I. Introduction
Recently, backscatter communication technology, characterized by ultra-low-power wireless communication, has emerged as a key technology in Internet of Things applications, including radio frequency identification (RFID), tracking devices, remote switches, medical telemetry, and low-cost sensor networks, laying the foundations for the advent of a hyperconnected era [1–6].
In particular, RFID systems can be classified as active, passive, or semi-passive based on the power supply of their tags. In the case of passive RFID, a reader initiates the generation of electromagnetic waves, while the tags receive the waves, modulate them to contain information bits, and then backscatter them to the reader [7]. The key idea here is the wave reflection from the impedance mismatch—a discontinuity implemented by connecting the antenna to the discontinuous impedance of the backscatter tag [1]. This idea has also led to alternative approaches to improving signal-to-noise ratios, such as harmonic transponding-based RFID tags and nonlinear radars [8, 9].
In this context, systems supporting various modulation formats, including binary phase shift keying, orthogonal phase shift keying, and orthogonal amplitude modulation, have been introduced to address the problem of modulation techniques related to low-frequency bands, which are typically limited to data rates of several hundred megabits per second [10]. Studies have also explored the possibility of using backscatter communication in commercial devices, such as universal serial bus (USB) cables and smart speakers [11, 12]. However, all these technologies require distinct antennas for carrying out backscatter communication, resulting in additional real estate requirements [13].
In this paper, an optimal hotspot estimation method for mounting nonlinear devices is presented by conducting an analysis of the excited resistance of PCB ground planes. Additionally, antenna-less backscatter communication is carried out by employing the on–off characteristics of a nonlinear device over a frequency band. The proposed method is distinct from other technologies in that, unlike most commercial systems, it does not require a separate antenna pattern.
II. Analysis of the Nonlinear Device
The proposed antenna-less backscattering system is set up as shown in Fig. 1, where an intentionally applied incident electromagnetic (EM) wave excites the ground (GND) planes of a printed circuit boards (PCB), while the gate of the nonlinear device is connected to a signal line that initiates the on-off signal for backscattering the EM wave. To realize this operation, a detailed analysis of the nonlinear device is performed.
The device used in this work is ATF-54143—a gallium arsenide (GaAs) pseudomorphic high-electron mobility transistor (pHEMT) that switches on when there is a positive bias. This device is considered appropriate for use in this study, since it can be switched on and off by most digital signals that swing between a positive and GND voltage. To measure the necessary parameters for system implementation, the measurement setup is configured as shown in Fig. 2(a).
Furthermore, a bias tee (Aeroflex 8810SMF2-18) is employed to separate the DC bias from the RF signal. The gate bias is swept by 0.05 V from 0 V to 0.6 V, and the drain bias is applied at 0 V, representing the same potential as the source. Subsequently, 2-port S-parameters are measured using a vector network analyzer at frequencies from 10 MHz to above 3 GHz.
The measured S-parameters are converted into Z-parameters Z22, which refers to the impedance between the drain and the source, including parasitic components, such as wire bonding [14]. In general, depending on the bias condition, the magnitude of drain–source impedance of the transistor is large when it is in the off state and very small when it is on.
Fig. 2(b) shows the variations in the Z22 parameter with changes in the gate bias and frequency. A clear distinction between the on/off states, characterized by high/low impedances, is observed at the threshold voltage of 0.3 V. Specifically, the off-state of a typical transistor exhibits high capacitance at DC, resulting in very high impedance, as evident from Fig. 2(b).
III. Hotspot Estimation Method
To locate the hot spot when an external EM wave is applied to the PCB, the characteristics of the nonlinear device along with the ground configuration of the PCB need to be analyzed when the system resonates. The simulation environment for this analysis is constructed using HFSS, as depicted in Fig. 3. A port is present at the top to create plane waves, and a test board is placed at the bottom. Additionally, in this section, a new concept called excited impedance is introduced for hotspot estimation. Excited impedance can be defined as follows:
which indicates the impedance, as expressed by dividing the magnitude of the electric field from the surface current formed on the test board when resonated by an incident EM wave. To apply this concept, two simplified GND structures are assumed to be excited by the incident EM wave.
1. Two Isolated GND Planes
In the case where two GND planes are isolated, each GND plane can be considered a transmission line with an open load, ZL = ∞. Therefore, as shown in Fig. 4(a) and 4(b), in regions without metal where I = 0, the electric field exhibits its maximum value. From this estimation, ZExcited of the GND plane can be easily calculated, as shown in Fig. 4(c). Assuming an open circuit, the location exhibiting maximum resistance is identified—approximately 9.8 kΩ, as denoted by ZExcited,MAX in Fig. 4(c).
2. GND Plane with a Narrow Bridge
The structure of two GND planes connected by a narrow bridge can be considered as two transmission lines with a short load (ZL = 0) in between. Similar to the previous case, as shown in Fig. 4(d) and 4(e), the electric field in the narrow bridge-shaped metal regions is minimized, while the surface current exhibits the maximum value. ZExcited is also calculated, as shown in Fig. 4(f). Assuming a short circuit, the location of the minimum resistance value (ZExcited,MIN) is approximately 0.26 Ω, as shown in Fig. 4(f).
3. Backscatter Communication by Hotspot Estimation
The reflection coefficient between the characteristic impedance of air and the excited impedance can be expressed as follows:
Here, ZAir is 376.6 Ω. Eq. (2) shows that when the circuit resonates, it approaches a situation of ZExcited = ZAir, meaning that the circuit is close to Γ = 0 . However, if the drain–source impedance of the analyzed nonlinear device is mounted at the ZExcited,MIN or ZExcited,MAX points of the GND plane, as denoted in Fig. 4(c) and 4(f), Eq. (2) has to be reformulated as:
Here, Z22 refers to the drain–source impedance of the nonlinear device. Unlike Eq. (2), which relates to a situation with a resonance point, a mismatch may emerge in this case due to the synthesis of ZExcited on the PCB and Z22 of the mounted nonlinear device. Therefore, in the situation described in Section I, the mismatch is maximized when the nonlinear device is in a high state. Meanwhile, in the situation described in Section II, the mismatch is maximized when the nonlinear device is in a low state. As a result, in this work, an amplitude shift keying system (ASK) is implemented based on the difference in return loss that occurs when the system in Fig. 5 is characterized by a resonance point at a specific frequency.
The overall process of the hotspot estimation method proposed in this paper is illustrated in Fig. 6.
IV. Fabrication and Measurement Results
To implement the proposed method, a test PCB of size 55 mm × 92.8 mm, roughly the size of a credit card, is fabricated, as shown in Fig. 7. This PCB pattern is intended to resonate at about 1.7 GHz.
The optimal location for mounting the nonlinear device is determined through excited impedance analysis, as explained in the previous section. Furthermore, clock signals are applied to the gate of the nonlinear device through the signal line on the PCB.
Fig. 8 illustrates the simulation and measurement results of the return loss for the fabricated test PCB. Fig. 8(a) shows a return loss of approximately 54.7 dB at 1.745 GHz when the signal applied to the gate of the mounted nonlinear component is in a low state, revealing resonant points distinct from other positions. In Fig. 8(b), where the signal applied to the gate is in a high state, no noticeable resonant points are observed as compared to Fig. 8(a). Furthermore, Fig. 8(c) represents the difference in the measured return loss, highlighting the maximum contrast obtained at the intended frequency. Therefore, this contrast in the reflection coefficient can be utilized for backscattering at 1.745 GHz.
Measurements in the time domain are performed by applying clock signals to the gate bias of the mounted nonlinear device through the signal line on the PCB. For measurement purposes, an RF signal generator, a TESCOM shield box, a pulse generator, and a digital oscilloscope are used, as shown in Fig. 9.
The results obtained by applying 5 kHz and 10 kHz clock signals to the gate of the transistor are shown in Fig. 10(a) and 10(b), respectively. It is evident that the backscattered signal represents an ASK waveform of a pattern with a constant voltage difference, depending on the applied clock signal.
V. Conclusion
In this paper, the feasibility of antenna-less backscatter communication using GND planes as the antenna was investigated. To implement the proposed model, the drain–source nonlinear impedance over a frequency band was analyzed, after which the excited impedance was calculated to locate the optimal mounting point for a transistor. A test PCB was fabricated for verification, and measurements were performed in the frequency and time domains. The frequency domain measurements confirmed that the maximum reflection loss varied by 29.68 dB at 1.745 GHz when switching the transistor state. Furthermore, time domain measurements showed that the ASK waveform is generated at a carrier frequency of 1.745 GHz.
PDF Links
PubReader
ePub Link
Full text via DOI
Download Citation
Print
