Electromagnetic Interference With the Mobile Communication Devices in Unmanned Aerial Vehicles and Its Countermeasures

One of the threats to industrial unmanned aerial vehicles (UAVs), which require a high level of safety, is the degradation of control wireless communication performance due to electromagnetic interference. Self-jamming caused by unintentional electromagnetic (EM) noises from the multiple electronic devices installed in a UAV, such as the power supply and camera, is a fundamental issue independent of the surrounding environment. UAVs are susceptible to self-jamming since electronic devices as noise sources are densely mounted inside their chassis. Therefore, evaluating the self-jamming problem and taking countermeasures to achieve high safety in wireless communication performance are important. In this paper, we focus on mobile communications used for UAV operation and evaluate the impact of EM noise generated by a commercial industrial UAV on the receiver sensitivity of an onboard mobile communication device. It was found that EM noise generated by the UAV degrades the sensitivity in the three communication bands, 800 MHz, 1800 MHz, and 2100 MHz, used over the air. In particular, receiver sensitivity was degraded by up to 29 dB in the 800 MHz band. These results indicate that the self-jamming problem is not negligible for the safe operation of industrial UAVs. In addition, countermeasures for self-jamming were taken at the mechanical design level to suppress EM noise to a level that does not degrade mobile communication performance. This paper clarified that countermeasures for self-jamming are an important element for safe UAV operation.


I. INTRODUCTION
The use of unmanned aerial vehicles (UAVs) has expanded over the last decade owing to their high mobility and low cost.UAVs were initially developed for military applications, The associate editor coordinating the review of this manuscript and approving it for publication was Valerio De Santis .
but have been used for recreational purposes and, more recently, for various industrial applications [1], [2], [3].Applications such as aerial video recording, pesticide spraying, and infrastructure inspections have become widespread.In addition, UAV applications are expected to expand to, for example, logistics, searching for people in distress, and surveillance [4].In these applications, UAVs need to be operated over a wide area.UAV operations using mobile communication systems such as long-term evolution (LTE) and fifth generation (5G) are being developed [5], [6], [7], [8], [9] since typical Wi-Fi-based UAV control limits the operating area to 1-3 km [10].
UAVs flying over densely populated areas with out-ofsight maneuvers must have reliable control communications, as control problems can lead to accidents involving human lives [10].A major factor that degrades the communication performance of mobile communications for UAV control is electromagnetic interference (EMI).
EMI is generally caused by unintentional electromagnetic (EM) noise generated by internal devices of the UAV and intentional or unintentional EM noise from other electronic devices [11].EMI due to unintentional EM noise generated by internal electronic devices, the so-called self-jamming or intrasystem EMI, limits the wireless communication performance of UAVs and is a fundamental issue that occurs independent of the surrounding environment.Self-jamming occurs when EM noise is mixed with wireless communication signals through the wireless receiver antenna, and the degree of degradation depends on the frequency, power, and waveform of the EM noise.
In previous studies, numerical analysis of the EMI due to EM noise in orthogonal frequency division multiplex (OFDM) systems, including LTE and 5G, has been conducted [12], [13], [14], [15], [16].Almost all electronic devices generate EM noise in response to their operation.The EM noise occurs in the wide frequency range used by mobile communication systems.
Moreover, various devices are densely mounted inside the chassis of an industrial UAV, which emanates complex EM noise.Therefore, the mobile communication device installed in the UAV may be affected by self-jamming, as shown in Fig. 1.To ensure the safe operation of industrial UAVs, it is necessary to analyze and establish countermeasures for self-jamming problems.Hereafter, self-jamming is simply referred to as EMI since this study only focuses on selfjamming.
In this paper, we present practical examples of EM noise evaluation, EMI analysis for mobile communications, and countermeasures for industrial UAVs, assuming that UAVs can be operated seamlessly even in areas with weak communication signals from base stations.We focus on mobile communication systems' downlink (DL) performance with the configuration shown in Table 1.We evaluated the most basic system using quadrature phase shift keying (QPSK) modulation because, in our evaluation, we intended to ensure that the wireless connection to a base station is always uninterrupted during UAV operations.
For the uplink (UL), the maximum transmission power of the user equipment (UE) is standardized to be +23 dBm [17], which is much higher than the EM noise generated by internal circuits.Hence, EMI due to EM noise is not considered in UL performance analysis.Another advantage of industrial UAVs using mobile communications is the high-speed transmission of high-quality video data they have captured [18].Although ensuring high-quality video transmission is important, it is unlikely that EM noise will interfere with video transmission owing to the abovementioned advantage.However, if EMI causes a packet error in the DL, the connection to the base station will be lost, and large-volume data transmission of UL will also be interrupted.Therefore, the evaluation and countermeasures for EMI problems in this study will improve the reliability of UAV control communications and indirectly improve the transmission performance of sensing data such as video.
This paper is organized as follows.In Section II, we present the evaluation of mobile communication interference by EM noise using an actual LTE-receiving device.In Section III, the noise sources are shown by the detailed analysis of EM noise.The EMI problems can be quantitatively evaluated by measurements and wireless-system-level simulations.In Section IV, we present the countermeasures for EMI problems inside UAVs and their effectiveness.The paper is concluded in Section V.

II. IMPACT OF EMI ON INDUSTRIAL UAVs
In Section II, we show experimental results of the impact of EM noise generated by the test UAV on the performance of an actual LTE-receiving device.
The test UAV was an industrial product already in practical use with a large chassis, a width of 1300 mm, a depth of 1400 mm, a height of 900 mm, and a weight of 4.7 kg.The four propellers were removed to ensure safety during the following experiments since the initial evaluation confirmed that the EM noise generated by the motors does not affect mobile communications.This is because the rotational frequency of the motor is 10-100 kHz, which is sufficiently lower than the frequency range used by mobile communication systems.In addition, the test UAV was powered by a series regulator-type power supply unit to stabilize the UAV operation, although it is normally powered by a battery.We evaluated the impact of EMI caused by EM noise generated by internal integrated circuit (IC) chips on the DL performance when the receiving device of mobile communication systems was mounted on this UAV.
The LTE-receiving device is separated into two parts, the receiving antenna and the circuitry board with a radio frequency (RF) front-end as shown in Fig. 2. The circuitry board is a commercial product, and its performance meets the third-generation partnership project (3GPP) standards [19], allowing simulated evaluation of actual EMI problems.A metal shielding is also mounted on the internal circuitry board of the receiving device to prevent EM noise from affecting the EMI evaluation [20].
For the evaluation using the actual LTE-receiving device, measurement instruments, the test UAV, and the horn antenna emulating the base station antenna, they were placed inside an anechoic chamber to shield the external environmental noise, and a wireless connection was made with an LTE base station tester outside the chamber, as shown in Fig. 3.The power supply unit connected to the test UAV is a series regulator type.Initial evaluations have confirmed that the power supply unit we employed does not emit EM noise in the mobile communication bands.The frequency bands evaluated were the 800 MHz (band 26), 1.7 GHz (band 3), and 2.1 GHz (band 1) bands currently allocated for over-the-air mobile communications use in Japan [21], [22], whose signal bandwidths are 10 MHz, 20 MHz, and 20 MHz, respectively.In particular, the 800 MHz band is expected to be used for UAV control because the signal can reach a wider area owing to the radio wave diffraction effect.
First, the receiver sensitivities of the LTE-receiving device were evaluated without the test UAV.The receiver sensitivities were determined from block error rates (BLERs), which is the percentage of data blocks that could not be received correctly out of the total number of received data blocks.The BLERs were evaluated on the basis of the link quality transmitted from the LTE-receiving device to the base station tester.Note that the throughput ratio is given by ( 1), TP means the throughput ratio, and r e means the BLER since the retransmission function was disabled in this evaluation.3GPP defines the minimum received power level that satisfies at least 95% of the throughput ratio as the minimum receiver sensitivity [23]; therefore, in this evaluation using the actual LTE-receiving device, the minimum receiver sensitivity was defined as the minimum received power level at which the BLER is at most 5%.The received power level is defined at the receiving antenna output port according to the 3GPP standard.The minimum receiver sensitivity of the LTE-receiving device in each band is shown in Table 2.The minimum receiver sensitivities meet the 3GPP standard [23].The bands evaluated are those used by telecommunications carriers in Japan for UAV operations, and their DL frequencies are 860-870 MHz, 1805-1825 MHz, and 2110-2130 MHz.
Next, we evaluated the impact of EMI caused by EM noise generated by the test UAV on the receiver sensitivity of mobile communication systems.The receiver antenna was placed at eight different positions on the side of the UAV, as shown in Fig. 4.Then, we evaluated the effect of EM noise on the receiver sensitivity by comparing the minimum receiver sensitivity when the UAV is driven with that when it is not driven.Figs.The degradation of sensitivity by 29 dB indicates that the communication range is reduced by approximately 1/13, as determined by solving Frith's transmission formula [24], assuming a line-of-sight space; this degradation is a huge problem for industrial UAVs.To focus on the effect of EMI and the effectiveness of countermeasures for UAVs, the following sections show the results of experiments at position 8, which showed the most degraded sensitivity.

III. EM NOISE COMPONENTS INSIDE INDUSTRIAL UAVs
In general, EMI effects depend on not only the noise power but also the time/spectral characteristics of the interfering noise.We measured the EM noise characteristics of UAVs observed in the communication bands, identified their source, and analyzed the interference mechanism, the results of which are presented in Section III.

A. EM NOISE MEASURED BY ANTENNA MOUNTED ON THE UAV
To evaluate the characteristics of the EM noise that was inputted into the antenna port of the LTE-receiving device, the antenna used for the communication quality evaluation in Section II was used, and the antenna was similarly placed at position 8. Fig. 6 shows the EM noise measurement setup.Mobile communications, which are standardized to operate even with low-power DL signals, can suffer from EMI problems even with weak EM noise.Therefore, this evaluation system is based on the measurement method in [25] and achieves high-sensitivity measurements using a low-noise amplifier (LNA) with a high gain of 45 dB and a low noise figure of 1.2 dB.First, EM noise was measured over a narrow bandwidth of 100 MHz, including each DL frequency band, to clarify the frequency components of the EM noise emanating from the test UAV.The measurement conditions for the spectrum analyzer were peak detection mode, resolution bandwidth (RBW) of 10 Hz, and three sweeps each in Max

B. EM NOISE GENERATED BY INTERNAL IC CHIPS
We disassembled the UAV and measured the EM noise for each internal PCB to identify the sources of the noise shown in Fig. 7.The internal configuration of the UAV is shown in Fig. 8.The 24 V power from an external power supply was stepped down to 12 V and 5 V by several power modules with DC-DC converters.The 5 V power was supplied to the microcontroller unit (MCU) mounted on the control modules including the main control module and other subboards, whereas the 12 V power supply provided power to the propeller.In addition, the MCU mounted on the control modules were connected to each other through controller area network (CAN) communication.EM noise above an IC chip that was mounted in each module was measured using a high-spatial-resolution magnetic field probe, and three major noise sources were found to be present in the test UAV.The characteristics of EM noise of up to 6 GHz from the three major noise sources (power module, control module, and camera module) are shown in Fig. 9.
The frequency characteristics of EM noise from the three major sources are completely different, indicating that the main noise components are different even in the three bands evaluated.The DC-DC converter in the power module is a silicon metal-oxide-semiconductor field-effect transistor (MOSFET) product with a switching frequency of 425 kHz, and the harmonic noise corresponding to its operating frequency extends to about 2 GHz.On the other hand, the MCU implemented in the control module is operated by an oscillator with a frequency of 16 MHz, and harmonic noise from the MCU was observed.The noise from the camera Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.FIGURE 9. Spectra of EM noise generated by camera module, power module, control module, and floor noise.These spectra were captured by a magnetic field probe and the horizontal axis shows the reading value from the spectrum analyzer (SA).
module was radiated by the pixel clock signal, and the pixel clock of this test UAV, which records full-HD-quality video, is approximately 74 MHz.These harmonic noises are emitted from the connector part of the HDMI cable for video transmission, and they extend over the frequency band of more than 6 GHz.In general, EMI in mobile communication devices in UAVs is caused by the mixing of these various EM noises inputted to the communication antenna.However, the frequency bands in which the EM noise generated by each module was predominantly observed are different; therefore, interference countermeasures are required according to the communication frequency band and the location of the antenna.

C. DEGRADATION OF DL PERFORMANCE BY EM NOISE GENERATED BY INDUSTRIAL UAVs
The effects of the EM noise so far on the receiver sensitivity of mobile communication systems were evaluated using a wireless communication system simulator.The simulator is based on a previous study [26], and a simulation setup was constructed to match the receiver sensitivities of the actual LTE receiving device as described in Section II.As shown in Fig. 10, the DL signal of the mobile communication generated by the simulator is added to the EM noise that was recorded by the EM noise measurement.The assumed communication parameters for the DL signal were the same as those in Section II.
The EM noise data were digital data of the measured EM noise within the communication bandwidth sampled for 1 s.The sampling frequency was 1.28 times than that in the communication signal band, and EM noise components within the communication band were recorded by extracting the I (in-phase) and Q (quadrature) components.The mixed signals were passed through the RF front-end model, and the throughput was measured after demodulation and error correction processes by the receiver model.The RF front-end model in previous studies was based on actual RFIC chips for parameters such as NF, gain, and the dynamic range of each part, which enabled a highly accurate simulation.For the RF front-end model shown in this paper, the NF of the first-stage LNA was adjusted to simulate the communication performance of the actual receiver module used in Section II.Specifically, the total NF of the RF front-end model was about 8 dB for the simulation of the 800 MHz band.The evaluation of the throughput with the receiver model was based on the RX data after error correction as shown in Fig. 11.At the transmitter, a cyclic redundancy check (CRC) bit sequence was generated for each block to check whether the information was correctly received.The data capacity per unit of time (throughput) and the ratio of the throughput to the maximum data capacity that can be transferred under the given communication conditions were calculated.
The results are shown in Fig. 12(a)-(c).The sensitivity of the mobile communications in the three bands degraded by about 33 dB, 21 dB, and 23 dB.These results are consistent with those presented in Section II, which showed that EM noise generated by the internal modules can significantly degrade the receiver performance of mobile communications.Thus, multiple EM noise components generated inside the 13.Metal shield for the DC-DC converter that is a commercial product.
UAVs spread to a wide frequency range used by mobile communication for UAV control in the air.Therefore, the receiver sensitivity is degraded by these EM noises.Since the EM noise components observed in each band are different, countermeasures for the EMI problem require the suppression of EM noise generated by each of the major noise sources: the power module, control module, and camera module.

IV. METHODS OF EM NOISE COUNTERMEASURES AND THEIR EFFECTS
Countermeasures to suppress the EM noise were taken for the industrial UAV with self-jamming problems.The countermeasures were not drastic methods such as sub-board integration; instead, parts were changed, the sub-boards rearranged, and the cables and chassis were shielded.In the Sections IV, we describe in detail the countermeasures for the three main targets: PCBs, cables, and chassis.
First, we will discuss the methods of countermeasures for PCBs, which differ between submodules and other parts.For submodules, the measures taken were using multilayered PCBs, changing the oscillator, and implementing bypass capacitors.For multilayered PCBs, a two-layer board was changed to a four-layer board to reduce noise by reducing the current paths in the board.For the originally four-layer and six-layer boards, GND was used on the surface layer to achieve low-noise wiring.As for the oscillator, a square wave oscillator was originally mounted on the board, but by changing to a crystal oscillator, the harmonic component of the clock signal input to the MCU was suppressed.The harmonic component of the waveform output from the MCU was also suppressed by a bypass capacitor.On the other hand, the main control board and DC-DC converter were difficult to customize on the boards since they are commercial products.Therefore, a metal shield was used to prevent the leakage of EM noise, as shown in Fig. 13.Note that a gap is provided for an air-cooling fan; thus, there is no problem with heat dissipation.
Next, we will explain the countermeasures for video transmission and CAN communication cables.The video transmission cables handle HDMI signals, and each pin is connected by a thin coaxial cable.However, the entire thin wire is unshielded, and the number of connectors has 11648 VOLUME 12, 2024 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.increased owing to the use of re-drivers.This generated much EM noise.Therefore, we covered the entire thin wire of the cable with a noise suppression mesh, as shown in Fig. 14, and reduced the number of connectors by reducing the number of re-drivers.For the CAN cables, standard wires were originally used, but USB type C was adopted to suppress the emission noise by using the coaxial cable and metal housing connector.
Finally, we discuss the countermeasures for the chassis as shown in Fig. 15.The original chassis was made of micro-carbon-fiber-filled nylon which has a conductivity less than that of metal, so shielding effectiveness can hardly be expected in the mobile communication band.Therefore, we applied a plating coating to the chassis to shield it from internal EM noise and prevent EM noise from entering the radio communication antenna mounted on the surface of the UAV.The weight of the chassis increased by approximately 400 g owing to the plating, and the antenna is covered with resin; thus, the antenna is not electrically connected to the plating chassis.
The effects of these countermeasures on the suppression of EMI problems are as follows.First, the EM noise characteristics in each band at antenna position 8 shown in Section III before and after the countermeasures are compared in Fig. 16  effect of EMI on mobile communications is small.A comparison of the minimum receiver sensitivity before and after the countermeasures at each antenna position using the actual LTE-receiving device in Section II is shown in Fig. 17 These results confirm that the countermeasures taken for the PCBs, cables, and chassis suppress EM noise from reaching the mobile communication antenna mounted on the chassis and prevent the degradation of communication performance due to EMI.Although these countermeasures are conventional EMC countermeasure methods, they are effective when used appropriately after understanding the problems such as resonance points in the electrically floating UAV that is not connected to the GND.Appropriate countermeasures for the EMI problems are required on the basis of the desired communication performance and the weight and volume of the UAV.

V. CONCLUSION
We evaluated and implemented countermeasures for self-jamming problems for industrial UAVs.EM noise is generated in a wide frequency range from electronic devices mounted on the UAV.EM noise may interfere with wireless communication signals since various modules are densely mounted.The degradation of communication performance in industrial UAVs due to EMI may cause accidents such as the loss of UAV control.In the case of the commercial industrial UAV used in this study, the receiver sensitivity of the mobile communication device mounted on the UAV was degraded by 29 dB owing to self-jamming in the 800 MHz band, which is mainly used for UAV control.In the 1800 MHz and 2100 MHz bands, degradation by more than 10 dB was also observed.Thus, self-jamming problems in industrial UAVs cannot be ignored.We were able to achieve effective noise suppression and improve the receiver sensitivity of the UAV in the three bands by identifying the major noise source by internal component evaluation.
In this study, the EMI problems were evaluated under the condition that a stable communication connection could be established even in areas with weak communication signals from base stations.However, the impact of EMI on mobile communication performance differs depending on the environment in which the UAV is used and the communication performance required of the UAV.The limit of EM noise level varies depending on the use of the UAV, and the countermeasure methods also vary.The countermeasure methods described in this paper are just some examples, but it is necessary to understand the kind of EMI risk each UAV poses, and then take the best EMI countermeasure for each purpose while balancing the trade-off between the noise suppression effect and the UAV performance, such as weight, volume, and cost.This is expected to ensure the safe and efficient operation of industrial UAVs.
RYOTA SAKAI (Member, IEEE) received the B.S. degree in computer science from Kobe University, Kobe, Japan, in 2021, where he is currently pursuing the M.E.degree in science, technology, and innovation.
His research interest includes signal processing for the time-domain analysis of electromagnetic noise from electronic devices.SATOSHI TANAKA received the B.S. degree in electronic engineering from Nihon University.
He was with the Development Department for Semiconductor in Hitachi Ltd. (currently Renesas Electronics Corporation), Tokyo, Japan, where he has involved in the design of radio frequency integrated circuit chips.He was an Instructor with the National Institute of Technology (KOSEN), Gunma College, Gunma, Japan.From 2014 to 2019, he was a Research Associate with Tohoku University, Sendai, Japan.He is currently a Visiting Professor with the Graduate School of Science, Technology and Innovation, Kobe University, Kobe, Japan.His research interest includes the evaluation and suppression of electromagnetic noise for interference with mobile communication devices.

FIGURE 1 .
FIGURE 1. Intrasystem EMI problems inside unmanned aerial vehicles (UAVs) may reduce the coverage area.

FIGURE 2 .
FIGURE 2. Configuration of LTE-receiving device used to evaluate EMI problems caused by the EM noise generated by the test UAV affecting the receiver sensitivity of mobile communication systems.

FIGURE 3 .
FIGURE 3. Overview of the evaluation of EMI caused by the EM noise generated by the test UAV affecting the receiver sensitivity.The LTE-receiving device connects the base station tester placed outside the anechoic chamber.

FIGURE 4 .
FIGURE 4. Antenna positions where differences in the impact on mobile communication systems are measured.Positions 1-4 are on the left side of the chassis and positions 5-8 are on the right side.

FIGURE 5 .
FIGURE 5. Degree of minimum receiver sensitivity degradation at each antenna position in the three bands: (a) 800 MHz band, (b) 1800 MHz band, (c) 2100 MHz band.

FIGURE 6 .
FIGURE 6. EM noise measurement setup near the test UAV.The measurement frequency range is from 500 MHz to 6 GHz.

FIGURE 7 .
FIGURE 7. Spectra of floor noise and EM noise generated by the test UAV in the three bands.The LTE antenna shown in Fig. 2 is used for the measurement, and the vertical axis shows the reading value from the spectrum analyzer (SA).(a) 800 MHz band, (b) 1800 MHz band, (c) 2100 MHz band.

FIGURE 8 .
FIGURE 8. Image and block diagram of electronic modules mounted on the test UAV.

FIGURE 10 .
FIGURE 10.Schematic of the simulation setup for evaluating the impact of EM noises generated by the UAV on the performance of the onboard mobile communication receiver.

FIGURE 14 .
FIGURE 14. Thin wire for HDMI signals covered by noise suppression mesh.

FIGURE 15 .
FIGURE 15.Plating coating for the chassis of the UAV.
FIGURESpectra of EM noise generated by the test UAV in the three bands with and without countermeasures: (a) 800 MHz band, (b) 1800 MHz band, (c) 2100 MHz band.

FIGURE 17 .
FIGURE 17.Effect of countermeasures for the minimum receiver sensitivity at each antenna position in the three bands: (a) 800 MHz band, (b) 1800 MHz band, and (c) 2100 MHz band.
(a)-(c).The results show that the communication performance has improved to almost the same level as the actual device performance at all antenna positions.

TABLE 1 .
DL Configurations of LTE communications for EMI measurement.

TABLE 2 .
Minimum receiver sensitivity of 3GPP standard and test LTE-receiving device.