Chapter 13

Enhanced LTE connectivity for drones

Abstract

This chapter presents new features introduced in 3GPP LTE Release 15 for enhancing wide-area connectivity for drones beyond visual line-of-sight. It starts with descriptions on different propagation channel characteristics for drones at a high altitude, and compare them to conventional terrestrial network channel models. These different propagation channel characteristics present certain unique challenges for providing cellular connectivity to drones from both device and network perspectives. This chapter continues on and describes how these challenges are addressed by the new features introduced in 3GPP LTE Release 15.

Keywords

Drones; Unmanned aerial vehicle (UAV); Aerial vehicle; Capacity; Mobility; Handover; Flight path; Power control; Interference; Inter-cell interference; Line-of-sight propagation; Antenna gain; UAV identification; Mobility event; Measurement

1. 13.1 Introduction 
2. 13.2 Propagation channel characteristics 
3. 13.3 Challenges 
4. 13.4 LTE enhancements introduced in 3GPP Rel-15 
   1. 13.4.1 Interference and flying mode detection 
   2. 13.4.2 Flight path information for mobility enhancement 
   3. 13.4.3 Subscription-based UAV identification 
   4. 13.4.4 Uplink power control enhancement 
   5. 13.4.5 UE capability indication 
5. References 

13.1. Introduction

Drones, also known as unmanned aerial vehicles (UAV), have in recent years gone much beyond being toys of hobbyists to becoming the centerpieces of many innovative use cases which have a potential to bring significant social-economic benefits. For example, drones are increasingly used for aiding search, rescue, and recovery missions during or in the aftermath of natural disasters like tornados [1], wildfires [2], hurricanes [3], etc. The commercial use cases of drones are also developing rapidly, including package delivery, inspection of critical infrastructure, surveillance, agriculture, etc. [4]. Wide-area connectivity is considered as one of the most critical technological components for fully realizing the potential of drone applications. An important aspect is command-and-control communications with drones beyond visual line-of-sight (VLOS). Reliable command-and-control communications are critical for drone operation safety and for enforcing aviation rules. Furthermore, many of the aforementioned drone use cases require connectivity, e.g., for delivering real-time imagery or video.

Recognizing an opportunity to expand the use cases of cellular networks for the flying things, 3GPP started a study on Enhanced LTE Support for Aerial Vehicles in 2017 [5]. The study focused on new aspects that are associated with connecting the flying things, in comparison with connecting the terrestrial devices. These aspects include.

1. • Propagation channel characteristics
2. • Downlink interference - aerial vehicles as victims
3. • Uplink interference - aerial vehicles as aggressors
4. • Handover performance
5. • Identification of aerial vehicles

All of these aspects are discussed in this chapter.

The study produced a comprehensive technical report [6] and concluded that LTE networks prior to Release 15 are already capable of serving flying drones; however, as the number of flying drones increases, certain challenges emerge. 3GPP then took action in Release 15 to introduce enhancement features for serving drones more efficiently and for better managing the impact on terrestrial devices should more flying drones are connected to LTE networks in the future. In Section 13.4, these enhancement features are described. Release 15 targets aerial vehicles of maximum height 300   m above ground level (AGL) and maximum speed 160   km/h [6]. Throughout this chapter, we will use drone, UAV, and aerial vehicle interchangeably. However, most descriptions in this chapter apply generally to cellular connectivity for all flying things which meet the aforementioned AGL and speed criteria.

13.2. Propagation channel characteristics

The propagation channel characteristics experienced by a drone is very different, compared to a terrestrial device. Therefore, understanding the difference in propagation channel characteristics provides a great insight in the potential drone communication problems that need to be addressed.

Fig. 13.1 Line-of-sight propagation channel probability for drones at different heights (rural macro-cell environment).

When a drone is at a height above buildings and trees, it has a much greater line of sight (LOS) probability to the base station antenna. Fig. 13.1 shows LOS probability in a rural macro-cell environment at various heights as a function of the horizontal distance to the base station, based on the channel model in Ref. [7]. It can be seen that the LOS probability increases as the drone flies higher. As the drone moves farther away from the base station, the elevation angle between the drone and base station antenna decreases. This increases the probability of having an object on the signal path, and therefore decreases the LOS probability. This phenomenon can also be interpreted from Fig. 13.1.

With a LOS propagation channel, there is no object on the signal path causing penetration, refraction, diffraction, and reflection loss. As a result, the overall path loss is smaller than in a non-line-of-sight (NLOS) channel when compared at the same distance. An example based on the channel model in Ref. [7] is shown in Fig. 13.2. Comparing the path loss between a drone at 50   m height with a LOS channel and a terrestrial device with NLOS channel in a rural macro-cell environment further illustrates this point. Observe that the path loss slope is smaller in the drone LOS case. The terrestrial device at 680   m from the base station experiences a 110   dB path loss. In comparison, the same path loss is experienced by the drone at 7380   m from the base station.

Fig. 13.2 Comparison of path loss in rural macro-cell environment with carrier frequency at 700   MHz between a terrestrial device at 1.5   m height with an NLOS propagation channel and a drone at 50   m height with an LOS propagation channel.

To complete the whole picture of drone propagation channel characteristics, it is also important to examine the base station antenna pattern. In a cellular network, a base station antenna is typically down-tilted to provide good terrestrial coverage. Antenna down-tilting also helps in alleviating inter-cell interference. Antenna down-tilting, however, is far from optimal in terms of aerial coverage. An example of base station antenna pattern is shown in Fig. 13.3 and 13.4 based on a column of 8 antenna elements. Fig. 13.3 shows antenna gain over azimuth angle ϕ . The antenna pattern in the example is a sector antenna, so the antenna pattern is designed to have sufficient antenna gains in the desired sector coverage area, which spans horizontally over certain azimuth angles centered at the boresight (0°). The sector antenna pattern is also designed to have an overlapped region between neighboring sectors in the sense that the antenna gain at the sector border is not too low. This is desirable for a terrestrial device to be handed over to an adjacent sector before the signal strength of its current serving sector drops too low. The antenna pattern example in Fig. 13.3 is for a 120-degree sector. The antenna gain at the sector border, i.e. 60° from the boresight, is approximately 7   dB below the peak.

Fig. 13.4 shows the antenna gain pattern over a range of elevation angles θ . In this example, the antenna is down-tilt by 3° and therefore the boresight points to the direction of 87-degree elevation angle. (The reference directions are 0 and 90° for the ground and the horizon, respectively.) Observe that less than 10-degree elevation angle above the boresight, there is a null, and in that direction, the antenna gain is very small-more than 35   dB lower compared to the peak antenna gain. Further increasing the elevation angle from the null, the antenna gain increases compared to the null, although still more than 15   dB lower than the highest antenna gain in the boresight direction. The elevation angles between the nulls, but not within the main-lobe in the boresight direction, can be said to be the antenna vertical sidelobes. These vertical sidelobes allow a drone flying above the base station antenna height to be covered. Although the antenna gain in a sidelobe is 10–25   dB lower compared to the main-lobe, the reduced path loss experienced by a drone with LOS can offset such reduction in antenna gain. For example, observe in Fig. 13.2 that the difference in path loss between the drone and terrestrial device is 24   dB   at a 1   km distance from the base station. This difference can very much compensate the reduction in antenna gain.

Fig. 13.3 Horizontal (azimuth) BS antenna pattern (sector antenna with a column of 8 antenna elements).

Fig. 13.4 Vertical BS antenna pattern (sector antenna with a column of 8 antenna elements).

The nulls of the antenna pattern at certain elevation angles however post a bigger challenge. When a drone flies at a constant height over a certain distance, the elevation angle changes, and the drone may enter the antenna null of its current serving cell. An example is illustrated in Fig. 13.5, showing the antenna pattern with warmer color (e.g. orange and yellow) representing higher antenna gains and cooler colors (e.g. blue and green) representing lower antenna gains.

Fig. 13.5 An example of a drone's flight path cutting through antenna nulls.

13.3. Challenges

The propagation channel characteristics as described in Section 13.2 gives rise to challenges in connecting drones to cellular networks.

1. • Due to higher LOS probability and reduced path loss, a drone flying above a certain height may receive downlink signals with good signal strength from a greater number of cells. One of these signals would correspond to the serving cell; however, all the other signals would contribute to interference. Thus, a drone could be a victim of pronounced inter-cell interference, which posts challenges to maintaining good signal-to-noise-plus-interference ratio (SINR) in the downlink. An example is shown in Fig. 13.6. It shows the distributions of the SINR for drones and for terrestrial devices, respectively. The scenario is set up according to the rural macro-cell scenario described in Section A.1 of Reference [6]. Furthermore, a worst-case scenario where all the cells are fully loaded is assumed. It can be seen that the SINRs experienced by drones are much lower compared to those of terrestrial devices.
2. • Also due to higher LOS probability and reduced path loss, the uplink signal from a drone flying above a certain height may reach a greater number of cells, causing interference to many cells. Thus, in the uplink, the drone may be an aggressor. As some of the drone use cases require drone transmitting a rather large uplink payload, e.g. video traffic, the interference generated by a drone in the uplink could significantly reduce the uplink capacity in the network.

3. 

   Fig. 13.6 Downlink SINR distribution in a rural macro-cell environment.

4. Table 13.1

   Impact of drones on terrestrial device uplink performance in a rural macro-cell scenario defined in Ref. [6]. (Case 1: no drones in the network; Case 2: 1 drone in every 10 sectors; Case 3: 1 drone per sector; Case 4: 3 drones per sector).

   Offered traffic per cell	3.85 [Mbps]				7.45 [Mbps]
   Drone density	Case 1	Case 2	Case 3	Case 4	Case 1	Case 2	Case 3
   Resource Utilization [%]	20.00	20.00	22.14	28.30	50.00	51.06	70.27
   5th-percentile terrestrial user throughput [Mbps]	7.59	7.59	6.64	4.50	2.91	2.71	1.11
   medium terrestrial user throughput [Mbps]	20.20	20.00	18.22	13.74	10.80	10.32	5.73
   Mean terrestrial throughput [Mbps]	18.17	18.06	16.69	13.32	11.79	11.37	7.28

   

The results shown in Table 13.1, taken from Section D.2.1 in Ref. [6], illustrate such phenomenon. The density of drones in the network is increased from Case 1 (no drones in the network) to Case 4 (3 drones per sector). However, the offered traffic per cell is fixed at either 3.85   Mbps or 7.45   Mbps. Thus, as the drone density increases, the traffic mix is more from drones.

The offered traffic represents the average rate of the traffic arriving at a cell, averaging over time and over all the cells in the network. When a packet arrives, how fast the packet can be delivered depends on the availability of radio resources and the SINR of the link between the device and the serving base station. The packet throughput statistics of a user is referred to as user throughput in Table 13.1. It can be seen that as the mix of drone traffic increases, the network radio resource utilization increases. This is mainly due to an increase in the network interference level in the uplink. Furthermore, the impact on terrestrial user performance can be significant when the network offered traffic load is high and when the drone density is high.

1. • The impact on uplink capacity makes it desirable for a cellular network operator to regulate air-borne connectivity. For example, an operator may require special subscription to authorize network connection while air-borne. Furthermore, in some regulatory regions, aerial vehicles are not allowed to connect to cellular network without authorization. To be able to enforce regulation and support authorization, the network needs to be able to identify an aerial vehicle and need to have a mechanism to verify the subscription or authorization status of the aerial vehicle.
2. • Due to a sharp drop-off in the antenna gain pattern when a drone is in coverage through the antenna sidelobes, the drone may need to complete handover from its current serving cell to a new cell in a shorter time frame compared to a terrestrial device.

Some of these challenges can be addressed using implementation-based solutions or using existing LTE features. For example, the downlink interference problem can be addressed by the LTE-M coverage extension feaures described in Chapter 7. In addition, the drone radio receiver can implement receiver beamforming to focus its receive beam to the direction of the desired signal. With this, the interference from other cells, not in the same direction as the desired signal, will not cause pronounced performance degradation. The uplink interference problem can be addressed using power control techniques described in Ref. [8] and coordinated multi-point based approaches described in Ref. [9].

Aerial vehicle identification problem can also be addressed through implementation-based solutions. Examples can be found in Ref. [10], where machine learning approaches are applied to device measurement reports to classify whether the device is in flying mode or not. The radio measurements can be, e.g., received signal strength indicator or reference signal received power.

In the next section, we describe the LTE features specified in 3GPP Release 15 for addressing the challenges described in this section.

13.4. LTE enhancements introduced in 3GPP Rel-15

13.4.1. Interference and flying mode detection

To assist the network in identifying situations where a drone is generating significant uplink interference in radio resource control (RRC) connected mode, LTE in Release 15 added two reporting events H1 and H2.

1. • Event H1: the aerial vehicle height is above a threshold
2. • Event H2: the aerial vehicle height is below a threshold

The threshold is configured by the network. 3GPP Release 15 supports the height threshold as high as 8880   m above the sea level. The two new events can trigger a drone reporting its height and location information. The drone may additionally include its horizontal and vertical speed information in the report. Such reporting helps identify a drone above or below the network set height threshold. Obviously, this feature can be directly used for flying mode detection. In addition, as discussed in Sections 13.3, when the drone is flying above a certain height it may be a victim of inter-cell interference in the downlink and an aggressor for uplink interference. This feature can therefore also be used for interference detection.

In addition to the two newly introduced events H1 and H2, extensions were also introduced to existing events such as A3, A4, and A5 to help interference and flying mode detection. Events A3, A4, and A5 are mobility events defined for RRC connected mode. The definitions of these events are described below.

1. • Event A3: A neighbor cell becomes better than the primary serving cell
2. • Event A4: A neighbor cell becomes better than a threshold
3. • Event A5: The primary serving cell becomes worse than a first threshold and a neighbor cell becomes better than a second threshold

The Release 15 extension is that the network can further configure a threshold N in terms of the number of neighbor cells satisfying one of these events. For example, if the number of neighbor cells satisfying event A3 is greater than N, the drone reporting will be triggered. Such a triggering condition corresponds to drone experiencing that a greater number of neighbor cells having better signal strength, e.g., in terms of reference signal received power, or quality, e.g., in terms of reference signal received quality, than the primary serving cell. This is an indication that the drone is experiencing significant interference from many neighbor cells, which can very well imply that the uplink signal from the drone will generate interference at these neighbor cells. Similarly, when event A4 or A5 is met by many neighbor cells for a drone, the drone may have an interference problem. Again, when a drone is experiencing an interference problem, the drone is mostly likely air-borne. Thus, this feature can be used for flying mode detection.

Both features described in this section allow the network to detect scenarios where a drone is experiencing or generating significant other-cell interference. By detecting which drones having interference problems, the network can take appropriate measures to address the interference problem. One example of such actions is to set appropriate device-specific power control parameters based on the features described in Section 13.4.4.

13.4.2. Flight path information for mobility enhancement

To improve mobility performance in RRC connected mode, 3GPP Release 15 introduces a feature that supports the network to request a drone to provide its flight path information to the network via RRC signaling. The flight path information can be used by the network to determine a suitable new serving cell for the drone.

The flight path information includes a list of waypoints along the drone's planned flight path. Up to 20 waypoints can be included. The flight path information may further include the time stamp of the planned arrival times at each waypoint.

13.4.3. Subscription-based UAV identification

To support the authorization of LTE connectivity to a drone, 3GPP Release 15 introduces subscription-based UAV identification. The signaling of UAV subscription is illustrated in Fig. 13.7. The subscription information is stored at the home subscriber server (HSS). Home subscriber server provides the subscription information to the Mobility Management Entity, which keeps the record of which base station is serving the drone and therefore can forward the subscription information to the base station that the drone is connected to. Based on the subscription information, the base station may deny the service to the drone if it determines that the drone is in the flying mode but has no UAV subscription. For X2-based handover, the existing base station can forward the subscription information via the X2 interface to the new serving base station.

13.4.4. Uplink power control enhancement

It is mentioned in Section 13.3 that a drone in flying mode may generate significant uplink interference in many cells, and therefore may have pronounced impact on network capacity. This problem is addressed in 3GPP Release 15 with the introduction of uplink power control enhancement.

Fig. 13.7 Signaling of UAV subscription information.

LTE open-loop power control for Physical Uplink Shared Channel (PUSCH) is described in Section 9.3.2.8.1. The open-loop power control determines the transmit power level of PUSCH based on the parameters below.

1. • The bandwidth of PUSCH. The higher the PUSCH bandwidth is, the higher the transmit power.
2. • Adjustment for modulation-and-coding scheme (MCS) used for PUSCH. MCS of higher spectral efficiency requires higher transmit power.
3. • A parameter that is related to the target received power level at the base station. We will refer to this parameter as P 0 .
4. • A pathloss compensation factor. If the path loss between the serving base station and the device is estimated to be L in dB, the pathloss compensation is α L dB, where α is referred to as the fractional pathloss compensation factor.

In LTE, up to 3GPP Release 14, the setting of the fractional pathloss compensation factor α is cell-specific, but not device-specific.

Motivated by the observation that a flying drone tends to have a different pathloss slope (see Fig. 13.2) and is likely to generate significant uplink interference compared to a terrestrial device, 3GPP Release 15 makes it possible to configure the fractional pathloss compensation factor α on a device-specific basis.

The P 0 parameter prior to Release 15 can already be configured on a device-specific basis. This is achieved by signaling a device-specific P 0 adjustment factor. However, its value range is from -8 to +7   dB. To increase the flexibility of PUSCH power control for a drone, Release 15 extends the value range of the device-specific P 0 adjustment factor to be from -16 to +15   dB.

The device-specific P 0 adjustment factor and α are included in a dedicated RRC signaling message.

13.4.5. UE capability indication

All the UAV features introduced in 3GPP Release 15, as described in this section, are optional for general LTE devices. A device can indicate whether it supports these features in the RRC information element UE-EUTRA-Capability [11]. However, for devices which have UAV subscription as described in Section 13.4.3, supporting the reporting of the two new height related events (H1 and H2) as well as the extension of mobility events A3, A4, A5 described in Section 13.4.1 is mandatory [12].

References

[1]. Drones in disasters: Eyes in the air play key role in Alabama tornado zone – and beyond . USA Today; March 6, 2019. https://www.usatoday.com/story/news/nation/2019/03/06/drones-disasters-alabama-tornado-recovery-relief/3077948002/.

[2]. California's fires face a new, high-tech foe: Drones.  CNET, . August 27, 2018. https://www.cnet.com/news/californias-fires-face-a-new-high-tech-foe-drones/.

[3]. These drones and humans will work together in hurricane florence recovery efforts.  Forbes, September 16, . 2018. https://www.forbes.com/sites/jenniferhicks/2018/09/16/these-drones-and-humans-will-work-together-in-hurricane-florence-recovery-efforts/#30703b38b714.

[4]. Lin X, Yajnanarayana V, Muruganathan S.D, Gao S, Asplund H, Maattanen H.-L, Bergstrom M, Euler S, Wang Y.-P. E. “The sky is not the limit: LTE for unmanned aerial vehicles”.  IEEE Commun. Mag., . July 2017;Vol. 56(4):pp. 204–210.

[5]. NTT DOCOMO INC., Ericsson.  RP-170779, New SID on enhanced support for aerial vehicles, 3GPP TSG RAN meeting #75 . 2017.

[6]. Third generation partnership project, Technical Report 36.777, v15.0.0. “Study on enhanced LTE support for aerial vehicles” . 2017.

[7]. Third generation partnership project, Technical Report 38.901, v15.0.0. “Study on channel model for frequencies from 0.5 to 100 GHz” . 2018.

[8]. Yajnanarayana V, Wang Y.-P. E, Gao S, Muruganathan S, Lin X.  “Interference mitigation methods for unmanned aerial vehicles served by cellular networks,”. 2018 IEEE 5G World Forum (5GWF), Silicon Valley, CA . 2018:pp. 118–122.

[9]. Määttänen H.-L, Hämäläinen K, Venäläinen J, Schober K, Enescu M, Valkama M. System-level performance of LTE Advanced with joint transmission and dynamic point selection schemes.  EURASIP J. Appl. Signal Process, . November 2012;Vol. 2012(1):247.

[10]. Rydén H, Redhwan S.B, Lin X.  “Rogue drone detection: a machine learning approach”, 2019 IEEE Wireless Commun. and Networking Conf., Marrakech, Morocco, April 15–19 . 2019.

[11]. Third generation partnership project, Technical Specifications 36.331, v15.3.0. “Evolved universal terrestrial radio access (E-UTRA) and evolved universal terrestrial radio access network (E-UTRAN); radio resource control (RRC); protocol specifications” . 2018.

[12]. Third generation partnership project, Technical Specifications 36.306, v15.3.0. “Evolved universal terrestrial radio access (E-UTRA) and evolved universal terrestrial radio access network (E-UTRAN); user equipment (UE) radio access capabilities”, 2019.