3.1. Fundamental Single-Line System
First, we numerically designed and evaluated waveform-selective metasurfaces using a co-simulation method of ANSYS Electronics Desktop 2020R2 (see “Simulation Method” of the Methods Section and Figure S3), which were experimentally validated (see “Measurement Samples” and “Measurement Methods” of the Methods Section and Figure S4). In the absence of waveform-selective responses, the monopole antennas used in Fig. 1 were numerically adjusted to efficiently radiate signals at approximately 2.4 GHz, as shown in Fig. 1d. However, the transmittance varied depending on the incident pulse width, as explained above and numerically demonstrated in Fig. 1e (see Figure S5 for power dependence and Figure S6 for pulse width dependence). In particular, our waveform-selective metasurfaces were also designed to operate at approximately 2.4 GHz by properly choosing design parameters including the conducting dimensions, substrate thickness, etc. Importantly, the pulsed sine waves used in this study (simply called pulses below) had very narrow spectra compared to the bandwidths of both the antennas and the waveform-selective metasurfaces, as shown in Fig. 1d (cf. the bandwidth of the pulse in Fig. 1e). This ensured that our antennas did not rely on the difference in the frequency spectrum of pulsed sine waves to change the transmission characteristics. With a limited power level, these metasurfaces showed the same transmittance in the time domain, as their diodes had not yet been turned on (Figure S7).
However, by increasing the input power to 10 dBm, each type of waveform-selective metasurface maximized transmittance at a different time period even at a fixed oscillating frequency of 2.4 GHz because of their transient, waveform-selective absorbing mechanisms. For instance, as seen in the left panel of Fig. 1f, when the L-based, the parallel and the C-based waveform-selective meatsurfaces were used between the transmitter and the receiver, a transmittance peak appeared near 30 ns, 300 ns and 10 µs, respectively. In other words, most of the energy of a short pulse, an intermediate pulse and a long pulse can be effectively transmitted by the L-based, the parallel and the C-based waveform-selective metasurfaces, respectively. Such transient transmittance was experimentally observed but with a small frequency change to 2.42 GHz as shown in the right panel of Fig. 1f (see Fig. 1a for one of the measured samples and Figure S8 for the frequency characteristics). Compared to the simulation results (the left panel of Fig. 1f) where each transmittance peak was at least 10 dB larger than the transmittances of the other curves, in the measurement results (the right panel) the vertical gaps between the transmittance peaks and the second highest transmittances were reduced to smaller values. This was in part because the lumped circuit components were soldered by hand, which led to a difference in the amount of the solder used and thus resulted in minor frequency shifts. With small frequency adjustments less than 40 MHz, larger contrasts between the three transmittances were experimentally obtained in Figure S9. More details are seen in Figure S10 and Figure S11. Note that these waveform-selective mechanisms can be characterized by the values of the circuit elements, such as capacitance and inductance.[49] For instance, the C-based waveform-selective metasurface shown in the right panel of Fig. 1f was used again in Fig. 1g but with C increased from 1 nF to 1 µF and 100 µF. As a result, the transient response appeared even over the order of several hundred milliseconds, which corresponds to billions of cycles (see Figure S12 for related simulation results). Note that this ultra-transient response was readily designed and experimentally validated by replacing discrete capacitor components of the C-based waveform-selective metasurfaces (i.e., replacing C of Fig. 1b with a larger capacitance). The other types of waveform-selective metasurfaces can operate in a similar time period by using larger circuit component values. Theoretically, conventional antennas without waveform-selective metasurfaces may be able to achieve the same performance by using an extremely large quality factor.[56] Realistically, however, this mechanism becomes ineffective due to the presence of even a minor lossy component.
3.2. Combined Selective Multi-Line System
Next, we numerically and experimentally demonstrated that the propagation direction of the transient antenna design introduced in Fig. 1 can be readily extended by increasing the number of variable transmission lines, namely, the number of waveform-selective metasurface lines. Figure 2a shows that a transmitting antenna expressed by an AC source, a switch and input impedance Z0 is connected to three variable transmission lines ZC, ZL and ZP representing C-based, L-based and parallel waveform-selective metasurface lines, respectively. Z1, Z2 and Z3 denote the impedances of the three receivers. This concept is specifically realized by the schematic and the measurement sample of Fig. 2b, where the three lines used in Fig. 1f are connected to form a Y shape. In this design, the transmitter is positioned at the centre to transmit a signal to the three receivers at the terminals of the metasurface lines. Figure 2c shows both numerically and experimentally that each receiver featured a maximum transient transmittance at a different time period since each waveform-selective metasurface line efficiently transmitted the signal during a different time slot.
As another planar scenario, hexagonal patches were fully deployed on a conducting ground plane to demonstrate that radiation characteristics can be changed over the entire 2D surface in Fig. 2d. This structure had a transmitting monopole and three types of waveform-selective metasurfaces, as shown in the Y-shaped structure in Fig. 2b. However, each waveform-selective metasurface occupied one-third of the area around the transmitter so that the surrounding waveform-selective metasurface was changed by every 120 degrees of reference angle to the transmitter. Under these circumstances, due to the presence of the different waveform-selective metasurfaces, the surface wave generated by the omnidirectional transmitter was expected to propagate over the ground plane unidirectionally, unlike conventional static metasurfaces.[17, 53] As shown in Fig. 2e, the change in the radiation pattern of the transmitter was measured with the simplified method using three monopole receivers. In such transmittance measurements, an average transient transmittance peak first appeared between 0.01 and 0.1 µs in the L-based waveform-selective metasurface receiver area (Rx3), and then another such peak appeared between 0.1 and 1 µs in the parallel waveform-selective metasurface receiver area (Rx2). Finally, a larger transient transmittance was measured in the C-based waveform-selective metasurface receiver area (Rx1) at approximately 10 µs. These results indicate that most of the energy of a short pulse, an intermediate pulse and a long pulse can be effectively radiated to the directions of the L-based, the parallel and the C-based waveform-selective metasurfaces, respectively. As fundamentally demonstrated in Fig. 1 and Fig. 2, these results confirm that the use of non-uniform waveform-selective metasurfaces can provide an additional degree of freedom to control surface waves and vary radiation characteristics even at the same frequency. Note that our antenna designs demonstrated in Fig. 2 were spatially constructed with different waveform-selective metasurfaces. This means that our antennas used the spatial dimensions as an additional degree of freedom, which is later exploited to selectively distinguish different waves even under simultaneous incidence.
3.3. System for Free-Space Wave Control
To demonstrate the applicability to waves propagating in a wireless network or in free space, we also show that the proposed concept can be used to design a selective system for free-space wave control, as illustrated in Fig. 3a. In this figure, as seen in Fig. 2a, a transmitter is represented by an AC source, a switch and input impedance Z0. In the case of free-space wave control, however, C-based, L-based and parallel waveform-selective metasurfaces are used as simple planar layers in parallel to the transmitter and, thus, represented by variable shunt impedances ZC, ZL and ZP instead of transmission lines that correspond to free space. Z1, Z2 and Z3, similarly, are the impedances of the three receivers, respectively. To specifically design such a system, a transmitting monopole antenna (18 mm tall) is surrounded by a set of waveform-selective metasurface panels in Fig. 3b. Note that, for simplicity, this scenario used a ground plane and the mirror image of the monopole to readily evaluate the radiation characteristics of a dipole antenna suspended in free space, unlike the surface wave cases in Fig. 1 and Fig. 2. The waveform-selective metasurfaces used in Fig. 3b were based on slit (or slot) structures[44, 57] and, thus, designed to strongly transmit incoming signals at a resonant frequency (see the structure and frequency characteristics in Figure S13 and Figure S14, respectively, and the design parameters in Table S3 and Table S4). However, this resonant mechanism was either maintained or disrupted depending on the three types of waveform-selective metasurfaces used (two panels each, as shown in Fig. 3b). Note that even though the overall metasurface size in two panels was comparable to the wavelength (the dimension was equivalent to 0.9λ × 0.65λ) the single meta-atoms are subwavelength (~ 0.25λ). Thus, the metasurface effectively worked as a homogenized surface for the monopole antenna.[21, 22]
To observe the change in radiation characteristics in a simple manner, we used three monopole receivers placed in front of each waveform-selective metasurface (see Figure S15 and Table S5 and Table S6 for details and design parameters). Under this circumstance, the waveform-selective metasurface-based antenna steered the main lobe of the radiation by increasing the input power to 30 dBm at 3.85 GHz, as plotted in Fig. 3c (the frequency dependences are presented in Figure S16). Note that compared to the surface wave demonstrations seen in Fig. 1 and Fig. 2, this configuration used a higher frequency near 3.85 GHz instead of the frequency region near 2.4 GHz to reduce the dimensions of the measurement setup. At first, the signal was most efficiently radiated out of the panels containing the L-based waveform-selective metasurfaces, as obtained by Rx1 in Fig. 3b and shown by the black curve in Fig. 3c. Then, the L-based waveform-selective metasurface gradually reduced the transmittance, but the parallel waveform-selective metasurface started strongly transmitting the signal, where Rx2 measured the largest transmittance near 1 µs (see the red curve in Fig. 3c). This large transmittance, however, then decreased, while the C-based waveform-selective metasurface eventually maximized the transmittance, as shown by the blue curve in Fig. 3c.
Next, we compared the simulation results with measurement results. The simulated results in Fig. 3c were entirely consistent with the measurements of the sample shown in Fig. 3d, as plotted in Fig. 3e, except minor discrepancies in, for instance, the magnitude of transmittance and shift of time scales. Such discrepancies arose primarily due to the presence of some additional factors in the experimental case, including extra parasitic circuit components. Nonetheless, these measurements ensured the experimental feasibility of the proposed concept for free-space waves.
Furthermore, to experimentally visualize the radiation pattern over a 2D plane, we rotated the hexagonal metasurface prism of Fig. 3d every 10º whilst monitoring the transmittance from a monopole receiver in a far-field distance as seen in Fig. 3f. Consistent with the above simulation and measurement results, the radiation patterns plotted in Fig. 3g indicated that the maximum antenna directivity was achieved towards three different angles depending on the time period of the signal. The radiation pattern also clarified that the corner direction of each metasurface panel gave the highest transmittance as seen, for example, in 0.02 µs (the black lines in Fig. 3g), where the maximum transmittance was observed at 60º angle corresponding to the corner direction of the L-based waveform-selective metasurface panels. The side-lobe level in this time was 10 dB lower than the peak, indicating that receiving antennas located in other directions would receive low-intensity signals from the transmitter (see Figure S17 for the full time-varying far-field pattern). Besides the above 2D far-field profile, to get a clearer picture of interference paths, such as crosstalk, we have estimated the coupling from the transmitter to a receiver via the other two external antennas, and the result is shown in Figure S18. The calculated crosstalk was much lower than the power through the main path while in the measurement they differed by at least 20 dB.
As a reference result, Figure S19 shows simulation results where all the transmittances remained the same and constant when the input power level was not large enough to turn the diodes on. Additionally, Figure S20 represents the simulation results using only one of the three waveform-selective metasurfaces for all six panels. As shown in Figure S20, transient transmittance varied if the incident power was sufficiently large, but all the receivers received the same amount of energy, which indicates that the radiation characteristic was transient but omnidirectional. Note that as mentioned in Fig. 2, the antenna design of Fig. 3 was also spatially constructed with different waveform-selective metasurfaces. In other words, spatial dimensions were used as an additional degree of freedom, which plays an important role to separate different pulsed signals even at the same time as demonstrated in the following part of our study. Also, as mentioned in Fig. 2, the results of Fig. 3 indicate that most of the energy of a pulsed signal is effectively radiated to different directions depending on the pulse width (e.g., a short pulse to the direction of the L-based waveform-selective metasurface receiver).
3.4. Passive Variable Sensor
Now that we have so far designed and validated the antennas to vary radiation characteristics at the same frequency depending on the pulse width of a surface wave and a free-space wave, we use the proposed metasurface-based antennas in three different situations to explore how such ultra-transient directional antennas can be used for practical applications, even under the restriction of a single frequency band due to a limited frequency resource. In Fig. 3, the antenna was shown to steer the main lobe or transmit a signal to different receivers, which fits in existing wireless communication environments to vary the radiation characteristics and avoid electromagnetic interference. However, we propose an additional application using the reflected waveform of the transmitted signal to detect a scattering object as a passive yet variable sensor. As shown in Fig. 4a, a copper plate (51 mm tall and 70 mm wide) was placed in front of the antenna shown in Fig. 3. Note that we changed the location of the copper plate, which also influenced the reflectance in the time domain as the radiation characteristics of the antenna depend on the pulse width. Therefore, these results were compared to the reflectance without the copper plate to detect the difference in between and where the plate was positioned.
Under these circumstances, as shown in Figure S21a and more clarified in Fig. 4b, the transient reflectance experimentally increased during different time slots depending on the conductor position. Specifically, in contrast to the result obtained without the copper plate (Figure S21a), when the copper plate was located in front of the L-based waveform-selective metasurface panels, which efficiently radiated an electromagnetic wave during the initial time period, there was a larger transient reflectance until 0.2 µs. Similarly, the transient reflectance increased at approximately 1 µs and 10 µs when the plate was positioned in front of the parallel and C-based waveform-selective metasurface panels, respectively. These differences in reflectance appeared even if more than one copper plate were deployed, as demonstrated in Fig. 4c and Figure S21b. As seen in the black curve of Fig. 4c, for instance, a reflectance increased during an initial time period and near 1 µs, when two copper plates were positioned in front of the L-based and parallel waveform-selective metasurfaces, since they independently increased transmittance during these time slots. These results indicate that multiple objects can be sensed by using several types of waveform-selective metasurfaces or non-uniform waveform-selective metasurfaces. Moreover, the distances between the copper plates and these waveform-selective metasurfaces were changed in Fig. 4d and Figure S21c. As seen in Fig. 4d, the increase in reflectance was relatively limited during an initial time period when the distance between one of the copper plates and the L-based waveform-selective metasurface increased. In contrast, the gap between the two curves in Fig. 4d was relatively small near 1 µs, since the distance of another copper plate to the parallel waveform-selective metasurface was fixed. Thus, these results demonstrate that the distance to a scattering object can be detected from the difference in reflected waveforms. Additional results are seen in Figure S22. In this figure, reflectances are shown as a function of the distance between the monopole antenna and a copper plate.
In conventional methods to detect the direction of a scattering object, an antenna or a sensor (or a radar) usually changes its direction or posture as seen in parabolic antennas, adjusts the input phase as seen in phased arrays or sweeps the frequency component as a chirp signal.[8] Without any of these active changes, however, our waveform-selective metasurface-based antennas can potentially detect the location of scattering objects since the antennas keep varying their main lobes depending on pulse width, which is exploited as a new degree of freedom in wireless communications. Also, the proposed method only used a single transmitter unlike phased arrays composed of several radiating elements and phase shifters or adjustment components. Moreover, no duplexer was needed to perform detection even though only a single antenna was used. It is noted that our prototype here preferentially selected only three different angles. To increase the angular resolution, the metasurface can be arranged as a polygon with a higher number of sides that are associated with different waveform-selective functionalities. The metasurface antenna here detected objects and instantly differentiated their distances, while the detection reliability was limited by the noise floor as seen in Fig. 4d. Therefore, the performance can be improved by several factors including the number of sides, complexity in the circuit realization and the monopole antenna setups.
3.5. Selective Reception under Simultaneous Incidence
The second application is the selective reception of simultaneous incidences. From the viewpoint of efficient communication systems/environments, many signals may travel at the same time to increase the entire spatial data transfer rate. However, none of the waveform-selective metasurfaces that have been reported thus far is capable of distinguishing different pulsed waves if the signals arrive at the same time since the pulses are received as a single combined signal.[42–48] We addressed this issue in Fig. 5, where the central grounded monopole (effectively dipole) shown in Fig. 3 was placed near different types of waveform-selective metasurfaces that were spatially constructed to face different directions. This implies that our configuration exploited spatial dimensions as an additional degree of freedom (see Fig. 3a). Therefore, the proposed antenna design can be used for selectively receiving a pulsed signal from a particular incident angle.
Figure 5a shows the simulation model tested using the same configuration as that in Fig. 3 but using the central antenna as a receiver and the three external antennas as transmitters, each of which generated a sine wave of 3.85 GHz (30 dBm) with a different phase offset, specifically, -120, 0 and + 120 degrees for Tx1, Tx2 and Tx3, respectively. In this case, if the three signals have the same input power level, the total power received at the central receiver becomes ideally zero (or practically low) since the three incident signals cancel out each other, unless these signals are selectively filtered by waveform-selective metasurfaces. Therefore, the phase offset of the external transmitters clarify which signal dominated others even at the same frequency, as well as the direction of the signal source. Additionally, the distance between the central receiver and the external transmitters was reduced to ensure a sufficiently large input power for the measurements (see Figure S23 using the previous far-field distance).
First, when only one of the transmitters generated a signal, the simulation result was obtained, as plotted in the left panel of Fig. 5b. According to this result, transient transmittance was maximized during different time ranges, depending on which transmitter sent the signal. In addition, when more than one transmitter was activated, multiple transmittance peaks appeared, as plotted in the right panel of Fig. 5b. In particular, the three-source case turned out to follow the largest values among the three individual curves (see the left panel of Fig. 5b). To more clearly analyse the primary signal source, the voltage of the received signal was investigated in the time domain. According to the phases calculated in Fig. 5c (or the time of zero voltage), only one of the three signals was found to be relatively dominant across three time periods (i.e., approximately 0.05, 0.45 and 10 µs). This result indicates that by combining non-uniform waveform-selective metasurfaces with angular (or spatial) characteristics, different signals can be selectively received at the same frequency, even if the signals enter the proposed antenna at the same time. In addition, experimental validation is provided in Fig. 5d, Fig. 5e and Fig. 5f. In these measurements, basically we used the same conditions as the ones applied to the simulations in Fig. 5a, Fig. 5b and Fig. 5c except for the input power adjusted to 36 dBm (see the detailed measurement method in Figure S24 and “Measurement Methods” in the Methods Section). First, as seen in Fig. 5b, the measurement results seen in Fig. 5e showed similar transmittance peaks with minor shifts in the time domain due to several factors including the absence of solder and some additional parasitic circuit parameters that were not fully included in the simulations and affected the time constants of Fig. 5e. Despite these time-domain shifts, however, a single transmitted signal was found to dominate over the others, as shown in Fig. 5f. When all the three transmitters generated signals, the magnitude of the transmitted voltage slightly changed due to the interference between all the three signals. However, the phase remained almost the same as that of each dominant signal. These results experimentally confirm that the proposed antenna is capable of selectively receiving different pulses even under simultaneous incidences. Note that, as mentioned above, conventional design of waveform-selective metasurfaces or related antennas suffered from the fact that multiple pulsed signals arriving at the same time appeared as a single combined pulse that could not be distinguished by previous waveform selectivities. However, this drawback is overcome in the proposed antenna design concept by using spatial dimensions as an additional degree of freedom.
3.6. Mutually Selective Communication System
As the third application, we show that waveform selectivity can be exploited by multiple antennas to design a mutually pulse-width-selective communication system. In Fig. 3 and Fig. 5, only one antenna was permitted to selectively transmit/receive a pulsed signal in accordance with the pulse width. In contrast, the system proposed here is one step closer to a realistic wireless communication environment where ideally several antennas are expected to transmit and receive signals at the same time to increase the data transfer rate of the entire space. To realize such a system, the waveform-selective metasurface panels used only for a transmitter in Fig. 3 can be deployed for the three external antennas, since these external antennas are also given waveform-selective radiation characteristics, which means that mutually pulse-width-selective communications are established between any pair of antennas.
For simplicity, such a communication system was demonstrated using surface waves together with the three waveform-selective metasurface lines used in Fig. 2b. However, the three metasurface lines were combined to form a triangle shape instead of a Y shape in Fig. 6a. Each corner had a grounded monopole so that all the paths between the three antennas were connected by either a C-based, an L-based or a parallel waveform-selective metasurface to independently use different types of waveform selectivities without severely interfering with the other paths. Additionally, the antennas were programmed to generate a sine wave (10 dBm at 2.36 GHz) with a phase offset of -120, 0 or + 120 degrees.
Under these circumstances, when only one of the three antennas generated a signal, the transmittances to the other two receivers were maximized in different time slots, as shown in Fig. 6b. This demonstrates selective transmission to the receiving antennas. Additionally, Fig. 6c shows that different signals were selectively received in different time ranges. In other words, the communication system shown in Fig. 6a exhibited not only the selective transmission capability (as presented in Fig. 2 and Fig. 3) but also the selective reception capability (as shown in Fig. 5), both of which were observed in every single antenna element of Fig. 6a.
The proposed system was experimentally validated using the measurement sample of Fig. 6d and the method introduced in Fig. 5 (see Figure S24). As a result, similar trends were obtained in Fig. 6e, where signals were selectively transmitted to receivers. In addition, two transmittance peaks appeared in each case of Fig. 6f even if two signals were simultaneously excited. Note that the locations of the two peaks were close to the peaks of the single source cases (i.e., compare the blue curves to the black and the red curves). These measurement results showed minor differences with the simulation results in Fig. 6b and Fig. 6c in terms of the magnitudes of the transmittances and the locations of the transmittance peaks. Nevertheless, the proposed mutually pulse-width-selective communication system was both numerically and experimentally validated.