Airborne IMT users in precision agriculture: Monte-Carlo analysis of UAV interference in 694–2690 bands

Introduction

Precision agriculture is increasingly adopting unmanned aerial vehicles (UAVs) to provide low-latency telemetry, real-time video, and on-the-fly analytics (Guebsi et al., 2024; Subhranil et al., 2021; Costa et al., 2012; Padhiary et al., 2024; Radoglou-Grammatikis et al., 2020; Toscano et al., 2024). The analysis of UAV utilization in precision agriculture for the period 2020–2024 showed the high relevance of this area of agricultural development for increasing crop yields (Subhranil et al., 2021; Toscano et al., 2024).

Recent studies expand this perspective by demonstrating how UAVs are applied in diverse agricultural tasks. For example, UAV-based multispectral imaging has been used to map soil salinization in large agricultural fields, providing essential data for agrotechnical decision-making (Mukhamediev et al., 2023). Advanced computer vision frameworks, such as YOLOv8x, enable fast and accurate plant detection in crop fields, supporting real-time analytics and variable-rate input application (Mukhamediev et al., 2025). In parallel, new algorithms for route planning in heterogeneous UAV groups allow more efficient coverage of agricultural areas with obstacles, going beyond traditional zigzag flight paths (Yakunin et al., 2025). These developments highlight the broad scope of UAV-based solutions in precision agriculture, which rely on stable and interference-free communications to be effective.

One of the important challenges in using UAV in precision agriculture is the problem of interference-free and stable operation of UAV communication systems, which must use reliable data transmission channels to ensure reliable control and data transmission in real time. Such channels can operate in the frequency bands defined for International Mobile Telecommunications (IMT) (ITU-R, 2015a), providing stable communication even in difficult interference conditions. Frequency bands, such as IMT bands, help drone communication systems minimize interference while extending the operational distances in compare with frequency bands, such as 2.4 GHz or 5.8 GHz.

Many IMT-based modern UAV communication systems operate in the 694–2690 MHz frequency bandss defined by 3GPP (3GPP 38.101-1 NR, 2022; 3GPP 38.104 NR, 2021), either directly or through commercial off-the-shelf cellular modems for command-and-control and payload telemetry. This trend reflects a broader integration of UAVs into IMT ecosystems, where they function as airborne user equipment (UE).

At first glance, UAV-mounted UEs seem identical to handheld terrestrial UEs, since both are constrained to the same maximum transmit power levels (typically 23 dBm). However, airborne operation fundamentally changes their interference footprint. Unlike conventional UEs, which are predominantly indoors or at ground level, UAVs operate outdoors and above the downtilt of base station antennas.

This effect has two major implications:

• Intra-operator interference–UAV transmissions can degrade throughput for neighboring cells in the same network. However, this aspect is largely under the control of each operator, who can apply scheduling, coordination, or resource management strategies as they see fit.
• Inter-operator adjacent-band interference–even when operating within emission masks, UAV out-of-band leakage can couple strongly into base stations belonging to other IMT operators. Unlike intra-operator effects, this cannot be managed by a single operator alone and instead requires regulatory analysis to determine appropriate coexistence rules, separation distances, and spectrum-sharing frameworks.

For this reason, our study focuses on inter-operator interference, as it directly affects how countries and regulators will implement UAV usage in IMT bands to avoid harmful interference across license boundaries. The risks are particularly acute in rural or semi-rural environments typical of precision agriculture, where large cell radii, minimal clutter, and high UAV altitudes exacerbate line-of-sight coupling.

The present work addresses this need by conducting a Monte-Carlo simulation of UAV interference across the 694–2690 MHz IMT bands. The analysis incorporates realistic antenna patterns, ITU-R propagation models, power-control behavior, and UAV operational envelopes. The main outputs are aggregate uplink interference, SINR degradation, and throughput loss for both edge cells and network-wide performance, leading to evidence-based recommendations for operational guidelines and regulatory coexistence frameworks.

Unlike most existing UAV–LTE/5G studies, which primarily focus on intra-operator mobility, coverage, and handover performance of aerial UEs, this work addresses inter-operator adjacent-band compatibility. The objective is not to optimize UAV connectivity, but to assess the regulatory risk that precision-agriculture UAVs pose to incumbent IMT networks operating in neighboring channels. The novelty of the paper lies in (i) a unified Monte-Carlo framework combining ITU-R and 3GPP models for regulatory-style compatibility analysis, (ii) a band-dependent derivation of UAV-to-cell separation distances across the entire 694–2690 MHz range, and (iii) a worst-case single-UAV perspective suitable for spectrum-management decisions. The study explicitly targets inter-operator scenarios, which cannot be mitigated by network scheduling and therefore require regulatory guidance.

The frequency range 694–2690 MHz was selected because it encompasses the most widely deployed IMT spectrum globally, including low bands (<1 GHz) used for wide-area rural coverage and mid/high bands (1–3 GHz) used for capacity enhancement. These bands are particularly relevant for precision agriculture, where large farms are typically served by macrocell deployments operating below 1 GHz, while UAV command-and-control modems often rely on commercial LTE/NR modules designed for the same allocations. Therefore, studying this range ensures that the conclusions are directly applicable to real agricultural communication scenarios and national spectrum regulations.

Interference scenario and challenges

From elevated positions, UAV will have line-of-sight (LoS) conditions to multiple surrounding base stations of other operators, sometimes over tens of kilometers up to distance to horizon. This leads to lower effective path loss, weaker antenna discrimination in the vertical plane, and a higher probability of simultaneous strong coupling to several cells. Consequently, a UAV transmitting at the same nominal power as a handheld UE can create disproportionately stronger uplink interference to non-serving base stations. Figure 1 shows a typical interference scenario where a UAV uplink signal affects a base station (BS) of another operator’s network.

Figure 1 illustrates the geometric interference configuration considered throughout this study. An airborne UAV establishes a serving link to its own base station while simultaneously creating unintended line-of-sight coupling toward multiple neighboring base stations belonging to other operators, which act as victim receivers.

To understand how the interference from UAVs differs from that of handheld UEs in adjacent bands comparing to the handheld UE it is required to see the propagation losses difference from UAV in the direction to the victim BS of the other operator and compare it for UAV and handheld UE. The antenna height of the handheld UE is typically around 1.5 m above ground whereas the UAV height is higher than 30 m above the ground. Figure 2 illustrates the difference in propagation losses for interfering signals transmitted from various altitudes using Recommendation ITU-R P.452 with 50% of time (ITU-R, 2015b). Even at a modest altitude of 30 m, a UAV experiences approximately 20 dB less propagation loss compared to a ground-level UE. This implies that, under otherwise identical transmission conditions, interference generated by a UAV will be at least 20 dB stronger than that from an ordinary UE.

Moreover, ordinary UEs are often located indoors, where penetration losses provide an additional 10–20 dB of attenuation. In addition, body loss effects contribute a further ∼4 dB of reduction. None of these mitigating factors apply to UAV-mounted UEs, which always operate outdoors with unobstructed line-of-sight conditions. Consequently, in a worst-case scenario, the aggregate difference can reach at least 35–40 dB, meaning that UAV interference levels may be orders of magnitude higher than those produced by conventional terrestrial UEs.

Parameters that are used for simulation

The following parameters of IMT networks were used for the analysis (ITU-R, 2013; ITU-R, 2014). These parameters were derived from the ITU reports and recommendations as well as 3GPP specifications (3GPP, 2023; 3 GPP, 2023). Table 1 contains the parameters for IMT networks below 1 GHz, whereas Table 2 contains parameters for IMT networks between 1 and 3 GHz bands.

TABLE 1. Deployment-related parameters for bands below 1 GHz.

TABLE 2. Deployment-related parameters for bands between 1 and 3 GHz.

Figure 3 below provides victim network base station antenna pattern based in horizontal and vertical planes. The study did not consider active antenna systems (AAS) with beamforming, since in the 694–2690 MHz bands most countries still deploy non-AAS base stations for LTE/5G networks.

The vertical and horizontal antenna patterns determine how much energy from high-elevation UAVs is received through side- and back-lobes rather than the downtilted main beam. These patterns therefore strongly influence the magnitude of airborne interference coupling.

At high UAV elevations (approaching zenith angles), the received interference is dominated by the vertical side- and back-lobes of the base-station antenna. These regions are explicitly modeled through the ITU-R F.1336 vertical pattern, which includes extreme elevation suppression. As a result, UAV coupling at near-90° elevation is not assumed isotropic but follows standardized attenuation behavior.

For UAV characteristics, the same transmit parameters as for UE were assumed, the difference is that body loss or indoor location were not applied to UAV and the height of the UAV was varied between 30 m and 2 km. The altitude range of 30 m to 2 km was selected to cover both typical agricultural missions (30–120 m for crop monitoring and spraying) and conservative worst-case high-altitude operations permitted under civil aviation rules. Including high altitudes ensures that the study captures extreme line-of-sight coupling conditions and therefore provides upper-bound interference estimates.

UAV structural self-shadowing is also not applied. This represents a conservative worst-case assumption, appropriate for rural precision-agriculture scenarios where UAV orientation often maintains clear line-of-sight to base stations. In practice, airframe shadowing could reduce interference by several decibels depending on mounting and orientation; therefore, the presented results should be interpreted as upper-bound interference estimates.

To estimate adjacent band interference, spectrum emission masks were used and are provided in Figure 4.

Adjacent-channel interference is computed using an ACIR (adjacent channel interference ratio) based formulation, in which the UAV emission mask is convolved with the victim base-station receiver selectivity. This follows standard ITU-R and 3GPP compatibility practice and accounts jointly for transmitter out-of-band emissions and receiver filtering characteristics.

Methodology of the study

This study addresses a single-UAV worst-case scenario. Aggregation effects of multiple simultaneously active UAVs are not modeled and may lead to nonlinear growth of interference. The results therefore provide conservative guidance for individual UAV operations rather than cumulative fleet behavior.

Active antenna systems (AAS) and beamforming are not considered in this study, as in most countries the IMT networks operating in the 694–2690 MHz bands are still predominantly deployed with conventional sector antennas. Moreover, the use of AAS would in general provide additional spatial discrimination and is therefore expected to reduce or redistribute the coupling between UAVs and non-serving base stations. By excluding beamforming, the analysis intentionally represents a conservative worst-case scenario. Consequently, the derived separation distances are primarily applicable to networks employing conventional sector antennas in rural and semi-rural deployments and should be interpreted as upper-bound regulatory guidance. AAS deployments would be expected to reduce average UAV-induced interference by several decibels and shorten required separation distances. Because many rural deployments still use passive sector antennas, this study intentionally models the more conservative non-AAS case.

A Monte-Carlo snapshot approach was used. Each snapshot generates the network geometry (19 tri-sector sites), populates UEs stochastically in the system area according to the rural deployment scenario case, assigns serving cells, computes path losses, and executes uplink resource/power control operations. Building on the algebraic modeling principles previously applied in digital signal processing using Galois fields (Bakirov et al., 2024; Kadyrzhan et al., 2024; Suleimenov et al., 2023), our Monte-Carlo simulation framework incorporates structured propagation and interference models to capture UAV-specific spectral leakage in IMT bands.

The analysis was repeated for 20,000 snapshots to obtain statistical distributions (CDFs) of the interference metrics. The interference levels and throughput losses were analyzed for the whole network and the edge cell (reference cell) which was the closest to the interfering UAV (ITU-R, 2017; ITU-R, 2002). Figure 5 below shows the topology of the victim network where reference cell is marked as red:

For each CDF curve, 95% confidence intervals were estimated using bootstrap resampling over the 20,000 snapshots. The resulting uncertainty was below ±1 dB, confirming statistical robustness.

Cluster layout, antenna parameters, user statistics, and spectrum-mask definitions were taken from the provided deployment tables and figures. Key assumptions include:

• − Cluster: 19 tri-sector sites (57 sectors).
• − Antenna height for BS: 30 m (with band-dependent variations where specified). Downtilt: 3°
• − Frequency reuse = 1.
• − User TX characteristics: maximum 23 dBm, antenna gain −3 dBi, body loss 4 dB (UAV: no body loss, always outdoor with power control applied.
• − UAV altitudes varied from 30 m up to 2 km.
• − Power control was also applied to the UAV.
• − To calculate propagation losses ITU-R P.452 Recommendation with random percentage of time was used, to calculate wanted signal inside the victim network channel model based on 3GPP TR 38.901 (3GPP, 2020a) was applied with Rural mode and LoS probabilities calculations
• − Tolerable limit of the throughput loss is 5% (3GPP, 2020b)

A cluster of 19 tri-sector sites (57 sectors) was adopted following ITU-R M.2101 system-level simulation guidelines. This size represents the minimum layout that preserves realistic first- and second-tier interference while avoiding edge truncation effects.

The 5% throughput-loss criterion follows 3GPP TR 36.942, which defines this value as the maximum acceptable degradation due to external interference for maintaining user experience and cell-edge fairness. This tolerable interference criterion follows the system-level coexistence methodology defined in 3rd Generation Partnership Project TR 38.876 (3GPP, 2020b), which specifically addresses interference assessment between terrestrial cellular networks and aerial/air-to-ground systems. In this framework, external interference is considered acceptable when the resulting average cell throughput degradation does not exceed 5%.

Recommendation ITU-R P.452 is employed for long-range interfering paths between the UAV and non-serving base stations, consistent with regulatory sharing studies. The P.452 time-percentage parameter is set to random, representing typical interference evaluation conditions.

3GPP TR 38.901 is used for intra-network wanted links, reflecting IMT system behavior.

Figure 6 illustrates simulations with Monte-Carlo analysis of interference from UAV to IMT network in the 694–2690 MHz frequency bands. The red dots are UAV, yellow dot is a base station that serves an interfering UAV, green dots are transmitting UEs of the victim IMT network and blue dots are victim BS receivers.

For i = 1 to number of snapshots inside the generated cluster were generated a number of UEs randomly throughout the system area. For each path from UE to all BSs the C/I was calculated in all cells (sectors) where for kth UE in the jth cell (U E j , k) C/I is denoted by C j , k I j , k where C j , k is the received power from U E j , k, at the j-th BS.

C j , k = P P U S C H j , k × p a t h C o u p l i n g L o s s U E j , k , B S j

The UE power control was used according to the following algorithm:where:

P PUSCH i = min P CMAX , 10 ⁡ log 10 M PUSCH i + P O _ PUSCH j + α j · P L

• P _PUSCH transmit power of the terminal in dBm
• P _CMAX maximum transmit power in dBm
• M _PUSCH number of allocated RBs
• P _{0_PUSCH} power per RB used target value in dBm
• α balancing factor for UEs with bad channel and UEs with good channel
• PL path loss in dB for the UE from its serving BS.

Power control followed the 3GPP open-loop formulation with parameters P ₀ = −85 dBm, α = 0.8, P _CMAX = 23P dBm. A higher α increases compensation of path loss and thus raises UAV transmit power at large distances, potentially worsening interference. Sensitivity tests showed that varying α between 0.6 and 1.0 changes interference levels by approximately ±2–3 dB, confirming that the chosen value represents a realistic yet conservative setting.

I j , k I j , k is the interference power which consists of intra system interference (interference from UEs allocated to same resource blocks in other cells in own network) I int r a j , k, and thermal noise N t:where:andwhere 10log₁₀(kT) is the noise power spectral density in dBm.

I j , k = I int r a j , k + N t

I int r a j , k = ∑ l = 1 , l ≠ j N c e l l P P U S C H l , k × p a t h C o u p l i n g L o s s U E l , k , B S j

N t = 10 ^ 10 ⁡ log 10 k T + 10 ⁡ log 10 b a n d w i d t h o f n R B s + N o i s e F i g u r e B S / 10

Add external interference Z _*y*, consisting of y interferers each with a power P _*ext,y* If interference only occurs in one or few cells, main interference should be placed in an active cell (i.e., a cell with a selected BS).

Then external interference from UAV was calculated to each UL C/I=(C j , k I j , k) for all UEs. Where C j , k, I int r a j , k and N t are the same as above.

Where:and where 10log₁₀(kT) is the noise power spectral density in dBm.

I j , k = I int r a j , k + I e x t e r n a l j , k + N t

I e x t e r n a l j , k = ∑ m = 1 y P e x t , m × p a t h C o u p l i n g L o s s Z m , B S j / × A C I R l i n e a r

N t = 10 ^ 10 ⁡ log 10 k T + 10 ⁡ log 10 b a n d w i d t h o f n R B s + N o i s e F i g u r e B S / 10

Figures 7–9 show cumulative distribution functions (CDF) of interference levels from UAV to the IMT networks within the frequency band in the range 694–2690 MHz.

As may be noted from figures above, the interference is higher for UHF bands (700 MHz, 800 MHz, 900 MHz) whereas for the bands 1800 MHz, 2,100 MHz, 2,300 MHz and 2,500 MHz it is significantly lower. This relates with the better propagation conditions in the UHF bands, as well as slightly different deployment characteristics for the bands higher than 1 GHz compared to the bands that are lower than 1 GHz. It should be noted that bands below 1 GHz are more commonly used in rural areas compared to the 1–2 GHz and 2–3 GHz ranges.

The combination of downtilted BS antennas (main lobe pointing slightly downwards) and UAVs above the downtilt means an airborne transmitter sees lower antenna attenuation in the vertical plane for some sectors (uplink main/side lobes). In many snapshots, the UAV falls within the near-far region relative to the victim BS, producing strong interference power at the receiver. Although power control helps to mitigate this effect, it cannot eliminate it, since UAVs may simultaneously experience low path loss to multiple base stations. Figure 10 below shows receiving gains towards the UAV of the BS in different elevation and azimuth directions.

The above figure quantifies how the effective receive antenna gain of the base station varies as a function of the UAV’s azimuth and elevation angles. The strongest coupling occurs when the UAV lies within the sector’s azimuth coverage and at moderate elevation angles (approximately 5°–30°), where the antenna side- and lower-lobe gains remain relatively high. Although the main lobe provides the highest nominal gain at very low elevation angles, it is mechanically downtilted toward the ground to serve terrestrial users; therefore, interference from UAVs arriving close to 0° elevation is unlikely in practice. Instead, airborne interferers are predominantly received through side- and back-lobes at slightly higher elevation angles, which explains why these regions dominate the observed interference levels.

To calculate the throughput loss of the IMT network the following equations were used to approximate the spectral efficiency over a channel with a given SINR, when using link adaptation:

S p e c t r a l e f f i c i e n c y S I N R b p s H z = 0 f o r S I N R ≤ SIN R MIN α f a c t o r · S S I N R f o r SIN R MIN < S I N R < SIN R Max α f a c t o r · S SIN R Max f o r S I N R ≥ SIN R Max

Where:

• S S I N R: Shannon bound, S S I N R = log 2 1 + S I N R bps / Hz
• α _ f a c t o r: Attenuation factor, representing implementation losses
• SIN R Min: Minimum SINR of the code set [dB]
• SIN R Max: Maximum SINR of the code set [dB].

The parameters, SIN R MIN and SIN R MAX can be chosen to represent different modem implementations and link conditions using Table 3 below.

TABLE 3. Parameters describing baseline link level performance for 5G NR.

The balancing factor α and related link-level parameters follow commonly used values in 3GPP system-level evaluations.

The obtained SINR values can be then mapped in accordance with Figure 11 curves that are based on the expressions above to calculate spectral efficiency reduction due to the external interference from UAV.

Spectral efficiency vs. SINR mapping used:

η S I N R = max 0 , A · ⁡ log 2 1 + S I N R B

Figures 12–14 below show simulated throughputs of the uplink for the whole network and in the edge cell (reference cell) band within the range of 694–2690 MHz.

Tables 4–6 present throughput losses of the edge cell and the whole network for varies number of the bands within the range 694–2,690 MHz.

TABLE 4. Results for 700 MHz, 800 MHz, 900 MHz bands.

TABLE 5. Results for 1,800 MHz, 2,100 MHz bands.

TABLE 6. Results for 2,300 MHz, 2,500 MHz bands.

Key observations that can be made based on the results:

• − Strong distance dependence: Throughput loss at the edge cell drops rapidly with separation; at 5 km losses can be severe (up to ∼24% at sub-1 GHz), while by ∼20–30 km losses are typically below a few percent.
• − Lower frequencies more harmful: For identical separation, lower bands (700–900 MHz) show the highest edge-cell degradation because of better propagation and lower path loss. Aggregate network impact is smaller than the worst-affected cell but still non-negligible for small separations.
• − Edge vulnerability: Most of the interference effect is concentrated on the edge/reference cell where the UAV has strong LoS to the victim sector(s). Network-wide impact is diluted due to geometry and scheduling.
• − UAV altitude and LoS: Higher UAV altitudes increase probability of LoS to multiple BSs and therefore can increase the aggregate interference footprint; this is especially relevant for rural deployments with tall cell radii. (Note: altitude sweeps were performed in the simulations per the supplied dataset.)

For low altitude (<100 m) operations close to populated areas, UAVs using IMT uplink can cause meaningful service degradation in nearby base stations, particularly at sub-1 GHz channels. For large agricultural fields where UAVs typically operate many kilometers from cellular infrastructure, aggregate network impact is often modest (network loss <∼1–2% for separations ≥20 km), but localized (edge cell) outages may still occur. Operators and regulators should be mindful that even a small number of airborne IMT users transmitting at UE power can produce disproportionate harm to a few well-aligned base stations.

The Monte-Carlo simulations quantified throughput loss in both the most affected edge cell and the network average as a function of the separation distance between the UAV and the victim base stations. The results reveal a strong dependence on operating frequency and distance.

For frequencies below 1 GHz (700/800/900 MHz), the impact was the most severe. At a separation of 5 km, the edge cell experienced a throughput loss of approximately 23.7%, far exceeding the 3GPP tolerable limit of 5%. Even at 10 km, the edge-cell loss remained at 11.4%, only falling below the 5% threshold at 20 km, where the value was 3.9%. In contrast, the network-wide average losses were lower, with 2.7% at 5 km and less than 1% beyond 20 km. These results highlight that the edge cells are disproportionately vulnerable compared to the average behavior of the network.

In the 1–2 GHz range (1,800/2,100 MHz), the interference effect was less pronounced but still significant at small separations. At 5 km, the edge cell showed a throughput loss of 12.3%, again above the tolerable limit. At 10 km, however, the loss fell to 3.6%, already within the acceptable range. For larger separations, such as 20 km or more, the edge-cell loss was below 2%, and the network average stayed consistently below 1%.

In the 2–3 GHz range (2,300/2,500 MHz), the edge-cell losses were the smallest. At 5 km, the throughput loss was 7.6%, slightly above the 5% tolerance. By 10 km, the loss decreased to 2.3%, and at 20 km it dropped below 1%. Network-average losses in this band were modest across all separations, staying below 1.3% even at 5 km and declining to below 0.5% beyond 20 km.

The recommended separation distances should be interpreted as conservative guidance for large, flat rural farms typical of precision-agriculture deployments. In smaller farms or in hilly/vegetated terrain, additional clutter, foliage, and partial shadowing can increase propagation loss and may in some cases allow shorter safe separation distances. However, because UAV interference predominantly follows long slant-path, line-of-sight air-to-ground propagation, and because rural agricultural environments generally lack buildings or dense obstacles that could provide effective shielding, terrain irregularities are unlikely to introduce substantial additional attenuation over the multi-kilometer distances considered in this study. Consequently, the simulated results already approximate near–worst-case coupling conditions, and dramatic reductions of interference due to terrain effects are not expected in practice. Overall, the results confirm that UAV interference is most critical at frequencies below 1 GHz, where acceptable performance is only achieved at separations of around 20 km or more. For the 1–2 GHz range, safe operation can be achieved at approximately 10 km separation, while for the 2–3 GHz range, distances of 5–7 km are sufficient to keep throughput losses within the 3GPP-specified tolerance of 5%. These findings provide clear, band-dependent guidance on how UAV deployments in precision agriculture should be managed to ensure compatibility with existing IMT networks.

Conclusion

Monte-Carlo analysis of UAV interference for airborne IMT users in the 694–2690 MHz range shows that UAVs, despite transmitting at the same nominal power as terrestrial UEs, can cause substantial localized uplink interference in victim base stations. At close separation, edge-cell throughput losses can exceed 20% at sub-1 GHz frequencies, 12% at mid-bands, and 7% at high bands.

According to 3GPP specifications, a 5% throughput loss due to external interference is the maximum tolerable degradation for IMT systems. Using this criterion, our simulations yield the following recommended minimum UAV-to-cell separation distances to ensure safe coexistence:

• − Sub-1 GHz (700/800/900 MHz): UAVs should maintain at least 20 km separation from IMT base stations.
• − Mid-band (1,800/2,100 MHz): A minimum 10 km separation is required.
• − High-band (2,300/2,500 MHz): A minimum 5–7 km separation suffices.

To reduce separation distances, it may be possible to do one or several following measures:

• − Enforce stricter out-of-band emission masks for UAV modems to reduce adjacent-channel leakage, particularly for operations near IMT incumbents.
• − Explore short-range coordination (via database or control channel) between UAV operators and mobile network operators (MNOs) to avoid scheduling UAV uplink in the adjacent RBs used by nearby cells, or to request temporary guard resources.
• − If operations must occur close to cellular infrastructure, prefer lower altitudes (within safe flight envelopes) to increase path loss to distant cells and concentrate coupling to fewer, more manageable victim sectors.
• − Require UAV operators to notify MNOs when large-scale agricultural missions are scheduled within risk zones, and log transmission parameters for post-incident analysis.

It is important to emphasize that this study does not address intra-operator interference, which is under the control of individual mobile network operators. Instead, the focus is on inter-operator adjacent-band interference, as it directly affects regulators and spectrum managers. Ensuring that UAV operations respect these recommended separation distances will be critical for countries to define coexistence frameworks, assign operational zones for UAVs, and avoid harmful interference across IMT license boundaries.

In summary, while UAVs bring clear benefits to precision agriculture, their unique interference footprint requires regulatory safeguards. Adopting band-dependent minimum separation rules, combined with UAV power management and emission mask enforcement, will ensure that agricultural UAV deployments can expand without compromising IMT service quality.

While our study has focused on the interference footprint of UAVs in IMT bands, secure and efficient communication remains another cornerstone of UAV deployment in precision agriculture. Lightweight group signature schemes based on PUFs (Sysoyev et al., 2025) complement interference mitigation measures by ensuring data authenticity, privacy, and robustness in resource-constrained UAV swarms. Taken together, interference-aware coexistence frameworks and lightweight cryptographic protocols provide a holistic foundation for safe, efficient, and secure UAV-based agricultural systems.

The present study is intentionally limited to a single-UAV worst-case scenario and to IMT networks employing conventional sector antennas. The aggregation effects of multiple concurrently operating UAVs, which may lead to nonlinear growth of interference, are not addressed and constitute an important direction for future work. Likewise, the impact of active antenna systems (AAS) and beamforming, which are expected to provide additional spatial discrimination and potentially mitigate UAV interference, has not been modeled. As AAS deployments expand into the 694–2690 MHz range, their influence on air-to-ground coupling and coexistence conditions should be systematically evaluated. Future extensions of this work will therefore consider multi-UAV scenarios and AAS-enabled networks to refine and generalize the separation-distance guidance presented here.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Peiyuan Qin, University of Technology Sydney, Australia

Reviewed by: Nameer Hashim Qasim, Alrafidain University College, Iraq

Xingchen Wei, Academy of Military Science of the Chinese People’s Liberation Army, China