RSSI Improved for LoRa Wireless Communication, Field-Tested in the Wide-Open Area

This paper presents an effective method to boost the RSSI, the received signal strength on a moving path of LoRa wireless networking without change in digital circuitry or power amplifiers. Replacing the wire antenna as the conventionally used sensor by a printed Yagi-Uda antenna as a novel one designed to have a higher gain signifies increments of the RSSI at distant positions in the LoRa system with the same battery. The electromagnetic properties of the novel antenna are given by observing its beam-patterns in the anechoic chamber following the design procedures and fabrication. And the field-test for the radio link is conducted to see the proposed antenna enhance the RSSI at positions along the street of a wide-open area for LoRa communication. The RSSI has risen by more than 10 dB at distances from 200m to 700m with DLP-RFS1280 as the LoRa device in comparison to the conventional configuration in the device.


I. INTRODUCTION
AS the era of Internet of Things (IoT) has begun, a plenty of applications and use-cases have been suggested to lead and comply with demands from mobile communication users and service providers. Accordingly, the industrial electronics for the mobile services has checked viability on what people need and adequacy of the legacy-and current wireless technologies from the view point of realizing the demands. This has been giving rise to various mobile technological frames and standards having variations related to frequency-band, bandwidth, modulation schemes, speed, data, protocol, hops, power, coverage and so on. Forming a radio network over cells in a large area for high data transmission rates aimed at wireless telephones, GSM, DCS, WCDMA, LTE, NR and NTN have been researched from generation to generation. Aside from mobile telephony, WLAN and WiFi are used to link laptops with the wireless network, often associated with Bluetooth or Zigbee which are said to work for formerly The associate editor coordinating the review of this manuscript and approving it for publication was Ravi Kumar Gangwar . M2M or D2D triggering RFID and NFC and presently IoT. Part of the well-established radio network, there is no difficulty in enabling BLE data and Zigbee data to travel far, but this is made possible by cellular and internet networks. The IoT includes sensor networks being built with BLE or Zigbee modules, which end up with limited distances. Long-Range(LoRa) networking technology has become a highlight on the stage in that its TX and RX devices communicate over hundreds of meters [1], [2], [3], [4]. This is attractive to those who hope to build sensor networks over a very wide area without setting mobile carriers' base-station antennas, and has expanded applicability to localization and electricitymetering. LoRa has its own gateways and nodes interacting with sensors distributed over real-estates like streets, campuses and fields.
Searching and checking articles and reports on LoRa, the followings were selected and the impressions of those representative projects are mentioned. While some of them address calculation-based prediction on the wireless communication, others verify the data from simulations by doing experiments gathering the data of the RX power in relatively small areas or large areas. Lim et al calculated spreading factors(SF) allocated to the traffic under unslotted ALOHA random access protocol to maximize massive connectivity in LoRa systems [5]. Using the stochastic geometry model, sub-optimal SFs could be found. Premsankar et al used the formulation of integer linear programming to get scalability making sure the LoRa networking reliable [6]. They saw their solution leading to a higher delivery ratio. Liu et al formulated the coverage probability which relates the signal transmission with power consumption and its adjustment [7]. Fine-grained information of the LoRaWAN was obtained based on the metadistribution of signal-to-noise ratio. Magrin et al presented the probability function vs. interference as performance metrics of a central-controller interacting with multiple LoRa links from industrial scenarios. [8]. Their configuration properly set in the IIoT brought them a high packet success rate. All the articles above are theoretical.
The following literatures conducted experiments in the building or outdoors showing convincing ideas over those of communication or network theories. Savazzi et al did channel characterization measurements on a floor of a building using LoRa devices and adopted Wiener-Hopf-based algorithm to correct errors in positioning [9]. Similarly, Xu et al took floors of a building as an test-case so as to observe the functions of the LoRa system adopted into room-to-room measurements over the whole floor [10]. They got probability functions hand-in-hand with the data. Aarif et al having the LoRa-based localization in mind checked the RSSI on the LOS path in indoor propagation environments with different LoRa configurations [11]. Oh et al suggested a method of localization by combining the LoRa technique with the UWB system to get over the shortcomings in a complicated indoor IoT space [12]. Goldoni et al used the measured RSSI to range the positions of the receiving module indoors and outdoors along with the channel characterization [13]. Kwasme et al derived a linear matrix resulting from the RSSI data of three base-stations realized by the LoRa modules for better localization [14]. Anjum et al introduced the machine learning technique to make the indoors and outdoors localization less erroneous [15]. Vazquez-Rodas et al took a ranging method similar to [14] setting up simultaneous equations [16]. The measurement was done in a farm and a soccer field. Torres et al tried sites greater than the cases above and plotted the RSSI at long distances [17]. Going over the latest relevant literature, the RSSI is very low and limited, since their wireless systems comprise the generally used elements including monopoles or dipoles [18], [19].
In this paper, a LoRa communication instrument is realized for improving the received power at several hundred meters in distance. The antenna as the sensor for the transmitting device of the LoRa system goes through a change from a wire antenna as the conventional one to a printed Yagi-Uda antenna mountable on the module. The proposed RF sensor is designed to radiate the RF power of a higher gain than that of the commercial antenna. Resonating at 2.45 GHz, the end-fire antenna outperforms the conventional one by nearly 9 dB when the far-field patterns are measured at the anechoic chamber. This enables the RSSI to be boosted for the proposed antenna-combined LoRa module by over 10 dB at far-distances on the path in the real experiments. This is a very effective way to the radio link more reliable even with the use of the same battery power. The street of the town is taken and the position of the TX device from the RX is varied from 200m to 700m as the far range after standing near the RX to 200m as the neighboring range for LoRa link instrumentation.

II. ELEMENTS FOR LORA NETWORKING DEVICES AND ANTENNAS AS THE RF SENSOR
The LoRa communication instrument is built with TX and RX devices each of which consists of the chipset drivin the wireless module, an antenna as the RF sensor and the battery. The RX is connected with a notebook computer that gives the command to the module and takes and stores the data generated by the chipset. Fig. 1 shows the TX and RX modules before and after being put in a LoRa test site. DLP-RFS1280 is adopted as the LoRa wireless module featured with supply voltage, frequency and output power of 1.8-3.7V, 2.4 GHz and 11.5 dBm maximum, respectively. In our own preliminary outdoor tests, the data revealing that this low-power consumption system delivers the 3RF power of -82 dBm at 200m and -90 dBm at 250m would make users doubt about credentials and usefulness of this wireless system as the signal becoming extremely week at the distance over 300m. To push up the RF power at the far-range without adding a power-amplifier or battery cells to the module, a cost-effective and technologically advanced way is to change the wire antenna to a new one. It should not be big but must overcome the drawbacks of the omni-directional antenna [20].
To make the beam pattern of the RF sensor oriented toward the receiver side, an end-fire antenna is adopted. It is smaller than an array antenna and simpler in RF-signal feeding. Unlike the original Yagi-Uda antenna which is supported and held in the air by a chain of plastic fixtures, a compact version is designed. As a PCB structure, Fig.2(a) is the initial antenna having metal elements for horizontal strips and T-shaped strips fed by a hook coupled with a slot on the FR-4 substrate. As in the side-view, the initial structure is covered with the top-and bottom parasitic layers(with bolt holes) as the improved structure, which concentrates the field-flux much more as in Fig. 2(b). As a result, the field is well guided to the end of the novel structure and leads to a high antenna-gain increased by about 9 dB as in Fig. 2(c). The design and simulation are done as in Appendix that the physical dimensions in the tables are decided by the parametric study meeting the operation at 2.4 GHz.
The initial and improved structures as the proposed RF sensors were manufactured as planar ones with the physical dimensions given in Appendix. The prototypes are shown in Fig. 3(a). The top and bottom layers are attached to the initial antenna through bolts instead of glue, which is practical. They are placed together with the wire antenna at the anechoic chamber as in Fig. 3(b). S 11 -curves in Fig. 3(c) present the antennas from the two steps in the design have good impedance match at 2.4 GHz. Fig. 3(d) shows the proposed antenna has the gain higher than the commercial antenna by around 9 dB. The difference of the measured S11 from the simulated S 11 is a frequency shift resulting from errors in dielectric constants and air-gap from layers. Besides, that the width and length of the fabricated geometry differ from those of the simulated geometry causes the discrepancy between the curves of the s-parameter. Also, the beam-patterns change to a degree due to the testing harness. The lengths of the cables for three antennas under test and how they are laid on the jig account for the difference between the simulated and measured beam-patterns.

III. FIELD-TEST EXAMINING IMPROVEMENT IN RSSI
The antenna developed and electromagnetically characterized as in the prior section is put to use in the low-power consumption long-range wireless networking. Here comes the scheme of the performance test with explanation.
The street of a sea-side town is selected as a wideopen area. The RX device using the conventional antenna is stationary and the location of the TX module is moved away from the RX side as in Fig.4(a) and Fig. 4(b). R as the distance increases from the vicinity of the RX through 200m to 700m. The entire range is divided into the neighboring range and the far range at first. The RF sensor of the TX device is connected with the wire antenna at first and then changed to the proposed antenna for each distance. The RSSI values are obtained along the positions in the two cases. Fig.4(d) shows the RSSI vs. the distance, which reveals the proposed antenna results in the RSSI 5.5 dB∼12 dB higher than that of the conventional one. Taking the TX device in the far range from 400m to 700m renders the experiment VOLUME 11, 2023 improved RSSI by the proposed RF sensor as in Fig.4(e). Covering the spot near the RX up to 700m all together, the RSSI-curve by the proposed antenna combined LoRa device is obviously greater than that by the wire antenna. This is validated by Fig. 4(f) that compares the curves and refers to the mathematical estimation of path loss which goes with the following equation that P and G mean power and gain each. This paper suggests an effective and advantageous method to build a LoRa instrument with significantly improved RSSI in the long range of radio communication. Limited in the battery power, without taking costly means like additional power-amplifiers or revamping digital circuitry, the RF sensor as the passive component in the LoRa wireless module has been changed from the wire antenna to the PCB Yagi-Uda antenna that turns out to provide the developer with improvement in the RSSI at long distances. Especially, the Yagi-Uda antenna as an end-fire structure is devised to focus the field-flux and guide the dense field toward the RX wireless module at the resonance frequency. The proposed antenna radiates the far-field wave with the gain approximately 9 dB higher than the conventional one. To deal with the stronger wireless networking, this novel sensor is connected to the LoRa module and taken to the test site to validate the method. In the field test, compared to the commercial antenna and cases reported in others' papers, the RSSI from the LoRa instrument is greater than the values at distances above two hundred meters by 12 dB(minimum)∼39 dB(maximum).

APPENDIX
The RF sensor, that is to say, the antenna is desired to be stronger at transmitting and receiving the RF signal differing from the omni-directional beam of the wire antenna. The endfire beam from the TX device to the RX module is proper. The Yagi-Uda is chosen due to its directional far-field and single-point feeding contrary to multi-point feeding of array antennas [21]. However, the Yagi-Uda antenna is generally is held with plastic fixtures and has a lukewarm gain. To solve the problems, the Yagi-Uda antenna should be realized as a PCB structure and made to concentrate the electromagnetic field flux to the direction of the radio link. This is explained by the following figure and mathematical expression.  The excitation element radiates the field and makes the adjacent elements generate the fields which interfere with one another at the far-zone. The interference is depicted with phase differences as follows.
This numerical test reveals the proper values of the spacing as well as the number of elements. Five elements are needed with the constraint on the size of the antenna. The initial value of the spacing is 0.22λ and modified in the full-wave EM simulation step. Meeting these requirements, the design of  the proposed antenna goes through two steps. The first step as the initial structure makes a planar Yagi-Uda antenna, and resonate at 2.4 GHz and increase in gain. The second step as the improved version can raise the antenna gain further by adding top-and bottom layers to the initial structure at the center the coupling of which concentrates the electromagnetic field from the feed point guided in the direction to the RX. These functions are achieved by conducting the parametric study on the geometries in Fig.1. Fig.7(a) to Fig.7(e) deal with S11 input reflection coefficient and the far-field patterns of the initially proposed structure as the values of the key parameters of the geometry are varied. On the other hand, Fig.7(f) to Fig.7(j) are the curves of S11 and the beam-patterns as the physical dimensions of the improved structure change. Before the initial structure is not given the top and bottom parasitic layers, the beam pattern seems smooth as shown in Fig.7(a) to Fig.7(e). g1 is varied from 5.13 mm to 15.57 mm, and 9 mm is good for the impedance match at 2.4 GHz as in Fig. 7(a). Fig.7(b) and Fig.7(d) present rapid changes in the S11-curve which means Lp1 and Lp2 as the lengths of the parasitic elements must be carefully chosen. When they are 30 mm, the resonance occurs at the target frequency. Five elements are proper according to Fig.7(e) and seven elements are not beneficial because of making the structure bigger. The initial structure is modified to have top and bottom layers to bring it an improved radiated field. The physical dimensions of the initial structure located at the center of the stacked geometry as in Fig.1(a) are kept the same as Fig.7(a)∼Fig.7(e), and what is left is to decide the physical dimensions of the top layer assuming that the top and bottom layers are geometrically identical. Among Fig.7(f) through Fig.7(j), S11 of Fig.7(g) and Fig.7(i) seem to drastically change as their parameters vary, while the far-field pattern from Fig.7(f) to Fig.7(j) becomes apparently directional. Varying g01 as the gap between the lower edge and the first parasitic element on the top layer from 5.13 mm to 15.57 mm, Fig.7(f) suggests 9 mm. Fig.7(h) proposes 22 mm as the gap from the lower edge to the second parasitic of the top layer. In Fig.7(g),25 mm is right for Lp1 as the length of the first parasitic of the top layer, which makes the antenna resonating at the operating frequency, say, 2.4 GHz. Lp2 as the length of the second parasitic of the top layer as in Fig.7(i) makes S11 below -10 dB which means the impedance match and generates the resonance when it is 25 mm. Fig.7(j) says that there is no big difference between 3, 5 and 7 elements. Considering a compact structure, 5 elements are taken. From the parametric study by checking the functions optimized as Fig. 7(k), the physical dimensions are determined as follows.
The antenna implemented with the geometrical parameters given in Table 1 is adopted to the LoRa radio system and it enables the receiver device to have largely increased RSSI levels greater than what others obtained. The comparison is made as Table 2. Observing the RSSI levels recorded in the table, it is discovered that the proposed instrument has −69 dBm and others have −82.5 dBm, −95 dBm and −108 dBm for the distance of 200m. While some experiments did not detect the RSSI over 500m, the received power of the proposed LoRa system becomes effective as remarkable enhancement over other technical experiments for the longer distances.
Extra tests are conducted to see the reciprocity exists, the two antennas being exchanged from the TX to RX modules. Case 1 has the wire antenna for the TX and proposed antenna for the RX. The antennas are switched between the two sides in case 2 as in Fig. 8(a). The RSSI values (nearly -60 dBm) measured from the distance of 50m to 100m for case 1 exactly agree with those of case 2 as in Fig. 8(b).