1. Introduction
Growing demands for higher data rates, lower communication latency, better energy efficiency, and higher system reliability present a multitude of technical challenges in the development of multifunction radio frequency (RF) electronics and systems amid emerging 5G/6G wireless applications such as intelligent Internet of Things (IoT) and multi-input multi-output (MIMO) systems [
1], [
2], [
3], [
4]. Indeed, these challenges are closely related to the foundational strain and incapability of the current front-end architectures of the wireless communication system, which necessitate a paradigm shift despite their great success over more than a century. A promising candidate to meet the above expectations is the in-band full-duplex (IBFD) technique. This scheme allows simultaneous in-band signal transmission and reception, which doubles the data throughput and drastically reduces system latency. Moreover, IBFD systems are set to naturally operate in both radio and radar modes, and they offer a straightforward way to develop joint radar and communication systems or integrated sensing and communication (ISAC) systems for the multifunction interplay and fusion of future intelligent wireless systems [
5], [
6], [
7], [
8].
A fundamental success in the realization of IBFD wireless systems hinges on the development of a robust suppression of transmitter (Tx)-to-receiver (Rx) cross-channel interference (also called self-interference, SI) to the noise floor over an appreciable frequency bandwidth in the analog domain. Analog domain SI cancellation (SIC) must be at least 60 dB [
9] to avoid saturation of the analog-to-digital (A/D) converter (ADC), which is typically realized in a two-stage implementation distributed along the front-end analog path. As the primary stage of SI suppression, active or passive circulators [
10], [
11], [
12], [
13], [
14], [
15] and dual-polarized (DP) [
16], [
17], [
18], [
19] antennas are two popular ways for realizing compact- and single-antenna configurations, which multiplex Tx and Rx signals to exploit wave propagation nonreciprocity and polarization orthogonality, respectively, leading to about 25-45 dB SIC.
To realize the IBFD concept in practice, secondary SIC schemes are necessary to power up the isolation to more than 60 dB, where active SIC techniques seemingly dominate the current state of the art [
9]. In principle, an active SIC scheme relies on the creation of a replica of the Tx signal, manipulating its phase and amplitude to match the residual SI signal and then subtracting it from the Rx chain to further enhance the Tx-to-Rx isolation. In active SIC circuits, different components such as attenuators and phase shifters are subject to various phase velocities related to the propagating signal, with different dispersion against the main path for the Tx and Rx signals. As a result, active SIC methods are essentially narrowband. To create wideband active SIC techniques, more taps should be used to manipulate the tapped Tx signal in more discrete points—albeit at the expense of significantly increased complexity. In addition, active SIC approaches can feature frequency tunability of the RF front-end, which is a vital characteristic for future wireless systems. The best performance reported to date for tunable integrated IBFD systems with a single antenna is associated with an active technique, in which an analog isolation of 65 dB over 80 MHz bandwidth was reported [
20], although the design suffers from a large insertion loss of more than 3 dB in both the Tx and Rx signal paths.
Nevertheless, wideband active SIC techniques are beleaguered by numerous disadvantages that pose an insuperable roadblock to the widespread and successful commercialization of IBFD systems. This is because their performance gain in terms of throughput and latency is not worth their many associated disadvantages, making IBFD systems unappealing in comparison with other existing architectures for wireless systems. A number of drawbacks can be anticipated, including substantial increased direct current (DC) power consumption, significant design complexity, increased production cost, limited tunability, low power-handling capability, and contribution to nonlinear distortions due to the use of active components [
21]. Hence, it is concluded that—despite numerous studies and many attempts in recent years [
20], [
21], [
22], [
23], [
24], [
25], [
26], [
27], [
28], [
29], [
30]—IBFD solutions are still far from being matured and practically useful for the much-publicized ubiquitous wireless systems. Therefore, a disruptive approach is required to achieve the essential combination of high performance (in terms of bandwidth, isolation, high-power handling, and tunability), low energy consumption, cost-effective realization, and simpler architecture.
Nonreciprocity and multiplexing are well-known methods to isolate two RF signal channels. While various multiplexing schemes create multiple separate orthogonal channels between the Rx and Tx to increase the network capacity or isolate Tx and Rx signals, nonreciprocal components are set to isolate Tx and Rx signals over the same channel by separating forward- and backward-directed waves. In this work, we come up with a way to combine ferromagnetic nonreciprocity and polarization-division multiplexing (PDM) schemes and achieve in-band ultra-high isolation between Rx and Tx channels. A DP antenna is designed to achieve PDM by multiplexing in-band transmitted and received signals using vertical and horizontal linear polarizations. The key component in this proposed solution for combining PDM and nonreciprocity is a concurrent dual-mode (DM) circulator, with each mode assigned to one polarization. We proposed and demonstrated a concurrent DM circulator for the first time in our recent work [
31]; it operates simultaneously over two orthogonal waveguide modes (TE
10 and TE
20). Moreover, the DP antenna and the DM circulator are integrated within a DM waveguide that plays an intermediate role between those two components by converting orthogonally polarized Rx and Tx signals into the orthogonal waveguide modes.
In addition, this work has led to the discovery and demonstration of a unique passive tunable secondary SIC scheme that is inherent to the proposed module and can boost the isolation substantially. In this scheme, mode conversion within the DM waveguide is exploited by imposing an asymmetrical magnetic bias to ferrite, which leads to the generation of a replica of the Tx signal whose amplitude and phase can be separately manipulated for the purpose of SIC. Finally, our approach is validated through measurements and system-level demonstrations. Through the experiments, this highly tunable module is shown to feature an analog isolation of 50, 70, and 80 dB over a 340, 101, and 33 MHz bandwidth at 5 GHz; excellent isolation tunability at adjacent frequencies; a drastically simple and low-cost design; and high power-handling capability and ultra-high linearity, due to being an all-passive structure.
2. Combining ferromagnetic nonreciprocity with PDM
The conventional circulator-antenna combination (
Fig. 1(a)) has been widely used in radar applications. However, this configuration cannot accommodate DP antennas, as conventional circulators operate in a single-polarization mode. Therefore, inspired by and based on this conventional technique, we propose a counterpart topology that involves a DP antenna, a DM circulator, and a DM waveguide to accommodate both polarization and mode diversity. In essence, we treat the DP antenna as a dual-function component for PDM. Therefore, to combine PDM with nonreciprocity, we first devised a DM circulator that can operate as a dual-function nonreciprocal component that is compatible with the DP antenna. Next, we integrated the DM circulator and the DP antenna within a DM waveguide. The conceptual schematic of this integrated DM circulator and DP antenna module is depicted in
Fig. 1(b), where it is compared with the conventional single-mode and single-polarization configuration (
Fig. 1(a)). In
Fig. 1(b), the antenna and circulator are represented by two lines with different colors to indicate the DP and DM nature of these components, respectively. In this schematic, a chained combination of the horizontal polarization and the first waveguide mode (TE
10) is assigned to the Tx signal, while the vertical polarization and second waveguide mode (TE
20) are assigned to the Rx signal. The DM circulator is set to nonreciprocally and simultaneously circulate both the TE
10 and TE
20 waves between the three ports. Therefore, the Rx signal is isolated from the Tx port due to both the nonreciprocity of the circulator and the mode-orthogonality of the signals. Moreover, as discussed in the next section, the ferrites within the DM waveguide offer an unprecedented passive tunable secondary SIC scheme that is fundamentally responsible for frequency-selective SI suppression.
Fig. 1(c) provides a schematic of the proposed module for demonstration, consisting of six layers (a detailed illustration is provided in Appendix A Section S1 Fig. S1 and the design parameters are provided in Appendix A Table S1). The first layer encompasses a substrate-integrated waveguide (SIW) [
32], a DM circulator, and a T-shaped radiating aperture for patch antenna excitation. The second and third layers are used to deploy shielded striplines to feed the Tx and Rx signals that excite the waveguide on one end and get connected to external Tx/Rx ports on the other. Although the Tx and Rx ports are implemented in each other’s vicinity, electromagnetic (EM) coupling between them is suppressed for two reasons. First, the stripline and SIW technologies are adopted since the shielding geometry of these types of transmission lines makes it possible to minimize EM radiation and avoid unwanted coupling between two signals. Second, a number of vias in these layers are deployed to suppress pernicious coupling related to discontinuities and fabrication tolerances. Layers 4 and 5 form a thick DM cavity to increase the bandwidth of the antenna. Finally, a square patch antenna is printed on layer 6 for radiation.
The EM wave propagation within the proposed module is simulated and depicted in
Fig. 2 for illustration. In this case, the Tx and Rx signals propagate orthogonally in both the air and circuit domain, due to the polarization and mode orthogonality, respectively. The DP antenna transmits and receives the wireless signal in two different polarizations. The T-shaped slot on the top wall of the SIW in substrate 1 orthogonally transforms the horizontal and vertical polarized waves to the corresponding TE
10 and TE
20 waveguide modes within the DM SIW, respectively. The orthogonal TE
10 and TE
20 modes can be labeled as common and differential modes, where their electric field patterns are respectively symmetrical and anti-symmetrical with regard to the center plane. In
Fig. 2, the two modes propagate in the DM waveguide within the bottom substrate and are distinguishable from their respective electric field patterns. The electric field patterns for the TE
10 and TE
20 modes have a lateral distribution like a half sine wave and a full sine wave, respectively. The circulator is integrated within the waveguide, simply exploiting the ferromagnetic nonreciprocity of two pieces of ferrite [
31]. The ferrites, which are biased with an external static magnetic field, circulate the waves inside the waveguide nonreciprocally relative to the direction of the wave, while operating similarly for both the TE
10 and TE
20 modes. This is the first time a circulator operates simultaneously with two modes, which is the main reason for materializing the proposed concept. To achieve the DM circulator, we have exploited ferrites exhibiting a negative effective permeability upon the application of a proper magnetic biasing [
31], [
33], [
34]. Ferrite slabs biased into a negative permeability exhibit three significant features conjointly. First, since the dielectric constant of the ferrite (> 10) is much higher than that of the waveguide substrate (2.9), the EM fields will basically be confined to the ferrite medium. In other words, the ferrites behave as a dielectric waveguide, trapping the EM waves. Second, since the effective permeability of the ferrite is negative, the wave vector inside the ferrite becomes imaginary, confining the waves more into the interface of the ferrite and forcing the waves to propagate locally at the ferrite interface. Third, due to nonreciprocity, the wave propagation occurs only at one interface of the ferrite, which interchanges as the direction of propagation or direction of magnetic bias changes. In this work, the wave incident from the right to left side of the waveguide propagates at the inner interfaces of the ferrite slabs, while the wave incident from the left to right side of the waveguide propagates at the outer interface of the ferrite slabs.
Generally speaking, when a vertically polarized Rx signal excites the patch antenna, it travels downward through the cavity and excites the longitudinal sections of the T-shaped aperture, resulting in the formation of the TE20 mode in the waveguide. The TE20 mode wave travels toward the circulator, and—due to the nonreciprocity of the ferrites—the wave travels between the interior interfaces of two ferrite pieces, reaching the other end of the waveguide, where it magnetically couples into a stripline at the top layer through a differential aperture. The aperture is differential because only the TE20 mode, which has longitudinal magnetic fields beneath the aperture, can couple into it; the TE10 mode cannot couple into it, since it has no longitudinal component of the magnetic fields beneath the aperture. The stripline is terminated into port 2, which is assigned to the Rx signal.
On the other hand, the Tx signal incident on port 1 propagates on a stripline, which is divided into two identical and parallel lines in common modes. Two signals in two striplines travel toward the waveguide-circulator junction through a via, which transfers the RF signal from the top layer to the bottom layer. Two common-mode signals incident from the side ports of the circulator are guided to the outer interfaces of the ferrite slabs as they propagate toward the antenna. Since two segments of the signal are in common mode, they excite the TE10 mode inside the waveguide. The TE10 mode excites the transverse section of the aperture and travels upward though the cavity, causing the patch antenna to radiate in the horizontal polarization.
The proposed module was fabricated, and the measurement results are reported in
Fig. 3; upon comparison with the simulation results, they show excellent agreement, which validates the proposed theory. It can be seen from
Fig. 3(d) that the module has a bandwidth of about 700 MHz for a return loss of 10 dB. Furthermore,
Fig. 3(e) reveals that the SIC from the Tx-to-Rx port is better than 50 dB over a 320 MHz bandwidth. The complete
S-parameter response of the device is depicted in Appendix A Fig. S2 which is available in Appendix A Section S2. It seems that isolation is limited by the fabrication tolerance, as asymmetries are created within the structure, leading to undesired cross-coupling between the two modes and/or polarizations, which in turn degrade the isolation. However, in the next section, we will show that the cross-coupling mechanism can be exploited to our advantage by creating intentional controlled asymmetry to cancel the residual SI.
3. Inherent passive tunable secondary SIC
In this section, we demonstrate that, by slightly perturbing the magnetic bias of the ferrites, the isolation can be boosted to more than 70 dB over a relatively wide bandwidth, due to an inherent passive tunable secondary SIC scheme that is naturally embedded in the proposed approach. It is generally known that the modes are coupled into each other in the presence of asymmetric perturbations in a waveguide [
33], [
34]. Hence, we propose that a coupling from the TE
10 (Tx) to TE
20 (Rx) signal can be identified as an SIC signal if both the amplitude and phase of the generated signal are controllable. In this work, phase and amplitude tuning are accomplished and demonstrated through a partially asymmetric magnetic bias perturbation (PAMBP). By “asymmetric” magnetic bias perturbation, we imply that the magnetic bias applied on two ferrite pieces is slightly different, resulting in a small mode conversion from the TE
10 to TE
20 mode and generating a copy of the Tx signal in the Rx channel. The amplitude of the generated signal depends on the level of the magnetic bias difference. In comparison, “partially asymmetric” denotes that the magnetic bias perturbation is only applied to a fraction of the ferrite. The length of the asymmetrical perturbation area is set to determine the phase of the generated signal and can be smoothly controlled, as will be shown shortly. Furthermore, as the Tx signal is input to the center of the ferrite junction and travels nonreciprocally toward the antenna end, it matters which end of the ferrite the asymmetry is applied to. In other words, the location of the PAMBP affects the phase of the generated signal as well. A schematic diagram is shown in
Fig. 4(a) to illustrate how the PAMBP disturbs the symmetry of the DM waveguide, leading to the generation of a tunable TE
20 signal. In this figure, a common-mode signal is applied laterally to a DM waveguide, exciting a TE
10 wave to propagate toward port 2. By applying a PAMBP, a TE
20 signal is generated, propagating toward both ports. The TE
20 signal directed toward port 2 is not shown in the figure, since the TE
10 signal is significantly stronger. As indicated in the figure, the amplitude and phase of the TE
20 wave can be tuned separately by adjusting the strength of the PAMBP and the length of the area over which the PAMBP is applied. More explanations and simulations about the mechanism of amplitude and phase tuning in the proposed passive tunable secondary SIC scheme can be found in Appendix A Section S3. To elucidate the concept of mode conversion, Fig. S3 in Appendix A conceptually illustrates the process of mode conversion for amplitude and phase tuning separately. Additionally, to substantiate the discussed concept, simulations of the electric field pattern for selected cases of amplitude and phase tuning are presented in Appendix A Fig. S4.
We have simulated the EM propagation of waves with the effect of the PAMBP to elaborate the proposed passive tunable secondary SIC scheme; the phase and amplitude variations are plotted in
Figs. 4(b) and
(c) for a center frequency of 5 GHz. In these figures, the Δ
H axis denotes the difference in magnetic bias strength in the two ferrite pieces. The
LfΔH axis denotes the length of the asymmetry for two orientations. The left section indicates the origin of the PAMBP on the left end of the ferrite close to the Rx port, which gradually increases toward the right side. The right section of the axis indicates that the PAMBP originates at the right end of the ferrite close to the antenna and increases toward the other end.
From the figures, it can be seen that the phase of the SIC signal is only dependent on the length and location of the PAMBP. Moreover, whether the PAMBP level is positive or negative, the phase undergoes an abrupt 180° phase shift. Considering all these aspects, it is observed that a continuous and greater than 360° phase tuning is accomplished. The amplitude of the generated SIC signal depends on all three parameters, although the “level” of the asymmetry has a dominant effect, which can be used to tune the desired amplitude independently. Thus, with the three variable parameters, it is demonstrated that a full manipulation of the generated SIC signal is always accomplished.
We should point out two notes here. First, since the PAMBP can change the phase of the SIC signal, it can change and tune the frequency of the peak isolation, as the phase and frequency are related to each other. Second, since the generated SIC signal is manipulated within the same waveguide along which the leakage signal is propagating, both signals have similar dispersions, which results in a larger isolation bandwidth when compared with other secondary SIC approaches. In other words, since the secondary SIC is inherent to the device, and all leakage and SIC signals are set to share the same medium for propagation, the dispersion for all signals is similar, and one can expect to yield a larger isolation bandwidth. The bottleneck in achieving such a larger isolation bandwidth is related to the dynamic of the antenna and reflection from nearby objects.
The proposed module has been measured under various PAMBP conditions; the results are depicted in
Figs. 4(d)-(f).
Fig. 4(d) depicts a frequency tunable response with more than 50 dB isolation over a bandwidth of 340 MHz at 5 GHz, while the frequency tunability is demonstrated over the whole operating bandwidth of the device with a small degradation. Similarly,
Fig. 4(e) indicates a more challenging criterion of 70 dB isolation, which is realized over a bandwidth of 101 MHz, while being effectively tuned at adjacent frequencies. Finally,
Fig. 4(f) shows an even larger isolation of 80 dB with a 33 MHz bandwidth at the center frequency, again with impressive frequency tunability. The tuning mechanism is achieved using RF magnetic sheets as depicted in Appendix A Fig. S5. More information about the frequency tuning of the proposed component is discussed in Appendix A Section S4.
These figures exhibit two important features: First, the isolation level can be boosted to as high as 80 dB over a considerable bandwidth, which stands out as a significant performance, given the current state of the art. Second, the peak isolation is quite frequency-tunable, which is a vital feature for full-duplex systems.
4. Full-duplex transceiver demonstration
To assess the proposed module’s performance in a real full-duplex system, two sets of measurements are performed. In the first demonstration, the proposed module’s isolation is measured while a real pulse-shaped modulated signal is applied to its Tx port in a realistic IBFD scenario. Moreover, to compare the proposed module’s performance with state-of-the-art approaches, we have designed a single-tap active SI canceler circuit as shown in Appendix A Fig. S6, which is used in conjunction with the DP antenna (see Section S5 in Appendix for more information).
The measurement results for two scenarios are shown in
Fig. 5 (photo of the measurement setup is provided in Appendix A Fig. S7). In these measurements, the Tx chain transmits a 64-quadrature amplitude modulation (QAM) signal with a bandwidth of 90 MHz and an average power of 10 dBm. As the device is passive, we could use a higher input power; however, the nonlinearity of the power amplifier (PA) would result in a significant channel spectral regrowth that would make the signal bandwidth control difficult.
The measurements for the DP antenna and the active SIC circuit are shown in
Fig. 5(b). This experiment demonstrates that up to 42 dB of SIC can be produced by the DP antenna alone over a wide bandwidth. However, by adjusting the active SIC circuit, the SIC is boosted over a significantly narrow bandwidth (less than 5 MHz). As an alternative,
Fig. 5(c) shows the measurement of the proposed module without the RF canceler. This figure illustrates that the proposed module can suppress the SI signal by up to 70 dB over the whole 90 MHz bandwidth of the input signal. The major difference between the two results and the outstanding performance of the proposed all-passive module demonstrates that our module is an enabling and effective solution for full-duplex wireless systems. More information regarding this measurement is available in Appendix A Section S6.
In the second measurement, a communication link is created between the proposed module and a separate transmitting antenna to evaluate the error vector magnitude (EVM) of the Rx in the presence of a strong SI signal from the Tx port. The measurement configuration for the communication link is depicted in
Fig. 6(a), where a distance of 60 cm separates the two nodes, resulting in a 42 dB RF path loss. The schematic and photo for this measurement setup is depicted in Appendix A Fig. S8. However, given that each antenna has an average gain of 8 dBi, while long cables, connectors, and the insertion losses of the antennas contribute to an overall RF loss of around 3 dB, the power loss from node 2 to node 1 would be approximately 29 dB. In this experiment, node 1’s Tx port broadcasts an 80 MHz, 256-QAM signal at 10 dBm average power, which is perceived as a noise or interference signal for node 1’s Rx. The Tx port of node 2 transmits an 80 MHz signal at various QAMs and various output powers. The bandwidth in this experiment is limited to 80 MHz, since the secondary vector signal generator (VSG) in our laboratory could provide modulated signals up to 80 MHz. The objective of this experiment is to assess the impact of SI on the received signal for a variety of modulation schemes and input power variations. Thus, the output power of node 2 varies from +10 to -30 dBm, translating to a received power at node 1’s Rx of -19 to -54 dBm.
Fig. 6(b) displays the measured EVM for several modulation schemes while the desired signal’s received power was swept as previously described. It can be seen that, when the input power drops, the EVM rises as the signal-to-noise ratio (SNR) rises. In addition, as the EVM is independent from the modulation scheme, measurements of EVM for all modulations produce similar results. However, deploying different modulation schemes requires different EVM levels in digital communications. For example, requirements for 5G QAM schemes specify that the maximum EVMs for quadrature phase-shift keying (QPSK), 16-QAM, 64-QAM, and 256-QAM signals are 17.5%, 12.5%, 8.0%, and 3.5%, respectively [
35].
The QAM constellations of the received signal at an input power of -19 dBm and for 256, 64, and 16 QAMs are shown in
Figs. 6(c)-(e), while the constellations for the 256-, 64-, and 16-QAM schemes are shown in
Figs. 6(f)-(h) at their maximum allowable values, which correspond to -34, -40, and -44 dBm input powers, respectively. In addition, eye diagrams of the received signals for QPSK and 16 QAMs are measured and compared at an input power of -19 dBm and their maximum allowable values of -47 and -44 dBm, respectively, as depicted in
Figs. 6(i)-(l). These measurements demonstrate that, in the implemented full-duplex radio under strong interference from the Tx, the Rx is capable of sensing and demodulating various complex signals. Furthermore, the measured EVM for each received modulated signal reveals that more than 70 dB of the SI is successfully suppressed. Detailed information regarding communication link measurement is provided in Appendix A Section S7. Moreover, information regarding methodology for device fabrication, simulation, and measurements are provided in Appendix A Section S8.
Table 1 [
20], [
24], [
27], [
39], [
40], [
41] summarizes and compares our work with published state-of-the-art works on full-duplex radio systems focusing on the analog domain. This table highlights the distinct advantages of our work across multiple dimensions. First, our approach achieves the highest figure of merit for the total isolation per bandwidth, surpassing all previous works. Second, our work presents excellent tunability—a special feature that is notably absent in many existing studies and solutions. Third, leveraging the SIW structure, our solution demonstrates the capability to operate at significantly higher power levels compared with its integrated circuit (IC) counterparts. This attribute facilitates seamless integration with other active SIC approaches, thereby achieving enhanced isolation in the analog domain for ultra-high-power applications. More importantly, our scheme is entirely passive, thereby presenting substantial simplification and cost reduction vis-à-vis those state-of-the-art alternatives.
5. Conclusions
In this paper, we proposed a fundamental approach to enabling the development of a full-duplex transceiver by achieving ultra-high isolation in the analog domain at the front-end. Theoretically, our approach is based on combining ferromagnetic nonreciprocity with PDM through a concurrent DM circulator. Furthermore, we demonstrated that the proposed DM circulator contains a unique inherent passive tunable secondary SIC scheme that can boost the isolation. We envision that the passive tunable secondary SIC scheme will be a game-changing solution for wideband SIC approaches, as it outperforms existing state-of-the-art techniques while being a passive component. Various measurements and system-level demonstrations validated and confirmed that our proposed solution reaches 50, 70, and 80 dB isolation over 340, 101, and 33 MHz bandwidth, respectively, with excellent frequency tunability.
To benefit from IBFD in future wireless systems and to deploy such systems ubiquitously, the proposed solutions need to be low-cost, low power-consuming, highly tunable, able to handle high RF powers, and able to avoid extra complex circuits. Our proposed all-passive solution has the potential to fulfill these requirements for IBFD systems, as it brings together all the mentioned criteria at once, while exceeding state-of-the-art solutions in terms of performance. Furthermore, the antenna in our solution can be easily extended into an array antenna or many other types of antennas, exhibiting great adaptivity.
Moreover, we have introduced a fundamentally new approach to the limited library of existing solutions for analog SIC. Interestingly, our approach can also be integrated with those techniques to target new horizons for IBFD systems. For example, for some specific applications in base stations, the Tx power may need to be as high as 46 dBm. In this case, the analog SIC must be greater than 100 dB, which is beyond the grasp of current systems due to both limited power handling and isolation. However, our solution can be adopted as the backbone architecture and integrated with existing active SIC techniques and/or large array antennas to boost the isolation to astonishing values of greater than 100 dB, while supporting 46 dBm power transmission at its output.
Finally, we envision that our theoretical foundation has the promising potential to open up new research horizons in a range of disciplines. For example, recent advances in materials engineering have led to remarkable breakthroughs in the development of novel magnetic, ferromagnetic, and multiferroic thin films, which can replace ferrite materials [
40], [
37], [
38]. These materials can be adopted in our approach to scale the design to small footprints and to millimeter-wave frequencies. Exploring the theory of combined mode and polarization diversity in different types of active circulators serves as another example.
Acknowledgments
This work was supported by a Natural Sciences and Engineering Research Council (NSERC)-sponsored Industrial Research Chair program, an NSERC Discovery Grant, and also in part by the Fonds de recherche du Québec Nature et technologies (FRQNT) Doctoral Fellowship of Amir Afshani funded by the Government of Québec Province. Also, the authors would like to thank Traian Antonescu, Steve Dubé, and Jules Gauthier of the Poly-Grames Research Center, Polytechnique Montréal, Quebec, Canada, for their support in the realization of the prototypes.
Compliance with ethics guidelines
Amir Afshani and Ke Wu declare that they have no conflict of interest or financial conflicts to disclose.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.02.007.
PDF
(3166KB)
