Aerial Cellular: Tutorial for UAV operators considering cellular-based C2 links Mark Davis, Vice President, Next Generation & Standards Feng Xue, Senior Research Scientist, Intel Labs Reza Arefi, Director of Spectrum Strategy, Next Generation & Standards Jingwen Bai, Research Scientist, Intel Labs Shu-ping Yeh, Senior Research Scientist, Intel Labs Disclosures No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document. Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade. Intel, the Intel logo, are trademarks of Intel Corporation in the U.S. and/or other countries. *Other names and brands may be claimed as the property of others. 2019 Intel Corporation 2
Technical bACKGROUND Agenda Technical Background - Performance, and problems specific to aerial operations. UAS Spectrum Considerations What is 5G What is UTM UAS Services Based on Cellular Antenna Issues Looking Forward What Are The Possibilities 4 Some definitions Some cellular terms are used throughout UE = user equipment = the modem on the mobile side (a cell phone, a UAV, etc.) BTS, eNb, gNb = the base station side (cell tower) serving a cell. Often sectorized, e.g. 3 sectors of 120 Downlink = BTS->UE; Uplink = UE->BTS
This is unfortunate choice of terms where UAVs are concerned, because it is backwards. But it is very embedded into cellular language, and will not change for our convenience. Basics of the aerial vs. terrestrial channel The cellular system was designed for a highly reflective terrestrial channel. Reflections arrive from several directions, and may result in constructive or destructive interference. Movements on the order of will change the channel randomly (result in a different sum of constructive/destructive interference). Result is fading with any movement and even a stationary UE will see fading due to motion of reflectors. 1GHz ~ =1 foot wavelength. Mainstream licensed bands from 700MHz 2.2GHz. Newer higher spectrum in some countries (e.g. CBRS, ~3.5GHz). 5G brings millimeter wave, e.g. 28GHz, 39GHz bands, and higher. (Yes, 30GHz is actually a centimeterbut all this is called mmw anyway). Path loss exponent >2, e.g. about 4 (i.e., path loss near 1/r 4) Basics of the aerial vs. terrestrial channel (2) In the air: Much higher probability of Line-of-sight (LOS)
LOS to too many base stations (called eNB or gNB) This can result in over-propagation, and cause interference and reduce throughput for ALL users (including terrestrial users who currently pay the bills). Path loss 1/r2 Antenna tilted down, so operation is often on sidebands Creating complex coverage situation 3gpp Technical Report with much detail on this topic is public, TR36.777. Main Challenges High Interference: BS antennas downtilt served by side lobes Aerial link has low attenuation Strong interference, and a drone sees more #cells 8 Problem 1 : Interference due to aerial UE Figure 4a from  DL per UE Throughput
1 0.9 0.8 0.7 0.6 0.5 # Drone UEs/sector = 0, IsDir = 0 0.4 # Drone UEs/sector = 1, IsDir = 0 # Drone UEs/sector = 5, IsDir = 0 0.3 # Drone UEs/sector = 10, IsDir = 0 0.2 0.1
0 0 1 2 3 4 5 6 7 8 9
Mbps With more aerial UE, interference rises and throughput drops. On both UpLink (i.e., reverse link, UE->eNB) and DownLink (i.e, forward link, eNB->UE) UpLink (UL) and DownLink (DL) are cellular terms originating from the fact that the eNB antenna is typically elevated and the UE is on the ground. But with an aerial UE, now the UpLink is going down and the DownLink is going up. Thus, reverse and forward may be better choices than up and down. Problem 2 : discontinuous coverage. Main lobe is down tilted In the air, the UE operates largely on side lobes RSRP from one sector at 0m, 50m, 100m, 300m from simulation. Downtilt 10. Red = higher RSRP (reference signal received power) At 0m, coverage matches sector well. Source:: Intel simulations.
Complexity of coverage picture Color codes the eNb with best connection to UE at various heights At altitude, the picture is complex This is challenging for handover algorithms design for terrestrial channel Source: Figure 5 from Mobile Networks Connected Drones: Field Trials, Simulations, and Design Insights, Xingqin Lin et al., Ericsson and Cornell University, https://arxiv.org/abs/1801.10508 Dropped Handovers due to RLF for aerial vehicles (omni antenna)
Various simulation contributions from 3gpp contributions [TR36.777, sec J.2.4] Independent contributors arrived at different numbers, but generally see more dropped HO due to RLF with altitude. This has been a topic of some controversy. The standard is not strictly prescriptive, and HO algorithms designed for terrestrial may vary in aerial performance. UAS Spectrum Considerations International & Domestic Regulations ITU-R Radio Regulations (RR), an international treaty governing the allocations to, and conditions of use of spectrum by, various services, is the ultimate reference document for operation of any wireless application. National regulatory bodies, e.g. FCC, generally adopt RR provisions and reflect in their
domestic rules, but exceptions/deviations also exist. Under ITU-R definitions, UAVs belong to the Mobile Service (MS), which consists of Land Mobile Service (LMS), Aeronautical Mobile Service (AMS), and Maritime Mobile Service (MMS). Unless otherwise specified, allocation of a band to MS means all three subcategories are allowed. An example of exception from RR: 14 Spectrum Needs of UAVs Technical characteristics related to the UAV mission and operational environment affect spectrum needs For instance TX power affects range and operational altitude, which could impact the choice of frequency range (low or high) Spectrum needs for payload and C2 very sensitive to link spectral efficiency (S) and required data rate (D); Bandwidth= D/S Example: Urban environment, D=30 Mbits/s average S=5.4 bits/s/Hz (urban) B=5.6 MHz
cell-edge S=0.15 bits/s/Hz (urban) D B= 200 MHz Spectrum for DAA a function of range, speed, required resolution 15 MBB Spectrum for UAV Operation (USA) // // // // // // Contact author for full table.
Several bands with an aeronautical prohibition in all or parts of the band: BAND 5 (850) and BAND 66 & 70 (Lower AWS below 1710MHz) ARE AFFECTED. Variance in amount of spectrum, associated technical rules, and level of operation Unclear if no restriction still requires aerial service rules (TBD by FCC, still under debate) No single band has the range, bandwidth, and suitable regulatory conditions that satisfy various UAV KPIs MBB spectrum collectively should be used to address all spectrum needs of UAVs 16 Spectrum Regulatory Challenges & Opportunities Protection of incumbents (co- and adj.) are important, pointing to reuse of existing regulatory framework to the extent possible Legacy terms and conditions (e.g. categorizing UAVs as aircrafts) might unduly burden small UAVs and need to be revisited LTE/NR protocols are capable of supporting all UAV radio functions (payload, C2, DAA, eID) through complex frame/message structure on the same radio channel, thus increasing spectral efficiency, and should not be burdened by regulatory segregation of services (LMS, AMS, RLS, etc.) 17 What is 5G?
What is 5G? There is not really one unambiguous answer to that question Operators can use the label with some flexibility. We saw this also in 3G, with multi-carrier HSPA being labeled as 4G by some operators, distinguishing it from 4G LTE. Some things generally associated with 5G: Release15 of the standard and beyond. A new waveform (NR = New Radio). But some Rel 15+ changes are relevant to the 4G (LTE) waveform, or are independent of the waveform.
Some people call such features 5G and others do not; there is no single correct answer. New bands, including millimeter wave. HOWEVER, all old bands also supported. So it is not correct to say that 5G necessarily implies mmw. Extending KPI in multiple directions, generally expected to be outside of the scope of what is useful for humans with smartphones (i.e., targeting M2M and vertical applications). New network architecture more amenable to flexible deployment (e.g. virtualization) 19 Example There is much discussion of applications and spider charts of targets, in various publications. 5G cannot achieve best of all in all dimensions, but can be flexibly tailored. For example, see many spider charts in: https:// www.5g-eve.eu/wp-content/uploads/2018/11/5g-eve-d1.1-requirement-definition-an alysis-from-participant-verticals.pdf 20
Significance for aviation folks 4G LTE is deployed by most operators worldwide. 4G capability is far from being a limiting factor for serving aviation use cases 5G is rolling out.gradually and piecemeal. 5G is often said to be focused on three pillars: URLLC = ultra-reliable low-latency communications EMBB = enhanced mobile broadband MMTC = massive machine type communication. For aviation purpose, it is convenient to think of 4 bins: a. Legacy cellular (2G-4G). This is more than adequate to get started. b. 5G - The three pillars can serve some aviation needs. c. 5G - General features that are not strictly these 3 pillars, but are useful for aviation. d. 5G - Features specifically intended to address UAVs, that happen to be in the same 3gpp releases as what are normally considered 5G (although some revise 4G). Such UAV features are currently in Releases 15, 16, 17. 21 UAV-specific features, brief summary (not including cell-on-wings)
Six CRs approved for Rel 15: 1. Aerial usage subscription 2. UL power control UE specific alpha; Extended range for UE-specific P0 3. Measurement triggering based on number of cells above a threshold 4.
Height report triggering based on H1 and H2 events 5. Adding vertical speed to location info (and thus height report) 6. Network polling of flight path plan Rel 16 : TS 22.125 UAS Support TR 22.825 Remote Identification (*) Rel 17:
TR 22.829 Enhanced UAV support (*) TS 22.125 Further enhanced UAV support based on TR 22.829 conclusions 22 Whats coming for UAV operators (dynamic situation). This means eventually you should have an aerial-specific SIM Several operators have already rolled out some Release 15 UAV changes Reduces network interference Some reduction of uplink throughput. Also can be used to restrict channel, although approaches vary Eventually might be tied to performance improvements, but this part is not available presently. Enforcement for any of this is also unclear
Ideas have been bandied about concerning some formal tie between FAA approvals of various types and registration with an aerial SIM. Very informal conversations, no plans yet (that I know of). Operators have various approaches. One example, is Verizon ALO Airborne LTE Operations available B2B, not through retail Accomplishes the goals using pre-Release 15 methods, because Rel 15 was not available when ALO was rolled out. 23 What is UTM Exponentially rising number of aerial vehicles is leading to increased encounters 350 300 250 200 150 100 50
0 Number of reports filed by manned aircraft pilots each month, reporting a UAV issue 25 Unmanned Traffic Management (UTM) The number of vehicles in the air is increasing exponentially. This is challenging for current systems and regulation. The existing ATM (Air Traffic Management) system is already stressed and having trouble coping with manned traffic Requiring ATM to deal with UAVs is impossible Not just an issue of scale, but of architecture. System needs renovation, not innovation. ATM is fundamentally a P2P (person-to-person) system There is much technical assistance, but in the end the Air Traffic Controller and Pilot are the key decision makers, and are very much in the loop." UTM requires an M2M (machine-to-machine) system Density and agility dictate that much control must be by machine. UTM (UAS Traffic Management) is the effort, worldwide, to define such M2M system.
26 UTM - a new M2M control system for UAVs  There has also been some discussion of use above FL600; see for example: https://www.researchgate.net/publication/330206407_Space_Traffic_Management_with_a_NASA_UAS_Traffic_Management_UTM_Inspired_Architecture 27 UAS ID / Remote ID Separately, law-enforcement wants an electronic license plate to deal with immediate problem of clueless/careless/criminal fliers. Variously termed UAS-ID, Remote ID, eID This will be rolled out before UTM They can be complementary, but it is TBD how much of that potential synergy will actually be achieved, given the different timelines 28
Slide 14 from https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20180002543.pdf Status of aviation standards what is UTM exactly Unfortunately, there is not just one worldwide body Situation is a bit chaotic, but settling. This will (probably) be a year of consolidation. Thus, UTM is not just one thing (yet). There are some commonalities and some differences; Interactive map of worldwide efforts: https://gutma.org/blog/2018/09/17/gutmas-map-of-utm-implementations/ UAV services based on cellular 3gpp concept of C2 over cellular (from TR22.829) Most people would think of something like Model D as using a cell modem on a UAV, with varying ideas of the function designated UTM in this drawing.
32 ASTM F38 02 Remote ID Cellular would be a common method of Networked Remote ID 33 Positioning Performance in the air of OTDOA can be quite a bit better than on the ground Redundancy of OTDOA is questionable because base station itself uses GPS as a time source. But looser methods that do not require tight timing are redundant. Source: Steve Caliguri, Acorn Technologies, https://gutma.org/portland-2019/presentations/ Full presentation is online and includes much more explanation. 34
Example hellaLOC from Acorn Source: Steve Caliguri, Acorn Technologies, https://gutma.org/portland-2019/presentations/ Terrestrial Example Aerial Example 35 Cellular V2X Source: Stefano Faccin, QUACOMM, https://gutma.org/portland-2019/presentations/ Full presentation is online and includes much more explanation.
36 Source: Stefano Faccin, QUACOMM, https://gutma.org/portland-2019/presentations/ Full presentation is online and includes much more explanation. 37 UTM, using NASA architecture as a model
Public/Private line in the center USS (on private side) is the primary UTM controller USS declares itself to be serving a geo; discovers all overlapping self-declared USS serving that same geo Thereafter syncs with those USS FIMS gives dynamic rules Like TFRs e.g. Super Bowl SDSP provide other services (weather, mapping, etc.) USS-UAV Link could be cellular, for example. Normal flow is very simple (normal cellular role in red): Declare series of 4D polygons Get approval Fly and report
Close flight plan. 38 Redundancy options You might find CAA approvals a bit easier with redundancy E.g., redundancy is specifically mentioned in SORA Annex A Various redundancy approaches that include cellular and are in the market: Natural Redundancy of Cellular due to multiple bands Unlicensed + Cellular Satellite + Cellular Cellular with Multiple SIM cards (e.g. SqwaQ) Multiple-tower network (e.g. Aeris) with one SIM card. 39 Im applying for BVLOS wavier, how good does my comm have to be?
This is a long topic with unclear status FAA UAS IPP waiver applications have variable and informal methodology, such as I flew a test there and the call didnt drop. This needs to be formalized more. But not clear how or by whom. ICAO has two documents (PBCS and GOLD) Performance-Based Communications and Surveillance (PBCS) Manual Doc 9869, June 2017 Global Operational Data Link Document (GOLD), Second Edition, 26 April 2013 RTCA DO-264 has rigorous definitions of Continuity, Availability, and Integrity of comms. DO-362 and DO-377 cite these in several target requirements for specific use cases. JARUS WG 5 will develop some requirements CTIA/FAA are working on a test metric list.
3gpp TR36.777 and TR22.829 suggest some goals. If you are applying for a BVLOS waiver or similar using cellular, should you attempt to use some of this? Might not be the best use of time at this point. If your use case closely matches DO-377, this might be worthwhile to use. 40 ANTENNA ISSUES Issues to be aware of when integrating a radio This section is very generalized and simplified You need to be cautious of antenna patter, and should consult an expert during integration of any modem A couple main issues: Items designed for terrestrial use may have a largely horizontal pattern
Thus there is danger of a zenith hole Mounting a module or antenna to your UAV may significantly alter the performance. Example link budget This is from Remote ID work, in unlicensed band. But principles apply to any case; this is just to show zenith issue This is only considering energy arriving at the receiver, and does NOT include the directionality of the receiving antenna, which may further accentuate the problem in the cellular case. Simplified view of a link budget. Transceiver Power Out Tx Losses Tx Antenna Pattern
Path Loss Rx Antenna Pattern Rx Losses Receive Power Required SINR Noise + Interference Floor Interference/Traffi c congestion kTB Thermal noise floor (around -174dbm/Hz) What happens with a vertical null as the drone goes overhead? This is simple line-of-sight model using an ideal dipole antenna (Friis equation, sin^2 toroidal pattern)
Highly idealized, only for the purpose of showing a closed-form solution. Example is from BT 2.4GHz at +8dbm conducted output power. 45 Simple ideal case Distance D in meters -40.002.5 5 Signal Strength vs. Distance for various constant altitudes 10 20 40 80
160 320 640 1280 2560 -50.00 -60.00 ^ | Received Signal power in
dbm | v -70.00 -80.00 -90.00 -100.00 -110.00 -120.00 -130.00 -140.00 Altitude A in meters 2.5 80 5 160
10 320 20 Sensitivity 40 46 Below the line, Remote ID is received; Above the line, Remote ID is not received.
The center line is the case from previous slide. The others are hypothetical curves if situation is -6dB to +6dB worse/better around this nominal case. So region of reception is like an upside-down half-bagel shape around the receiver A= Altitude in meters This is the boundary in D and A where sensitivity is met 700 600 500
400 -6dB -3dB Nomina +3dB +6dB 300 200 100 0 0 100
200 300 400 500 600 700 800 D = Horizontal distance in meters 47 Explanation of previous graph
Equation is shown on next page. Rxdbm = sensitivity level, and A is found as a function of D such that the receiver barely receives the sensitivity level; this then defines the boundary. Generally there will be two Ds for each A: When D is too small, the Remote ID cannot be received because of then downward antenna null. When D is too large, Remote ID cannot be received due to high path loss (simple distance). Avoid over-interpreting this.
The point is simply this: a terrestrial module might not be designed to radiate up/down, and this can lead to a hole vertically. Explanation of previous graph RxdBm=TxEIRP Horizontal Plane dBm +GainsLosses dBm ( 2 + 2 )= A Note: this is a generalization referring to all gains/losses, such as antenna gains, fading margin, insertion losses, etc. They are itemized in ASTM F38 02 working documents
LOOKING FORWARD - What are the possibilities Performance Enhancements Existing networks are more than sufficient We do NOT have a chicken and egg situation where todays cellular networks are only practical with expensive optimization, and such optimizations are only worthwhile after the market exists. However, certain improvements can be made. Here are three possibilities: Automated coverage mapping UE, Network, and combined performance optimizations 5G as a distributed, connected computing environment for aviation 51 Automated coverage planning Many of us have applied for BVLOS waivers using cellular modems. The first question the FAA asks (as they should) is how do you know there is cellular coverage on your
route. This results in a long test and characterization process, with two major problems: It laborious and limited to well characterized areas, and thus provides no method for scaling to any arbitrary flight area. Such a process is far too slow to capture the dynamic nature of the C2 environment. Coverage is dynamic for many reasons (antenna adjustments, power adjustments, etc.) Also handover is a dynamic event, and is the source of most call drops. Beamforming/MIMO can alter coverage intentionally to cover a flight path. To have predictable and scalable automated flight, it is necessary to change this slow process involving paper and people to an instant machine-to-machine exchange. M2M exchange between USS (or flight planner invoking the USS) and the 3gpp network. Several vendors offering solutions; standardization TBD. 52 Optimization: UE-centric for interference challenge Horizontal/Vertical
0 65 deg HPBW UE directional -5 antenna model UE based directional Tx/Rx 35 deg HPBW 15 deg HPBW -10 UAV can use directional pattern for tx and rx from wanted cell(s) -15
Directional pattern comes from special antenna or MIMO -20 -25 Wideband SINR CDF 1 0.9 0.9 0.8 0.8 0.7 0.7
0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.1 0 -10 -5
0 5 10 15 Wideband SINR, dB 20 25 0.1 0 -150
-100 -50 0 50 100 150 Case 5, Omnidirectional Case 1, Omnidirectional Case 5, 65 HPBW, DoT Case 5, 65 HPBW, LOS Case 5, 35 HPBW, DoT Case 5, 35 HPBW, LOS 0.2
30 -30 Angle (deg) 0.3 Omni 65 HPBW, DoT 35 HPBW, DoT 65 HPBW, LoS 35 HPBW, LoS 0.2 UL UE average post-processing IoT 1
0 5 10 15 20 25 IoT, dB 53 Some Observations on using directional Tx/Rx 5 drones per cell. Directional antenna has 65 degree HPBW. Downlink: From BS to UE low traffic
load 0.9 packets/cell /sec high traffic load 1.65 packets/cell /sec With UAV, With UAV, directional, Omni DoT Uplink: From UE to BS With UAV, directional, LoS
No UAV 27.9 28.63 21.74 19.21 21.37 21.51 6.82 22.54 33.37
0 26.42 30.88 31.88 19.89 10.18 14.77 19.49 17.25 8.12
16.28 16.48 0 2.48 6.57 19.91 0 8.84 22.59 24.81
Tput (Mbps) No UAV Ground UE 28.7 22.06 UAV 0 Ground UE UAV With UAV,
With UAV, directional, Omni DoT With UAV, directional, LoS 54 Solutions: Network-centric for interference challenge Full dimensional MIMO and massive MIMO: Utilize MIMO capability for better resolution CoMP (Coordinated muilti point) Coherent transmit/receive from multiple base stations with joint signal processing eICIC (Enhanced inter-cell interference coordination)
Resource (freq+time) coordination among neighboring cells 55 Solutions: UE-Network joint solutions for interference challenge More real-time interactions between UE and network Enhance all aforementioned solutions Directional Tx/Rx FD-MIMO/ Massive-MIMO eICIC/CoMP 3GPP Release 15 new features: New capability for UAV uplink power control. Instead of cell specific power control parameters, now one can apply UAV specific power control More reporting triggers for UE to report link qualities etc. 56 Solutions: mobility support enhancement More measurement reports are supported now based on height, speed, etc.
Flight plan can be reported to network Combine with coverage map and flight plan, one can optimize handover design, beamforming, cross-cell coordination for better mobility support such as handover. For example, simulation has shown that one can dramatically reduce handover failure rate with rout-aware scheduling 57 Tower Convergence. Premise: 5G can serve not only as C2 for UTM, but can serve as the distributed, connected compute structure upon which UTM is deployed. Rationale: 5G rollout requires dense new networks, for capacity and use of mmw. The most efficient way to do this is multi-purpose towers/sites (already a trend) Host multiple licensed operators, unlicensed systems, Smart city sensors, etc. Aviation/UTM functions map naturally to a 4G/5G network Aviation/UTM by nature is localized Objects and data are relevant locally (traffic, weather, RF environment) UTM needs distributed physical sites with tower/power/backhaul/fronthaul
UTM has software components that need to be near the edge for multiple reasons. UTM/USS also needs to be near its C2 network The 4G/5G network itself meets all these criteria. 58 Multi Stage Edge Compute COMMON THEME IN INDUSTRY: The need to distribute applications along multiple processors, from vehicle to deep cloud. Vehicle: lowest latency
Weight constrained Bandwidth constrained (WAN) Low federated awareness Landing / Charging Site: Longest latency (post flight) High bandwidth / lowest communication cost from UAV (LAN) Process latency-tolerant data before forwarding Filter, or convert to metadata Local USS: Serves specific area Full federated awareness in that area Low latency for dynamic re-planning, (e.g., ATM / UTM interactions) Central office to Deep Cloud (various): Cloud
Increased federated awareness Increased latency and communication path Increased economies of scale for storage & compute Lowest cost per bit / flop 59 UAV Landing / Charging Converged tower conceptual diagram (*) Edge Compute S&A unction Motion analytics Payload processing Smart City Sensors Wifi, etc. Unlicensed Modems
Fronthaul Xn Interface USS - USS Interface Etc. GNSS RTK Reference Node 1090 Mhz 978 MHz Remote ID Flarm Etc. Aviation Band Rx Local Micro-weather USS sub-6
Path-based beamforming Primary surveillance UTM / C2 Coordination gNb MMW Other Tower Converged Tower Backhaul NG Interface FIMS, SDSP Interface Etc. (*) aviation items in orange; each is
expanded/described in Backup. Aviation functions on a Cell Tower. 1. UAV Landing/Charging: Especially information collection case needs network of small nesting sites with tower, power and backhaul. Various companies make such nest type products. 61 Aviation functions on a Cell Tower (2). 2a. Edge Compute : S&A / DAA Function (Sense and Avoid = Detect and Avoid) There is no clear framework for the role of DAA in UTM. ASTM F38 02 is working on USS-USS and DAA standards. Compute may be performed at edge to: (1) offload vehicle (2) provide federated information Federated DAA could provide a path forward for UTM/ATM integration
Agile M2M system gets out of the way of non-participating objects. (Federated DAA seems a more popular concept in automated driving than in automated flight) 62 Aviation functions on a Cell Tower (3). 2b. Edge Compute (contd) : Payload-related processing examples: Inference - Mobile AI-directed sensors add another dimension to analytics, but the SW structure and rationale is similar to existing analytics, e.g. OpenVino Add analytics to command a new view, and combine with pole-mounted sensors Also gives opportunity to combine inference and SFM/photogrammetry Related to both payload and UTM: 3d point cloud from photogrammetry can be used to update physical and C2 map Can be used for flight planning as an SDSP (Supplementary Data Service Provider), or for general purpose (Smart City, etc.) 63
Aviation functions on a Cell Tower (4). 2c. Edge Compute (contd) : Motion analytics: AI-based inference can easily determine if UE is aerial. This information can be used in many ways: Optimize network power control Help create airspace picture (from position of flying UEs) Always-on sentinel checking for flight anomalies such as : UE without proper Rel 15 aerial designator; UE without correct flight plan; UE without Remote ID; UE position not matching position reports in USS. Note, this is of interest to security-related agencies, not just aviation related agencies. 64 Aviation functions on a Cell Tower (5). 3. RTK reference network: Release 15 already standardized eNB as an RTK reference node.
Need to have an accurate antenna position; not all installations have this. This has use far beyond UTM; could even foresee consumer usage of RTK in the not-toodistant future. Could be in every smartphone eventually. Can be used to enhance the update physical and C2 map 65 Aviation functions on a Cell Tower (6). 4. Aviation-band Rx. See for example Involi box (ADS-B, Mode-S, Flarm). Currently implemented as separate HW attached to cell towers. This could be implemented as SDR modem in some cases. Remote ID broadcasts can be added to this. This is of high interest to regulators: cell network can also serve as a dense, automated network of UAV sentinels, automatically reporting many types of non-compliant UAVs. This has use beyond UTM Combining this with cell tower allows tower companies or MNOs to provide such aviation monitoring service. 66
Aviation functions on a Cell Tower (7). 5. Local Microweather Two main methods: Sensors mounted at the BTS Exploiting channel model at BTS to infer weather information without additional sensors Both can be augmented with sensor data from vehicles. 67 Aviation functions on a Cell Tower (8). 6. USS USS by nature handles a local area. Needs local interconnect to other USS. Needs low latency Also a USS needs low-latency communication with the C2 network (in this case, eNB/gNB). Hosting USS within the eNB/gNB provides easy path for this. Recently a new element is defined (DSS), sitting above the many peer-level USS. This also needs a distributed implementation, and could be a target.
68 Aviation functions on a Cell Tower (9). 7. eNb/gNb enhancements. Note that there a ample opportunities to improve C2 performance, especially given access to USS information. There are many academic papers on this (*) There are also aviation-directed 3gpp-based networks in the world (e.g., Smartsky, European Air Network) These are built specifically for aviation, and thus have some advantages (*)  S.D. Muruganathan, X. Lin, H.-L. Maattanen, Z. Zou, W.A. Hapsari, S. Yasukawa, An overview of 3GPP Release-15 study on enhanced LTE support for connected drones, Online: https://arxiv.org/ftp/arxiv/papers/1805/1805.00826.pdf.  Mark Davis, Feng Xue, Reza Arefi, et al. Aerial Cellular: What Can Cellular Do for UAVs With and Without Changes to Present Standard. Published in Proceedings of AUVSI Xponential, May 2019  Intel labs recently submitted a paper to Globecomm on Handover improvements. 69 Aviation functions on a Cell Tower (10).
8. Mmw enhancements. Mmw gives several opportunities for UAVs 1. Agile beamforming increases safety and decreases interference, even if the panel is downtilted. 2. Research in using mmw panel for primary surveillance shows feasibility of detection of a 10cm object at ranges of 3-5km 70