Implementation of an Integrated LoRa and Dual-Band GNSS Antenna in a Compact Package

Article information

J. Electromagn. Eng. Sci. 2024;24(6):632-640

Publication date (electronic) : 2024 November 30

doi : https://doi.org/10.26866/jees.2024.6.r.269

Seongjung Kim, Sangwook Nam

School of Electrical and Computer Engineering, Seoul National University, Seoul, Korea

^*Corresponding Author: Sangwook Nam (e-mail: snam@snu.ac.kr)

Received 2023 November 28; Revised 2024 February 13; Accepted 2024 March 1.

Abstract

I. Introduction

The simultaneous utilization of long-range (LoRa) and global navigation satellite system (GNSS) technologies has emerged as an effective solution to meet the increasing demand for efficient wireless communication and precise positioning systems [1, 2]. While LoRa is renowned for its long-range capabilities and low-power operation, making it well suited for various Internet of Things (IoT) applications, GNSS provides global positioning services through a network of satellites, enabling accurate localization and navigation. Integrating these two technologies offers a promising opportunity to enhance the performance and functionality of wireless communication systems, particularly in scenarios that require reliable connectivity and precise location information.

However, the integration of LoRa and GNSS antennas into size-limited applications, such as tracking devices for animal pests [2], presents significant challenges. In such applications, minimizing the size of the antenna is highly desirable. However, the LoRa band in Korea operates at low frequencies, specifically 920–925 MHz, which necessitates large physical dimensions for antennas to ensure normal operation [3–5]. Similarly, the GNSS band operates in the 1.164–1.189 GHz (L5) and 1.559–1.609 GHz (L1) frequency ranges, thus requiring large physical dimensions for antenna operation. Moreover, GNSS signals transmitted by satellites propagate in accordance with right-handed circular polarization (RHCP), necessitating a large-sized feeding network or resonant structure to receive matched polarization [9–14]. Therefore, integrating LoRa and GNSS antennas into a central control board inevitably results in a bulky physical size.

Various techniques have been researched to reduce this size, such as utilizing meander lines that fold the current path [3, 4, 6–8]. However, the cancelation of opposite current paths along the meander line may degrade radiation efficiency. Furthermore, the use of lumped elements introduces ohmic losses, which pose challenges in achieving precise narrow-band resonance for a desired frequency range (920–925 MHz) [5]. Despite such efforts to incorporate different techniques, the LoRa and GNSS antennas designed thus far remain relatively large.

The practical implementation of antennas requires a control board, a battery, electric cables for connection, and a protective case to ensure the system’s integrity and protect it from external factors. However, these components have adverse effects on the antenna’s operation, including reduced radiation efficiency, altered resonance frequencies, and distorted radiation patterns, among others. Unfortunately, only a few studies have investigated the effects of these components on antenna design [3]. Most antenna designers have focused solely on antenna performance without examining the impact of its electrical connection to the control board and battery. This limitation arises due to the lack of antenna ports for measurement when the antenna is connected to these components, as shown in Fig. 1, which makes it challenging to assess antenna performance unless measured separately. The connection introduces grounding effects due to the presence of large metallic scatterers, resulting in resonance frequency shifts and gain variations. Therefore, a new method that measures the antenna while accounting for these effects is needed.

Fig. 1

Connection of an antenna board to a control board and a battery via electric cables.

In this paper, we propose an integrated LoRa (0.92–0.925 GHz) and GNSS (1.164–1.189 GHz [L5], 1.559–1.609 GHz [L1]) antenna in a compact package. Our first contribution to the literature is the introduction of an ultra-compact LoRa antenna, designed to provide an efficient current path without cancelation and unwanted coupling with the GNSS antenna, resulting in high radiation efficiency. The second contribution is the integration of LoRa and GNSS antennas into a small and thin printed circuit board (PCB). Lastly, we propose a method for measuring the antenna’s performance when it is electrically connected to the control board and battery, reflecting its actual operating condition.

The rest of this paper is organized as follows. Section II describes the operational principles of LoRa and GNSS antennas, including the effects of electric cables, the control board, and the battery. Section III presents and compares the simulated and measured results, including the S-parameters, radiation efficiencies, three-dimensional (3D) average gains, and radiation patterns. We also compare the proposed antenna with other LoRa and GNSS antennas. Finally, Section IV concludes the paper.

II. Design of Antenna Structure and Operation

Fig. 2 presents the proposed integrated antenna in a package consisting of an integrated LoRa and GNSS antenna, a plastic case, an aluminum battery, a control board, and LoRa and GNSS cables. The plastic case has a dielectric constant (ɛ_r) of 2.5 and a loss tangent (tanδ) of 0.01. The control board is made of FR-4 material with an ɛ_r of 4.3 and tanδ of 0.025, while the antenna board is composed of Taconic RF-43 material with an ɛ_r of 4.3 and tanδ of 0.0033, respectively. The thickness of the antenna board is 0.6 mm. The dimensions of all the components are listed in Table 1. The battery is connected to the control board through a direct current (DC) line, while the control board is connected to the antenna board via LoRa and GNSS cables to ports G1 and G2 (U.FL to U.FL cables). Thus, both the control and antenna boards are grounded in the aluminum battery. Although this electrical connection is bound to affect the performance of the antenna, it nonetheless represents the most practical setup for providing DC enengy from the battery. To measure the LoRa and GNSS antenna performance, two spare U.FL connectors (ports 1, 2) are connected to a vector network analyzer (VNA) using U.FL to subminiature version A (SMA) cables. In the examination, ports G₁ and G₂ were not connected to the signal lines for excitation, as shown in Fig. 3. Instead, they were left open and grounded to the bottom layer using vias. Furthermore, for the simulation excitation of the antenna, discrete ports with 50 Ω impedance are employed, as shown in Fig. 3(a).

Fig. 2

Proposed antenna configuration: (a) the antenna enclosed in a plastic case, (b) without the case, (c) without the case, battery, and control board, (d) connected to the measurement equipment, and (e) inner view of the case.

Table 1

Dimensions of the proposed antenna structure (unit: mm)

Fig. 3

(a) The antenna board, (b) fabricated top layer and (c) bottom layer of the board.

The LoRa antenna (port 1) is excited at an offset location from the middle of A_y, primarily due to the antenna’s input impedance, which depends on the feeding location. As shown in Fig. 4, at 0.92 GHz, as the location moves closer to the middle, the input impedance decreases. Conversely, as the location moves closer to the open end following the slot line, the input impedance increases due to the stronger electric field at the end. Furthermore, the slot line along the board’s edge is open-ended at both sides and has a half-wavelength at 0.92 GHz. This configuration allows for the achievement of a compact antenna with high radiation efficiency, as it prevents the cancelation of y-polarized electric fields due to the inverse phase and bending along the slot.

Fig. 4

Electric field distribution at 0.92 GHz on the antenna board when port 1 (LoRa) is excited. The open-ended half-wavelength slot of the antenna is individually illustrated on the right.

The GNSS antenna (port 2) in Fig. 3(a) is excited at an offset location at the right open end for a similar reason as the LoRa antenna, although its input impedance decreases as the feeding line inches closer to the open end, since the feeding line is capacitively coupled to the antenna without a via. Notably, the GNSS antenna configuration is drawn from [15] and [16]. As shown in Fig. 5, this configuration generates linear polarization, which eliminates the need for bulky structures for RHCP, albeit with a maximum polarization loss of 2.3 dB. As shown in the slots in Figs. 4 and 5, the polarization of the antenna operating at the LoRa and GNSS bands is linear in the y-direction. The length of the top slot line, gl₁ + gl, corresponds to a quarter-wavelength at 1.17 GHz (L5). Similarly, the bottom slot line has a quarter wavelength at 1.58 GHz (L1), as shown in Fig. 5, with the electric field being dominantly strong at the open end along the slot. At L5, a +y-directed electric field is primarily made at the top slot, while the two slots are simultaneously excited at L1. While a–y-directed electric field is formed at the top slot at L1, a +y-directed electric field is dominantly generated at the right-side gap. Thus, at the GNSS bands, the proposed antenna operates with y-directed polarization. In fact, the bottom slot line shares the feeding line and a partial slot line with a length of gl. The gap, g, controls the resonance frequencies of the GNSS antenna, with a smaller value of g, resulting in a longer effective length of the slot line and lower resonance frequencies. Therefore, the GNSS antenna enables the realization of an ultra-compact antenna within the constraints of an extremely limited PCB size.

Fig. 5

Electric field distribution on the antenna board when port 2 (GNSS) is excited at (a) 1.17 GHz (L5) and (b) 1.58 GHz (L1).

To obtain high radiation efficiency, it is crucial to minimize mutual coupling between the LoRa and GNSS antennas, because coupling results in the dissipation of energy instead of radiation. In Fig. 6, a significant amount of mutual coupling (blue dashed line) is observed around 1.58 GHz (L1). This coupling, according to the reciprocity theorem, can degrade the receiving gain of the GNSS antenna. However, this mutual coupling can be reduced by implementing a shorting line (SL) of width t, as depicted in Fig. 3(a). Therefore, when port 2 is excited at 1.58 GHz, a strong induced current distribution is formed on the antenna board (left image in Fig. 7), reaching the feeding line of port 1. However, when the slot in front of the feeding line is blocked by an SL (right image in Fig. 7), a closed loop that bypasses the current path is formed, effectively reducing mutual coupling, as seen in Fig. 6. At the same time, when port 1 is excited at 0.92 GHz, the LoRa antenna operates without any disturbance. In this context, it is worth noting that if the feeding line was positioned below the SL in the opposite y-direction, it would be more vulnerable to coupling. In fact, the field distribution at 0.92 GHz, as shown in right image of Fig. 4, corresponds to the odd mode since the polarization directions on both sides are opposite to the middle of the slot. In contrast, as observed in Fig. 6, the proposed configuration comprising an SL generates an additional even mode at 0.83 GHz, maintaining the same polarization direction on both sides. This is because the slot line with SL, which is excited by the structurally asymmetric feeding position, separates the electric field distribution into even and odd modes.

Fig. 6

S-parameters with and without the SL. S-parameters of the two cases overlap at around 0.92 GHz (odd mode).

Fig. 7

Surface current distribution at 1.58 GHz without and with SL when port 2 (GNSS) is excited.

A parametric study was conducted on the key factors of the antenna board, as shown in Fig. 8. Factors such as t, l, and f₂ were found to be related to S₁₁, while the others were related to S₂₂. The width of the SL, referred to as t, determines the ratio of the even and odd modes—a thicker width strengthens the even mode and weakens the odd mode. Therefore, considering that the SL itself effectively reduces mutual coupling, it is advisable not to choose a thick SL. For instance, at t = 4 mm, the even mode was found to be dominant at 0.89 GHz, while the odd mode became weaker at 0.93 GHz. Thus, for this study, we chose t = 0.2 mm to generate a strong odd mode. Furthermore, the slot line, l, influences the resonance frequency of LoRa antennas, with longer lines resulting in lower resonance frequencies. Therefore, to design a compact antenna that can be used at a low frequency band, we chose l = 41.4 mm as the longest possible slot length. As discussed earlier, the feeding position or line length, f₂, determines the antenna’s input impedance. Notably, a length of f₂ = 9.3 mm was found to correspond to an impedance of approximately 50 Ω. Deviating from this length would lead to an impedance mismatch, as shown in Fig. 8. Similarly, the line length of the GNSS antenna, f₆, determines the antenna’s input impedances at the two bands. Thus, an appropriate length had to be selected to simultaneously obtain a value close to 50 Ω (in this case, f₆ = 8.1 mm). Furthermore, the top slot length, gl₁, determines the two resonance frequencies of the GNSS antenna, with longer lengths resulting in lower resonance frequencies for both bands. Therefore, we set the lower band at L5, regardless of the higher band. At the same time, since the bottom slot length, gl₂, only affects the higher resonance frequency, it was set at L1 in sequence.

Fig. 8

Parametric study on key factors of the antenna board. (a)–(c) t, l, f₂ are related to S₁₁, (d)–(f) while the others are related to S₂₂. Notice that the abscissa ranges are different from each other.

With regard to the packaging effects on the antenna, it is important to note that the electrical connection between the antenna board, battery, and control board significantly impacts antenna performance, as verified in Fig. 9. Notably, connection refers to the GNSS and LoRa cables in Fig. 2(e) that are connected to the antenna board (G₁ and G₂ in Fig. 3(b)). The S-parameters and 3D average co-polarization gains exhibited notable variations depending on the connection. The co-polarization results indicated y-principled linear polarization (LP) for the LoRa antenna and RHCP for the GNSS antenna. The electrical connection between the antenna board, battery, and control board made them a single radiating antenna, resulting in changes in the antenna’s input impedance and radiation pattern. Therefore, it is crucial to have prior knowledge of these effects before proceeding with fabrication and measurement. According to the method proposed in this study, this knowledge can be acquired by connecting the dummy ports G₁ and G₂ to the control board connectors, as shown in Fig. 2(e). This connection reflects the grounding effect. Furthermore, the antenna can be excited and measured at the same time via VNA cables (see Fig. 2(e)). Moreover, the plastic case (YYT-35-55-20 made in Shenzhen Yong Ye Tai Electronics Co. Ltd.; ɛ_r = 2.5, tanδ = 0.01) could also reduce the radiation efficiency by 20%–30% and down-shift the resonance frequencies.

Fig. 9

Case study on the connection between the antenna board and the control board via the GNSS and LoRa cables: (a) S-parameters, and (b) 3D average co-polarization gains when either port 1 or port 2 is excited.

In summary, the antenna needs to be measured by accounting for the connection (G₁ and G₂) made by feeding ports 1 and 2 to verify antenna operation. Therefore, G₁ and G₂ have to be sequentially removed, and ports 1 and 2 need to be connected to the control board instead of G₁ and G₂, which represent the end product. Fig. 10 illustrates the measurement setup for estimating the radiation pattern of the proposed antenna. One of the ports (LoRa or GNSS) is connected to the VNA through the cable, while the other is terminated by 50 Ω. To avoid the influence of the measurement cable, the antenna was positioned such that its polarization (y-pol.) is approximately vertical to the cable (z-pol.).

Fig. 10

Measurement setup for estimating the radiation pattern of the proposed antenna.

III. Simulated and Experimental Results of the Proposed Antenna

In this section, we present the simulated and experimental results of the proposed antenna design. Fig. 11 presents the verification results for the S-parameter, radiation efficiency, and 3D average co-polarization gain. After conducting fine-tuning trials, the simulated and measured resonance frequencies were found to be in exact agreement. Notably, the measured impedance bandwidths (IBWs) were consistently wider than the simulated IBWs despite a slight increase in mutual coupling. The measured IBW under −10 dB at port 1 covered the LoRa band (0.92–0.925 GHz) at a range of 0.89–0.934 GHz. At port 2, the measured IBW under −10 dB was 1.15–1.17 GHz and 1.58–1.6 GHz. Under −6 dB, it extended to 1.139–1.191 GHz and 1.558–1.615 GHz, covering the GNSS bands (1.164–1.189 GHz, 1.559–1.609 GHz). Furthermore, the measured mutual coupling reached a maximum of −12.5 dB. A second parasitic resonance was observed for the measured S₁₁, which occurred due to bending of the LoRa cable, which was the shortest prototype cable. As a result, the measured radiation efficiencies and 3D average gain exhibited some differences compared to the simulated results, owing to the uncertainty in the material information of the plastic case and the inability to accurately model the exact control board. The measured efficiency at the LoRa band was 64.9%, while the simulated value was 44.6%. At the low GNSS band, the measured efficiency was 46.3%, compared to the simulated 69.7%. Similarly, at the high GNSS band, the measured efficiency was 54.2%, while the simulated efficiency was 66.2%. Furthermore, the measured 3D average copolarization gains at the LoRa, low GNSS, and high GNSS bands were −5 dBi, −6.48 dBi, and −7.88 dBi, respectively, while the peak co-polarization gains of the proposed antenna were −3.52 dBi, −3.66 dBi, and −3.36 dBi, respectively. In this context, it should be noted that for antennas mounted on randomly moving platforms, such as animals, a non-directive radiation pattern is preferable over a directive pattern. In fact, RHCP gains at the low and high GNSS bands were lower than the y-principled LP gains by 2.3 dB and 1.8 dB, respectively. This indicates that the antenna’s polarization was not purely linear due to the orthogonally induced current on the aluminum battery. As a result, the antenna exhibited RHCP with a high axial ratio, and an ultra-compact RHCP GNSS antenna was obtained with only 2.3 dB and 1.8 dB degradation.

Fig. 11

Simulated and measured (a) S-parameters, (b) radiation efficiencies, and (c) 3D average co-polarization gains.

Fig. 12 presents the normalized radiation patterns in the three principal axes. The simulated and measured radiation patterns exhibit good agreement. Since the antenna board was linearly polarized in the y-direction and the induced current on the battery was linearly polarized in the x-direction, it is observed that the cross-polarization level is high at the angles where the co-polarization level is relatively low.

Fig. 12

Normalized radiation patterns in three different planes when ports 1 and 2 are excited. Co-polarization is represented by red color, cross-polarization is shown using blue color, solid lines indicate simulated gain, and dashed lines trace the measured gain.

Finally, Table 2 compares the performance of the proposed antenna with that of the existing LoRa and GNSS antennas. Among the works listed, only one study [3] designed an antenna that accounted for packaging effects. However, the approach adopted posed a significant handicap in terms of achieving a compact size, high radiation efficiency, and gain. As evident from Table 2, the size of the proposed antenna is considerably smaller than the others, even when considering its packaging and the integration of the LoRa and GNSS antennas. The IBW listed in the table refers to the range within which the reflection coefficient was found to be below −10 dB. It is worth noting that many GNSS antennas have been designed with wideband IBW rather than dual-band IBWs. In contrast, the proposed antenna focuses specifically on the essential LoRa and GNSS bands. Additionally, the radiation efficiencies of the proposed antenna are relatively high despite the presence of the lossy plastic case. Unlike most previous works that designed directive antennas with high peak gains, the ultra-compact nature of the proposed antenna results in non-directive radiation patterns and relatively low gains, with an approximate −3 dB degradation caused by ohmic losses. For further comparison, the performance of the antenna without packaging is also presented (including only the antenna board) in the table. Since ohmic losses and scattering are removed in this case, the radiation efficiency and the peak gain are significantly improved, and the size of the antenna is smaller.

Table 2

Performance comparison of LoRa and GNSS antennas

IV. Conclusion

In this paper, we present a design for an integrated LoRa and GNSS antenna specifically tailored for size-limited applications. The antenna was successfully integrated into a small and thin PCB, thus addressing the challenges associated with limited space. Furthermore, we carefully considered the effects of the electrical connection between the control board, battery, and electric cables on the antenna’s performance. It was observed that these factors resulted in shifts in resonance frequencies and variations in antenna gain. To overcome these challenges, we propose a novel measurement method that allows for the examination of antenna performance when it is electrically connected to the control board and battery via electric cables. This enabled us to assess the antenna’s behavior considering actual operating conditions in which DC power was supplied by the battery, thus accounting for the grounding effects caused by the connection via cables. Consequently, we were able to adjust the operating resonance frequencies by accounting for these effects. Moreover, we achieved high radiation efficiency by employing a novel configuration that prevents the cancelation of electric fields. Furthermore, it is demonstrated that an SL can significantly reduce unwanted mutual coupling between ports, resulting in improved radiation efficiency. Through simulation and experimental validation, we confirmed the performance of the proposed antenna in terms of its IBW, radiation efficiency, and radiation pattern.

The compact and integrated nature of the proposed antenna makes it highly suitable for IoT applications that require reliable wireless communication and precise positioning. By addressing the challenges of size limitations and optimizing the antenna’s performance, the proposed design offers a promising solution for such applications.

Article information Continued