the experiment by using a well-performing electrical hybrid and power-matched RF cables, the image interference is still a crucial part of the in-band interferences. Note that the FSR of the comb is 10 GHz, the image interferences would be non-negligible when the instantaneous bandwidth exceeds 5 GHz. In the experiment, we simulate a broadband signal with a swept frequency signal from a microwave source and observe the image interference with a spectrum analyzer. Given that the sweep range is from 2 GHz to 4 GHz, the interferences are distributed between 6 GHz and 8 GHz. It can be seen from Fig. 4 that the image suppression ratio is 25.2 dB, indicating that the influence of image interferences is effectively eliminated. Therefore, the response is still reliable when the instantaneous bandwidth exceeds half of the FSR. For a higher image interference suppression ratio, an I/Q modulator with a better extinction ratio can be applied. Besides, it is also important to realize the matching of all branches in the SC-SSB modulation and the balanced Hartley structure.
To realize arbitrary configuration of the spectral response, the tap coefficients are programmed independently by the OSS. By controlling the comb line number N, bandwidth scaling responses can be observed. The FSR difference ΔωFSR is fixed at 100 MHz and we program An to enable a different number N. As Fig. 5(a) depicts, flat-top responses with band-widths of 2.1 GHz, 4.1 GHz, and 6.1 GHz are realized in the experiment when N is set to 21, 41 and 61. In addition to the scalable bandwidth, the center frequency of the spectral processor response is also tunable. In the proposed processor, the optical spectral profile of the comb can be directly mapped to the RF response. Consequently, enabling the comb lines at different positions in the optical spectrum results in a center frequency shift in RF response. In this case, we demonstrate the tunable center frequency performance with a 2 GHz bandwidth response. As shown in Fig. 5(b), flat-top responses centered at 3 GHz, 5 GHz and 7 GHz are demonstrated with orange, purple, green solid lines, respectively. It is clear from the Fig. 5(b) that the center frequency of the response can be flexibly tuned by controlling the optical tap coefficients.
With flexible tap coefficients, a series of arbitrary shape responses can be configured. We preset the comb with a notch profile and show the programmed comb with attenuated comb lines in Fig. 6(a). The corresponding notch shape response (denoted as the purple solid line in Fig. 6(a)) agrees well with the simulated response (denoted as the gray dashed line in Fig. 6(a)). In this way, an arbitrary waveform response can be achieved by shaping the comb with the corresponding profile. By changing the comb profile from a notch shape to a Gaussian profile, a Gaussian profile comb and corresponding Gaussian shape RF response are shown in Fig. 6(b). Considering the requirements of practical scenarios, a dual-passband response and a single-passband response are shown in Fig. 6(c) and Fig. 6(d), respectively. It is worth mentioning that since shape change only requires OSS configuration, shape response switching is quite stable compared to the thermal tuning of microring resonators. In addition to the arbitrary shape responses, the phase response of the spectral processor is also reconfigurable. By applying phase control tap-by-tap, a first-order linear phase response, a second-order nonlinear phase response and a third-order nonlinear phase response are demonstrated by the simulation in Fig. 7. As a result, the arbitrary shape response and reconfigurable phase response contribute to the excellent reconfigurability of spectral processor.
Fig. 6. The reconfigurable shape of the spectral processor. (a) The notch profile comb and corresponding notch shape response. (b) The Gaussian profile comb and corresponding Gaussian shape response. (c) The dual-passband profile comb and corresponding dual-passband shape response. (d) The single-passband profile comb and corresponding single-passband shape response.
Fig. 7. Reconfigurable phase response results. (a) A first-order linear phase response and its amplitude response. (b) A second-order nonlinear phase response and its amplitude response. (c) A three-order nonlinear phase response and its amplitude response.
In order to realize arbitrary shape configuration, it is necessary to achieve not only flexible tap coefficients, but also precise control between taps. Our comb-based spectral processor is fundamentally based on the concept of tap-by-tap manipulating, so the bandwidth that can be achieved with a single tap is the frequency resolution of the response configuration. Typically, tap coefficients are controlled in optical domain by optical filters with tens of gigahertz resolution. However, relying on the Vernier effect we implemented in proposed processor, the frequency resolution of the RF response can reach hundreds of megahertz. To investigate the frequency resolution of the proposed spectral processor, we explore the 3-dB bandwidth that can be achieved by a single tap. The response of the single-tap passband is shown in the inset figure of Fig. 8, and the response after zooming in is shown in the large one. From Fig. 8, the frequency resolution of a single tap can reach 96.5 MHz. Benefit from the high resolution of the processor, the response can be precisely controlled tap-by-tap with a 96.5 MHz resolution, which enables truly arbitrary reconfigurability of the spectral processor. Table 1 compares the performance between the proposed spectral processor and previously reported spectral processors. As can be seen, although some works can achieve high-resolution [11–18], the response shape is limited by the optical filtering devices. In actual scenarios, it is difficult for multifunctional spectral processors [19–21] to be arbitrarily configured to adapt to changes in requirements. In the exploration of achieving reconfigurable responses [22–25], this work realized the precise high-resolution manipulation in a broadband response. This is, the proposed spectral processor achieves high resolution, reconfigurability and large bandwidth at the same time, which is of great significance in the RF signal spectral processing.
Table 1. Performance comparison between the proposed spectral processor and previously reported spectral processors
4. Conclusion
In this paper, we demonstrated a reconfigurable spectral processor based on an OFC. With sufficient taps provided by the OFC and the interference suppression technique, a large instantaneous bandwidth can be realized without the restriction of the image interferences. Using tap-by-tap manipulating, arbitrary shape, tunable center frequency and adjustable bandwidth responses are available to realize. Moreover, the conversion of the comb profile to the spectral processor response enables high-resolution shape configuration. In the experiment, we demonstrate flat-top responses with bandwidths of 2 GHz to 7.1 GHz. Multifunctional and arbitrarily shaped responses are shown, such as single-passband, dual-passband, notch filter and Gaussian profile responses, which demonstrates the excellent reconfigurability of the spectral processor. The reconfigurable phase control is shown with programmable phase responses. Furthermore, we investigate the precise shape configuration of the spectral processor and measure a high frequency resolution of 96.5 MHz.
Funding
National Key Research and Development Program of China (2021YFB2800802, 2021YFB2800805); National Natural Science Foundation of China (62071055, 62135014).
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. J. Capmany, B. Ortega, and D. Pastor, “A tutorial on microwave photonic filters,” J. Lightwave Technol. 24(1), 201–229 (2006). [CrossRef]
2. J. Yao, “Photonics to the rescue: a fresh look at microwave photonic filters,” IEEE Micro. Mag. 16(8), 46–60 (2015). [CrossRef]
3. V. Polo, B. Vidal, J. L. Corral, and J. Marti, “Novel tunable photonic microwave filter based on laser arrays and N/spl times/N AWG-based delay lines,” IEEE Photonics Technol. Lett. 15(4), 584–586 (2003). [CrossRef]
4. A. Loayssa, J. Capmany, M. Sagues, and J. Mora, “Demonstration of incoherent microwave photonic filters with all-optical complex coefficients,” IEEE Photonics Technol. Lett. 18(16), 1744–1746 (2006). [CrossRef]
5. J. Mora, B. Ortega, A. Diez, J. L. Cruz, M. V. Andres, J. Capmany, and D. Pastor, “Photonic microwave tunable single-bandpass filter based on a Mach-Zehnder interferometer,” J. Lightwave Technol. 24(7), 2500–2509 (2006). [CrossRef]
6. D. Liang, Q. Tan, W. Jiang, W. Zhang, D. Wang, J. Gong, X. Li, L. Li, Z. Zhang, and Y. Liu, “Spectrum-continuous narrowband filter with an enhanced flat-top passband and sharp roll-off performance,” Opt. Lett. 46(21), 5397–5400 (2021). [CrossRef]
7. H. E. Kotb, Y. M. Sabry, M. S. Abdallah, and H. Omran, “MEMS-SOA spectrum-sliced auto-equalized source enabling uniformly tunable microwave photonic filter,” IEEE Photonics Technol. Lett. 33(1), 15–18 (2021). [CrossRef]
8. M. Song, C. M. Long, R. Wu, D. Seo, D. E. Leaird, and A. M. Weiner, “Reconfigurable and tunable flat-top microwave photonic filters utilizing optical frequency combs,” IEEE Photonics Technol. Lett. 23(21), 1618–1620 (2011). [CrossRef]
9. W. Dai, W. Rao, H. Wang, and H. Fu, “Microwave photonic filter by using recirculating frequency shifter to generate optical frequency comb,” IEEE Photonics J. 13(2), 1–8 (2021). [CrossRef]
10. H. Shu, L. Chang, Y. Tao, B. Shen, W. Xie, M. Jin, A. Netherton, Z. Tao, X. Zhang, R. Chen, B. Bai, J. Qin, S. Yu, X. Wang, and J. E. Bowers, “Microcomb-driven silicon photonic systems,” Nature 605(7910), 457–463 (2022). [CrossRef]
11. W. Zhang and R. A. Minasian, “Widely tunable single-passband microwave photonic filter based on stimulated Brillouin scattering,” IEEE Photonics Technol. Lett. 23(23), 1775–1777 (2011). [CrossRef]
12. A. Mahendra, Y. Liu, E. Magi, A. Choudhary, D. Marpaung, and B. J. Eggleton, “High link performance of Brillouin-l based microwave bandpass photonic filters,” OSA Continuum 1(4), 1287–1297 (2018). [CrossRef]
13. M. K. Varun and Ravi Pant, “Efficient microwave photonic bandpass filter with large out-of-band rejection, high-resolution and low loss up to 40 GHz,” J. Lightwave Technol. 39(21), 6724–6732 (2021). [CrossRef]
14. W. Li and J. Yao, “A wideband frequency tunable optoelectronic oscillator incorporating a tunable microwave photonic filter based on phase-modulation to intensity-modulation conversion using a phase-shifted fiber Bragg grating,” IEEE Trans. Microwave Theory Techn. 60(6), 1735–1742 (2012). [CrossRef]
15. E. Xu and J. Yao, “Frequency-and notch-depth-tunable single-notch microwave photonic filter,” IEEE Photonics Technol. Lett. 27(19), 2063–2066 (2015). [CrossRef]
16. H. Qiu, X. Zhang, F. Zhou, J. Qie, Y. Yao, X. Hu, Y. Zhang, X. Xiao, Y. Yu, and J. Dong, “A continuously tunable sub-gigahertz microwave photonic bandpass filter based on an ultra-high-Q silicon microring resonator,” J. Lightwave Technol. 36(19), 4312–4318 (2018). [CrossRef]
17. Y. Liu, J. Hotten, A. Choudhary, B. J. Eggleton, and D. Marpaung, “All-optimized integrated RF photonic notch filter,” Opt. Lett. 42(22), 4631–4634 (2017). [CrossRef]
18. J. Ge and M. P. Fok, “Reconfigurable RF multiband filter with widely tunable passbands based on cascaded optical interferometric filters,” J. Lightwave Technol. 36(14), 2933–2940 (2018). [CrossRef]
19. N. Shi, T. Hao, W. Li, N. Zhu, and M. Li, “A reconfigurable microwave photonic filter with flexible tunability using a multi-wavelength laser and a multi-channel phase-shifted fiber Bragg grating,” Opt. Commun. 407, 27–32 (2018). [CrossRef]
20. H. Chen, M. Lee, Y. H. Won, B. Nakarmi, and S. Pan, “High-speed switchable dual-passband microwave photonic filter with dual-beam injection in an SMFP-LD,” J. Lightwave Technol. 39(24), 7966–7972 (2021). [CrossRef]
21. Z. Tao, Y. Tao, M. Jin, J. Qin, R. Chen, B. Shen, Y. Wu, H. Shu, S. Yu, and X. Wang, “Highly reconfigurable silicon integrated microwave photonic filter towards next-generation wireless communication,” Photon. Res. 11(5), 682–694 (2023). [CrossRef]
22. E. Hamidi, D. E. Leaird, and A. M. Weiner, “Tunable programmable microwave photonic filters based on an optical frequency comb,” IEEE Trans. Microwave Theory Techn. 58(11), 3269–3278 (2010). [CrossRef]
23. H. J. Kim, D. E. Leaird, and A. M. Weiner, “Rapidly tunable dual-comb RF photonic filter for ultrabroadband RF spread spectrum applications,” IEEE Trans. Microwave Theory Techn. 64(10), 3351–3362 (2016). [CrossRef]
24. M. Tan, X. Xu, J. Wu, R. Morandotti, A. Mitchell, and D. J. Moss, “Photonic RF and microwave filters based on 49 GHz and 200 GHz Kerr microcombs,” Opt. Commun. 465, 125563 (2020). [CrossRef]
25. J. Li, Y. Dai, F. Yin, W. Li, M. Li, H. Chen, and K. Xu, “Megahertz-resolution programmable microwave shaper,” Opt. Lett. 43(8), 1878–1881 (2018). [CrossRef]
26. J. Li, S. Fu, X. Xie, M. Xiang, F. Yin, Y. Dai, and Y. Qin, “Optically magnified dispersion of microwave signals with a wide and flexible tunable range,” Opt. Lett. 47(5), 1057–1060 (2022). [CrossRef]
27. F. Yin, Z. Yin, X. Xie, Y. Dai, and K. Xu, “Broadband radio-frequency signal synthesis by photonic-assisted channelization,” Opt. Express 29(12), 17839–17848 (2021). [CrossRef]
