In this paper, a photonics-based dual-band linear frequency-modulated continuous wave (LFMCW) radar receiver is proposed. The system core is a microwave photonic in-phase and quadrature (I/Q) mixer, whose inherent large bandwidth, high I/Q balance and favorable uniformity enable the receiver to operate over an extremely wide frequency range. An integrated dual-band waveform offers the possibility of independent detection, allowing the sharing of hardware resources and joint dechirp processing of dual bands. In the proof-of-concept experiment, the distance measurements of S- and C-bands are implemented, with a high and uniform image rejection exceeding 28 and 30 dB, respectively. The image rejections of the two bands can be further improved to 43 and 41 dB at least by digital signal processing (DSP). The proposed photonic-assisted receiver is thus able to simplify the architecture and improve performance for the multispectral sensing application.
© 2017 Optical Society of America
Nowadays, an increasing attention has been paid to multispectral sensing and radar systems. The capability to operate at more than one frequency can increase the likelihood of capture and anti-jamming effectiveness . Different characteristics of the scene can be extracted from multiband sensing data, helping to enhance classification and environmental surveillance [2,3]. In such applications as the dual-frequency precipitation radar , separation of sea ice types at L- and C-band  and dual-frequency altimeter , multispectral sensing is necessary to offer the integrated information. Moreover, the multiband radar system is also able to enhance the resolution and target recognition by using data fusion and classification algorithms [7–9].
Linear frequency-modulated continuous wave (LFMCW) radar system with the advantages of the simple structure and low cost is widely applied to high range resolution radar applications such as synthetic aperture radar (SAR)  and indoor precise positioning . Dechirp processing is used to process wide bandwidth linear frequency-modulated (LFM) waveforms in the way of homodyne detection. Compared with the conventional matched filter, dechirp processing substantially reduces the effective intermediate frequency (IF) bandwidth and allows digitization using low-speed analog-to-digital converters (ADCs) . In addition, the homodyne downconversion and a reduced processing bandwidth are beneficial for the multiband radar to simplify complexity of systems. Several schemes of multiband LFMCW radar has been proposed [12,13]. However, their main drawbacks are disability of simultaneous operation at different bands and the limitation of functionalities.
The multiband radar system requires minimizing the implementation complexity of the radio frequency (RF) receiver, entailing the necessity of sharing hardware resources. However, the conventional electronic receiver can only operate on a fixed single band, and hardly expand further due to the inherent bandwidth limitation. Then, the leakage problem associated with the homodyne detection could saturate the receiver first-stage low-noise amplifier (LNA) and degrade the performance of receiver . Low local oscillator to radio frequency (LO-RF) isolation and the substrate coupling are the main sources of leakage for the conventional receiver mixers. The leakage becomes more severe for multiband receivers. Moreover, the non-uniform amplitude and phase response as a function of frequency would cause joint processing to fail. A dechirping-based multiband radar concept using a single mixer and ADC was addressed . However, these disadvantages of electronic receiver mentioned above would severely restrain its implementation.
Microwave photonic techniques with several inherent benefits, such as large instantaneous bandwidth, favorable uniformity over entire RF bands, high RF isolation, and immunity to electromagnetic interferences, have been widely studied for microwave measurements and radar systems [16,17]. All these features indicate microwave photonics to be a promising solution to the multiband radar systems. A photonics-based transceiver of multiband coherent radar was achieved by a single mode-locked laser . The system operating in the S- and X-bands completed aerial and naval target detection by field trials. Based on a reconfigurable Hilbert transformer, a microwave photonic in-phase and quadrature (I/Q) detector for frequency agile radar was designed . The system was demonstrated over a frequency range of 3.5-35 GHz with a maximum phase imbalance of 7.1°. Based on an ultra-high-Q band-pass filter, a full-band RF photonics receiver with the SFDR larger than 111.6 dB·Hz2/3 was obtained .
Recently, we proposed and experimentally investigated a direct-conversion receiver with high port isolation and I/Q phase balance using microwave photonic I/Q mixer . This scheme architecturally employs optical I/Q modulators to achieve single-sideband carrier-suppressed (SSB-CS) modulation, instead of the conventional Mach–Zehnder modulators (MZMs) and tunable optical filters. In this paper, we further propose and demonstrate a dual-band LFMCW radar receiver based on the microwave photonic I/Q mixer in . An up-chirp and a down-chirp LFM waveforms falling on two different bands are used as the transmitted signals. The sensing information from different bands can be obtained independently by using the I/Q mixer. This waveform design can avoid the appearance of false targets in the concept of . Furthermore, we experimentally demonstrate the distance measurements of S- and C-bands, in which the delay of echoes is emulated by optical links because the stringent phase and noise requirements can be met in fiber-optic links . The proposed photonic receiver allows a single hardware to support different sensing signals. The results show a high and uniform image rejection of IF signal, that is crucial to the independent and unambiguous detection of the dual-band receiver. In addition, image component can be further suppressed using subsequent digital I/Q imbalance compensation due to the favorable uniformity over entire RF bands of the microwave photonic I/Q mixer.
2. Dual-band waveform for LFMCW radar
LFM waveform is chosen for continuous wave radar system because of its simplicity and the possibility of using dechirp processing to simplify range compression. The important feature of an LFM waveform is that the signal’s instantaneous frequency changes with time linearly. In the traditional single-band LFMCW receiver, as shown in Fig. 1(a), the received echo waveform with a delay is mixed with a replica of the transmitted waveform using a typical continuous sawtooth modulation. This operation produces the IF signal, usually referred as the “beat signal,” which is subsequently low-pass filtered and acquired. This approach effectively converts time delay into the beat frequency containing the sensing information. The bandwidth of the beat signal is usually much lower than the transmitted bandwidth. Hence, this dechirp processing can deeply reduce the ADC and processing requirements.
In order to achieve dual-band operation of the receiver and minimize the hardware, an integrated sensing waveform is necessary for satisfying the requirements of functionality and performance. Functionally, the dual-band receiver should possess the capability of receiving simultaneously and detecting independently. This means the receiver can operate different bands simultaneously while the sensing information from different bands should not interfere each other. In , a dechirping-based solution for a multiband LFMCW radar has been proved through simulation. However, in this scheme, the beat frequencies of different bands may overlap at the same range window, and this could lead to the appearance of false targets. The imaging of large field and detection of multiple targets cannot be achieved. We propose a radar waveform for the dual-band LFMCW radar system, enabling each band to complete detection simultaneously and independently, as shown in Fig. 1(b). An up-chirp and a down-chirp LFM waveforms whose chirp rates have opposite signs are adopted for two bands respectively. It should be noted that the quadrature demodulation is needed to provide complete information of echo signal. Through an I/Q mixer and low-pass filters (LPFs), the beat signals of I and Q channel are combined into a single complex signal I + jQ. After applying a fast Fourier transform (FFT) over the complex signal, the beat frequencies generated from the up-chirp and down-chirp bands will distribute in the positive and negative spectrum respectively. The sensing information from different bands can be obtained independently through this approach.
In fact, the operation of dechirp processing is equivalent to the homodyne receiving. The mixing of the received signal with a replica of the transmitted occurs at the RF level. Considering the performance of the receiver, an ultra-wideband and low leakage I/Q mixer is a crucial component. In addition, the image component of beat frequency resulting from the amplitude and phase imbalance of I and Q channels is a serious problem, especially for the dual-band waveform proposed. The image component of beat frequency from the up-chirp band could overlap with the beat frequency from the down-chirp band and vice versa. The false target will appear if the image component cannot be rejected sufficiently. The impact of image component on threshold detection is shown in Fig. 2, where a single target is detected by the dual-band receiver we proposed. A 2-MHz and a −4-MHz beat frequencies are detected by the up-chirp and down-chirp bands respectively. The −2-MHz image component crossing the threshold becomes a false target. Hence, the image rejection is an important factor for the dual-band receiver we proposed.
Furthermore, for the conventional electronic I/Q mixer, usually a 90° phase shift is realized by a microstrip line. However, this method cannot provide the desired and uniform phase shift and the precise amplitude matching of I and Q channels over a very broad frequency range due to the frequency-dependent phase deviation and loss of RF phase shifters. The dual-band receiver may operate on two frequency bands with a large frequency difference (such as X- and Ka-band). Therefore, a high and uniform image rejection are hard to be guaranteed by conventional electronic I/Q mixer.
3. Photonics-based I/Q receiver
Figure 3 shows the schematic diagram of the proposed photonics-based I/Q receiver for the dual-band LFMCW radar. The system consists of a continuous wave (CW) laser, two I/Q modulators, a 90° optical hybrid, two balanced photodetectors (BPDs) and a pair of ADCs. The CW laser is first generated and then split into two paths. The replica of the transmitted LFM waveforms and the echo signal regarded as the LO and RF signals, respectively, are applied to the optical carriers on the two paths by two optical LiNbO3 I/Q modulators. Two RF quadrature hybrids and automatic bias controllers (ABCs) are employed to generate the SSB-CS mode of two optical I/Q modulators. Thereafter, the optical signals from the two paths are injected into the 90° optical hybrid together and then detected by two BPDs. Then the I and Q beat signals are acquired by a pair of ADCs for subsequent digital signal processing.
Mathematically, the transmitted ith band LFM waveform in one period can be expressed as
The optical carriers on the LO and RF paths are modulated by a replica of the transmitted signal and the received signal, then the optical outputs of the SSB-CS mode can be expressed respectively as
Then the optical signals on the LO and RF paths are mixed through a 90° optical hybrid which provides accurate phase shifts of 0°,180°, 90°, −90° for optical signals. The four outputs of the optical hybrid can be written as
As previously mentioned, a high image rejection of the beat signal is the key to the proposed receiver. For the microwave photonic I/Q mixer proposed, there are two main sources of image component. One is the I/Q imbalance brought by the imperfections of the part electro-optic modulation. It includes the phase deviation of RF hybrid and the phase and amplitude mismatch among the two branches of I/Q modulator. These will cause the SSB-CS modulation with the insufficient suppression of the useless sideband. Another is the phase and amplitude imbalance of two BPDs outputs, originating from the imprecise phase shift of the optical hybrid, the inconsistent responses of BPDs, and the power mismatch among the paths after the optical hybrid.
To analyze the image rejection of the microwave photonic I/Q mixer, numerical simulations were carried out based on the setup shown in Fig. 3. The transmitted signal is a single-band LFM waveform with 100-MHz bandwidth at a 2-GHz carrier frequency. Note that we only need to consider the relative signal power in the simulations.
Figure 4 shows the image rejection ratio (IRR) of beat signal as a function of sideband suppression ratio (SSR) of two I/Q modulators. Here, we assume that output of the I/Q modulator on each path has the same SSR. It can be seen that obtaining a 40-dB IRR needs a 20-dB SSR of I/Q modulators, which is caused by an 11° phase deviation of RF hybrids. And the image rejection will decrease only when SSB-CS modulations of two paths are both faulty. This indicate that some imperfection of RF hybrids is tolerable. Therefore, high image rejection can be easily satisfied by using broadband RF hybrids with a general performance.
Next, we investigate the more immediate causes of image generation related to the phase and power imbalance of two BPDs outputs. Figure 5 illustrates the IRR as a function of the phase and power imbalance. As shown in Fig. 5(b), a phase mismatch of 5° gives an IRR of 27dB. In the proposed I/Q mixer, the phase imbalance of beat signal is determined to the phase-shift imprecision of optical hybrid which can be guaranteed to be within ± 5° for some commercial products. In practice, the slight I/Q imbalance is tolerable because it can be compensated for by subsequent digital signal processing. However, for the multiband receiver, we concern not only the magnitude of I/Q imbalance, but also the uniformity of I/Q imbalance between different bands. The inconsistent imbalance cannot be compensated.
4. Experiment and results
The experimental setup and the corresponding optical and electrical spectra in several monitor points are shown in Fig. 6. An arbitrary waveform generator (Tektronix AWG70001A) is used to generate dual-band LFM waveforms containing an S-band up-chirp signal and a C-band down-chirp signal. They are generated simultaneously at 3.3 GHz and 5.3 GHz carriers with 100-MHz and 200-MHz bandwidths respectively and have the same period of 200 μs. Then the signal generated is divided into two branches. In the upper branch, an electrical amplifier (EA) with a gain of 26 dB boosts the signal up to a total power of 14.9 dBm (11.6 dBm for S-band and 12.2 dBm for C-band). Note that the EA only has a 3 dB bandwidth of 2-6 GHz, mainly limiting the operating frequency range. The amplified signal is regarded as the replica of transmitted signal (i.e. LO signal) whose power remains unchanged. In the lower branch, a delay module using external modulation optical links, as shown on the right of Fig. 6, provides RF delays for the echo signals (i.e. RF signal) of point targets. The power of RF signal can be tuned by the optical attenuator. The optical carrier at 1551.3 nm with an optical power of 18.7 dBm is generated by a 1MHz-linewidth CW laser. After an optical coupler, the optical carriers are injected to two I/Q modulators driven by the LO and RF signals. Then the optical signals on two paths are mixed by a 90°optical hybrid. The two low-speed BPDs both embedded with around 20 dB LNAs are used to detect the optical signals. The input optical power to the BPDs is around −9.5 dBm. As ADCs, a two-channel digitizer (ADLINK PXIe-9852) of 14-bit and 200 MS/s has been employed. Digital signal processing (DSP) is carried out by Matlab. After a digital I/Q imbalance compensation based on the Gram–Schmidt algorithm, the extracted I/Q beat signals are combined together to be a complex signal. The beat frequencies containing the distance information can be obtained by FFT to the complex signal padded with zeros.
Firstly, the delay module uses a 1 km single-mode fiber (SMF) to generate the echo signal of a single point target. When the echo signal power of the S- and C-bands are −23.1 and –20.7 dBm, respectively, the sampled I/Q beat signals in time and in the frequency domain are shown in Fig. 7. In Figs. 7(a) and (b), the temporal I/Q outputs in four periods and the part of temporal waveforms is shown. The power spectrum of beat signals is obtained by FFT to the I/Q complex signal, as shown in Fig. 7(c). The beat frequencies of S- and C-bands are located in the negative and positive spectrum with the frequency of around −2.5 MHz and 5 MHz. They have the different frequencies because of the different chirp rates of two bands. High image rejection of 28.4 dB for S-band and 30.9 dB for C-band is obtained. The corresponding I/Q amplitude imbalance of the down-converted signals is within 0.4 dB, and the phase imbalance is less than 2.8°. In fact, the phase imbalance is superior to that of commercial broadband microwave I/Q mixers, which is typically ± 5°. After the I/Q imbalance compensation, the I/Q phase imbalance is measured to be around −0.6° and 0.6° for S- and C-bands. The image rejection can be further increased to 44.3 dB and 43.9 dB for S- and C-bands, as shown in Fig. 7(d). In addition, the residual DC offset from device’s imperfection is also compensated for in the digital domain.
Figure 8 shows the power of I/Q beat signal of each band as a function of RF power. It can be observed that the frequency convert has a good linearity in the measured power range. The inconsistent conversion gain of two bands is caused by the nuanced LO power and frequency-dependent loss of electronic devices. In addition, the conversion loss can be further reduced by optimizing the LO power and inserting optical amplifiers to compensate for the optical loss.
Subsequently, the IRR of the beat signal is measured when inputting different RF power, and shown in Fig. 9. The IRR higher than 28 dB and 30 dB for S- and C-bands is obtained without I/Q imbalance compensation. This is better than the conventional electronic I/Q receiver in the broadband operation. Furthermore, I/Q imbalance of the beat signal of two bands can be compensated for simultaneously, so that the IRR is increased by around 15 and 10 dB, raising to 43 and 41 dB at least for S- and C-bands. Due to the favorable uniformity of optical processing, the proposed I/Q receiver has a uniform I/Q imbalance of each band. This means that the compensation operating on the dual bands can be realized effectively. It should be noted that the compensation algorithm is more inclined to compensate I/Q imbalance at the band with a lower image rejection. The characteristics of both the receiver and the compensation algorithm ensure a uniform image rejection of different bands.
In order to demonstrate the performance of multi-target detection, two optical links of about 1 km and 2 km are used to simulate two point targets (target A and target B), respectively. The RF signal power of target A is fixed at −21.7 dBm and −20 dBm for the S- and C-bands, while the RF power from target B is regulated by an optical attenuator. Note that the bandwidth of C-band transmitted signal is changed into 150 MHz from 200 MHz in order to discriminate image component clearly. A special situation in which the two targets have a power difference of almost 30 dB is demonstrated in the experiment. In Fig. 10(a), the image of the strong target (target A) will interfere the beat frequency of the weak target (target B) before I/Q imbalance compensation. After the compensation, the weak target can be distinguished due to the increased IRR as shown in Fig. 10(b).
Figure 11 shows the IRR of target A and target B as a function of the RF power of target B. Note that the IRR cannot be measured when the image power is lower than the noise in Fig. 11(b). It can be observed that the two targets have a uniform image rejection (higher than 28 dB without compensation and 40dB with compensation) as the change of received power. The high image rejection and good compensation effect are maintained for multi-target detection.
5. Discussion and conclusion
It should be noted that the high microwave performances of the microwave photonic I/Q mixer have been confirmed experimentally in . The I/Q phase imbalance is below 3° over the entire RF bands below 20 GHz. An extremely high LO-RF isolation can mitigate the leakage problem of LFMCW radar receiver. And a high spurious-free dynamic range (SFDR) can increase the linearity range of sensitivity for the multiband operation. The performance can be further improved by reducing the optical loss, using pre-LNA and optical modulators with a larger bandwidth. Moreover, the introduction of balanced detection in the optical domain can suppress the common-mode noises effectively. These fundamental features facilitate the implementation of the dual-band LFMCW radar receiver we proposed.
A main constraint of the proposed receiver is the requirement of broadband RF hybrids, nevertheless the perfectness of their phase shift is undemanding. The dual-band LNA and antenna are required for a complete radar receiver system. Compared with the electric multiband radar using separated receivers for different bands, the photonics-based dual-band radar receiver offers the clear advantages of reducing the hardware usage and avoiding a doubling of the complexity and footprint. Photonic integrated techniques are expected to reduce the size and power consumption of the proposed architecture. Moreover, the dual bands can be selected flexibly due to the large bandwidth of the optical processing, while the operation frequency of conventional receivers is fixed. Therefore, the reconfigurability is a main and undisputed advantage of the proposed photonic receiver.
Functionally, the simultaneous operation on two bands is based on the integrated LFM waveform of the up and down chirp. The range windows of different bands are distinguished by the multiplexing of positive and negative frequency. Compared with the approach in , which distinguishes beat frequencies by a suitable selection of the chirp rates, the proposed concept has an essential difference. The overlapping of range windows between different bands can be avoided, therefore it is possessed of stronger generality for more scenarios such as imaging and detection of large field and multiple targets. In addition, the approach is compatible with subsequent signal processing such as target Doppler analysis and imaging because the beat signals contain the complete information of dual-band echoes.
In conclusion, we have proposed a dual-band LFMCW radar receiver based on microwave photonic I/Q mixer. The sharing of hardware resources and joint dechirp processing of dual bands are implemented by applying an integrated dual-band LFM waveform. Experimentally, we demonstrate the independent and unambiguous distance measurement of S- and C-bands for multi-target detection. The image rejections are estimated to be more than 28 dB and 30 dB for the S- and C-bands, which can be further improved to 43 and 41 dB by digital I/Q imbalance compensation. These benefit from the high I/Q balance and excellent frequency uniformity of the receiver proposed. It is hard to be achieved by the conventional electronic homodyne receivers. Therefore, the proposed photonics-based receiver is a promising solution with compactness and high performance for the multispectral sensing and radar systems.
National Natural Science Foundation of China (NSFC) (61431003, 61601049, 61625104); National 863 Program of China (2015AA016903); State Key Laboratory of Advanced Optical Communication Systems and Networks, Shanghai Jiao Tong University, China.
References and links
1. M. I. Skolnik, Radar Handbook, 3rd Ed (McGraw-Hill,2008).
2. Z. Zhou, P. Caccetta, and N. C. Sims, “Multiband SAR data for rangeland pasture monitoring,” in Geoscience and Remote Sensing Symposium (IEEE, 2016), pp. 160–173. [CrossRef]
3. D. B. Trizna, C. Bachmann, M. Sletten, N. Allan, J. Toporkov, and R. Harris, “Projection pursuit classification of multiband polarimetric SAR land images,” IEEE Trans. Geosci. Remote Sens. 39(11), 2380–2386 (2001). [CrossRef]
4. S. Seto, T. Iguchi, and T. Oki, “The basic performance of a precipitation retrieval algorithm for the global precipitation measurement mission’s single/dual-frequency radar measurements,” IEEE Trans. Geosci. Remote Sens. 51(12), 5239–5251 (2013). [CrossRef]
5. J. A. Casey, S. E. L. Howell, A. Tivy, and C. Haas, “Separability of sea ice types from wide swath C-and L-band synthetic aperture radar imagery acquired during the melt season,” Remote Sens. Environ. 174, 321–328 (2016). [CrossRef]
6. N. M. Frew, D. M. Glover, E. J. Bock, and S. J. McCue, “A new approach to estimation of global air-sea gas transfer velocity fields using dual-frequency altimeter backscatter,” J. Geophys. Res. 112(C11), C11003 (2007). [CrossRef]
7. P. Van Dorp, R. Ebeling, and A. G. Huizing, “High resolution radar imaging using coherent multiband processing techniques,” in IEEE Radar Conf. (IEEE, 2010), pp. 981–986.
8. M. Vespe, C. J. Baker, and H. D. Griffiths, “Automatic target recognition using multi-diversity radar,” IET Radar Sonar & Navigation 1(6), 470–478 (2007). [CrossRef]
9. J. Tian, J. Sun, G. Wang, Y. Wang, and W. Tan, “Multiband radar signal coherent fusion processing with IAA and ap FFT,” Signal Process. lett. 20 (5), 463–466 (2013). [CrossRef]
10. M. Edwards, D. Madsen, C. Stringham, A. Margulis, B. Wicks, and D. Long, “microASAR: A small, robust LFM-CW SAR for operation on UAVs and small aircraft,” in Proc. in Geoscience and Remote Sensing Symposium (IEEE, 2008), pp. V-514–V-517. [CrossRef]
11. G. Wang, C. Gu, T. Inoue, and C. Li, “A hybrid FMCW-interferometry radar for indoor precise positioning and versatile life activity monitoring,” IEEE Trans. Microw. Theory Tech. 62(11), 2812–2822 (2014). [CrossRef]
12. N. Joram, B. Al-Qudsi, J. Wagner, A. Strobel, and F. Ellinger, “Design of a multi-band FMCW radar module,” in 10th Workshop on Positioning Navigation and Communication (IEEE, 2013), pp. 1–6. [CrossRef]
13. M. M. Jatlaoui, F. Chebila, P. Pons, and H. Aubert, “New micro-sensors identification techniques based on reconfigurable multi-band scatterers,” in Asia Pacific Microwave Conference (IEEE, 2009), pp. 968–971. [CrossRef]
14. Z. Li and K. Wu, “On the leakage of FMCW radar front-end receiver,” in Global Symposium on Millimeter Waves, (IEEE, 2008), pp. 127–130. [CrossRef]
15. J. M. Muñoz-Ferreras and R. Gómez-García, “A deramping-based multiband radar sensor concept with enhanced ISAR capabilities,” IEEE Sens. J. 13(9), 3361–3368 (2013). [CrossRef]
16. X. Zou, B. Lu, W. Pan, L. Yan, A. Stöhr, and J. Yao, “Photonics for microwave measurements,” Laser Photonics Rev. 10(5), 711–734 (2016). [CrossRef]
17. P. Ghelfi, F. Laghezza, F. Scotti, G. Serafino, S. Pinna, D. Onori, E. Lazzeri, and A. Bogoni, “Photonics in radar systems: RF integration for state of-the-art functionality,” IEEE Microw. Mag. 6(8), 74–82 (2015). [CrossRef]
18. P. Ghelfi, F. Laghezza, F. Scotti, D. Onori, and A. Bogoni, “Photonics for radars operating on multiple coherent bands,” J. Lightwave Technol. 34(2), 500–507 (2016). [CrossRef]
20. H. Yu, M. Chen, Q. Guo, M. Hoekman, H. Chen, A. Leinse, R. G. Heideman, R. Mateman, S. Yang, and S. Xie, “All-optical full-band RF receiver based on an integrated ultra-high-Q bandpass filter,” J. Lightwave Technol. 34(2), 701–706 (2016). [CrossRef]
21. J. Li, J. Xiao, and X. Song, “Full-band direct-conversion receiver with enhanced port isolation and I/Q phase balance using microwave photonic I/Q mixer (invited paper),” Chin. Opt. Lett. 15(1), 010014 (2017).
22. I. L. Newberg, C. M. Gee, G. D. Thurmond, and H. W. Yen, “Long microwave delay fiber-optic link for radar testing,” IEEE Trans. Microw. Theory Tech. 38(5), 664–666 (1990). [CrossRef]