Airport Connectivity Infrastructure for Autonomous Vehicles

Research Report: 5G, CBRS, and Wireless Networking for Airside Autonomous Operations

DFW Airport 5G --- CBRS Deployment at Scale
Changi Airport --- Private 5G for Autonomous Tractors
CBRS 2.0 --- Regulatory Evolution and Airport Implications
Private 5G vs Public 5G --- Why Airports Need Dedicated Networks
URLLC --- Ultra-Reliable Low-Latency Communication
Network Architecture for Autonomous GSE
WiFi 6/6E as Alternative
Mesh Networking Between Vehicles
Bandwidth Requirements
Redundancy and Failover
Airport RF Environment
Cost Model
Vendors

1. DFW Airport 5G --- CBRS Deployment at Scale

Overview

Dallas Fort Worth International Airport (DFW) represents the most significant CBRS-based private 5G deployment at any US airport. In 2023, DFW signed a five-year, $10 million contract with AT&T to deploy a comprehensive wireless platform combining public WiFi upgrades with a private 5G network across its 27-square-mile campus --- comparable in size to Manhattan.

Infrastructure Deployed

Component	Details
CBRS transmission sites	~33 sites (Nokia equipment)
WiFi access points	~200 new Cisco APs + updates to ~800 existing hotspots
Coverage area	27 square miles, indoor and outdoor
Deployment timeline	Outdoor deployment completed in under 12 months
Network ownership	DFW owns the core; runs it inside the airport
Spectrum	CBRS General Authorized Access (GAA) --- no spectrum licensing fees

Network Architecture

DFW takes a converged access approach, routing both private 5G CBRS traffic and WiFi traffic through a single management platform. This unified architecture simplifies operations and reduces the management overhead of running two parallel wireless networks.

Nokia supplies the CBRS radio access network (33 transmission sites)
Cisco supplies the WiFi access points and the converged management platform
AT&T provides the integration, deployment, and managed services

What It Enables

DFW completed three comprehensive proofs of concept (PoCs) in ramp, cargo, and terminal services before committing to the full deployment:

Concession monitoring: Tracking whether 160+ concessionaires are open/closed for real-time passenger displays
Conveyance tracking: Monitoring 180 escalators and moving walkways for immediate breakdown alerts
Remote camera connectivity: Solar-powered cameras in remote locations connected without fiber
Autonomous shuttle testing: Tests for autonomous shuttles in parking facilities
Cargo asset tracking: Evaluating CBRS for tracking cargo assets across the airfield
Security and baggage handling: Enhanced monitoring of passenger traffic, security systems, and baggage
Digital twins: Building digital representations of airport operations

Performance Metrics

Transaction speeds: 50--70% faster than public cellular networks in pilot testing
Latency: "Much lower latency than what we would see using public cellular"
Cost advantage: Under $1,000 per field router vs. $50,000+ per fiber endpoint for remote connectivity
Security cameras: 40+ outdoor cameras operate solely on the private wireless connection

Key Lessons

Indoor deployment proved more complex than outdoor, requiring extensive RF mapping and additional hardware
Device compatibility varies; older equipment may require adapters for CBRS connectivity
Airport leadership predicts CBRS will evolve from optional technology to "critical infrastructure" for modern airport operations

Sources

2. Changi Airport --- Private 5G for Autonomous Tractors

5G Aviation Testbed

Singapore's Changi Airport launched a 5G Aviation Testbed in March 2023 at Terminal 3's airside, operated by Singtel in partnership with the Civil Aviation Authority of Singapore (CAAS), Changi Airport Group (CAG), Singapore Airlines, and IMDA. This two-year testbed enables companies operating airside to leverage 5G's high bandwidth, high-speed connectivity, and ultra-low latency.

Key network details:

Provider: Singtel (Singapore's primary telecommunications operator)
Coverage: Terminal 3 airside initially; plans to extend 5G to public areas in all terminals
Upgrades: ~4,000 Singtel 4G corporate mobile lines at the airside received complimentary 5G upgrades
Safety measures: CAAS established restrictions on transmission power and antenna tilt angle to ensure flight operation safety

Autonomous Tractor Fleet

In January 2026, Changi deployed its first fleet of fully driverless autonomous tractors for airside baggage operations, built by UISEE (a Chinese autonomous driving technology company):

Parameter	Specification
Vehicles in operation	2 (initial), expanding to 8 in 2026, 24 by 2027
Route	7 km between Terminal 1 and Terminal 4 baggage handling areas
Capacity	Up to 4 baggage containers, ~10 tonnes combined weight
Sensors	10+ per vehicle (LiDAR, cameras, RTK positioning, inertial navigation)
Detailed sensor suite (from other UISEE deployments)	4 LiDAR sensors, 6 HD cameras, RTK high-precision positioning, inertial navigation system
Testing	5,000+ test trips over nearly one year of trials
Safety record	20,000+ km of accident-free operation
Conditions	Day, night, and rain operations
Certifications	ISO 21434 cybersecurity, ISO 27001 information security, Singapore TR68 compliance

5G's Role in Autonomous Operations

The 5G testbed specifically enables:

Real-time teleoperation: High-definition video streams with low latency and high transmission stability for remote monitoring of AV operations
Continuous monitoring: Operators can supervise AV operations remotely in real-time
Secure flight data transfer: Singapore Airlines uses 5G to transmit critical flight data (weather, airport information) to aircraft, replacing fiber optic cables
Remote aircraft inspection: Advanced video analytics with AI for predicting aircraft turnaround times

Future Expansion

6 additional autonomous tractors deploying on a different route between Terminal 2 and aircraft stands (CAG-SATS collaboration)
Total fleet expanding to 24 by 2027
Future applications include autonomous towing of cargo and equipment (not just baggage)
UISEE also piloting at Hamad International Airport (Qatar) and Beijing Daxing International Airport

Sources

3. CBRS 2.0 --- Regulatory Evolution and Airport Implications

What is CBRS?

The Citizens Broadband Radio Service operates in 150 MHz of spectrum in the 3.5 GHz band (3550--3700 MHz). The FCC established a three-tiered access framework managed by a Spectrum Access System (SAS):

Tier	Name	Details
1	Incumbent Access	Federal users (primarily US Navy radar), Fixed Satellite Service earth stations. Receive full protection from interference.
2	Priority Access (PAL)	Licensed on a county-by-county basis via auction. 10 MHz channels in 3550--3650 MHz. Must protect Tier 1, protected from Tier 3.
3	General Authorized Access (GAA)	License-by-rule, open to all. Can operate across full 3550--3700 MHz band. No interference protection. Free to use.

The SAS is a cloud-based automated frequency coordinator that manages spectrum assignments and transmit power to prevent harmful interference to higher-priority users. Approved SAS operators include Federated Wireless, Google, CommScope, Amdocs, and Key Bridge Wireless.

Environmental Sensing Capability (ESC) sensors detect transmissions from Department of Defense radar systems and relay that information to the SAS for real-time spectrum management.

Key CBRS 2.0 Changes (2024--2025)

Dynamic Protection Area (DPA) Reduction

In June 2024, the FCC, NTIA, and US Navy collaborated to reduce the size of Dynamic Protection Areas --- zones along coastlines and around federal facilities where Navy radar can displace commercial CBRS users. The changes:

Modified the aggregate interference model used in the 3.5 GHz band
Expanded unencumbered CBRS coverage to approximately 72 million additional Americans
Total coverage now reaches roughly 240 million people nationwide
Affected states include Texas, Pennsylvania, North Carolina, Georgia, and Arizona
Airport implication: Coastal and military-adjacent airports gain more reliable CBRS access with fewer potential interruptions from incumbent users

NPRM for Further Changes (August 2024)

The FCC released a Notice of Proposed Rulemaking seeking comment on:

Higher power CBRS devices: Whether to add one or more classes of higher-power CBSDs (Citizens Broadband Service Devices)
Alignment with 3GPP standards: Whether to align end-user device power levels with international 3GPP specifications
Multiband device approval: In May 2025, Ericsson and Samsung received conditional waivers to manufacture devices operating in both the 3.5 GHz CBRS band and the 3.7 GHz C-Band --- enabling more flexible, higher-performance equipment
Out-of-band emissions: Whether to align CBRS base station emission limits with those adopted in the 3.7 GHz service
ESC sensor modifications: Changes to Environmental Sensing Capability requirements

Controversy: Spectrum Reallocation Proposal

In 2025, the DOD contemplated a proposal (supported by AT&T) to move low-power CBRS users to the lower 3 GHz band to share with DOD users, freeing 3.55--3.7 GHz spectrum for exclusive, high-power 5G mobile use and auctioning it to the highest bidder. This remains contentious --- smaller operators and cable companies oppose higher power levels that could undermine shared spectrum access.

Implications for Airport Deployments

More reliable coastal airport coverage: Reduced DPAs mean airports near coastlines (many major hubs) face fewer CBRS interruptions
Higher power = better outdoor coverage: If approved, higher power levels would improve airfield coverage with fewer access points
Multiband devices: Equipment operating across both CBRS and C-Band provides fallback spectrum options
Regulatory uncertainty: The potential reallocation of CBRS spectrum to exclusive auction use poses a risk to airports that have invested in GAA-based private networks

Sources

4. Private 5G vs Public 5G --- Why Airports Need Dedicated Networks

The Fundamental Problem with Public 5G for Autonomous Operations

Public 5G networks are designed for consumer traffic --- shared bandwidth, best-effort delivery, and coverage optimized for revenue-generating areas. Autonomous vehicles at airports require the opposite: dedicated bandwidth, guaranteed latency, and coverage in non-public areas like airfields, taxiways, and remote cargo areas.

Comparison Matrix

Attribute	Private 5G	Public 5G
Spectrum	Dedicated (CBRS, licensed, or leased)	Shared with all subscribers
Latency	Sub-10ms achievable, URLLC capable	20--50ms typical, variable under load
Reliability	99.999% achievable with URLLC	99--99.9% typical
Coverage control	Deployed exactly where needed (airfield, ramp, taxiways)	Carrier determines coverage; airfield gaps common
Bandwidth guarantee	Dedicated capacity; no contention	Shared; degrades with user density
Network slicing	Full control over slice configuration	Carrier-managed; limited enterprise control
Data sovereignty	All data stays on-premise	Data traverses carrier infrastructure
Security	SIM-based authentication; isolated network	Shared infrastructure; potential attack surface
Handoff control	Tuned for vehicle speeds and routes	Optimized for pedestrian/vehicle patterns
SLA	Self-managed or contractual with full control	Carrier SLA; limited recourse for airfield coverage
Customization	QoS policies tailored to AV requirements	One-size-fits-all policies
Cost model	CapEx + OpEx; owned infrastructure	Per-device/per-GB; recurring carrier fees

Why Private Networks Win for Autonomous Airside Operations

Deterministic QoS: Private 5G uses centralized scheduling to allocate dedicated wireless access to each client. Public networks use contention-based access where devices "fight" for bandwidth. For autonomous vehicles making safety-critical decisions, deterministic access is non-negotiable.
Coverage where it matters: Public carriers optimize coverage for terminals and parking (passenger revenue areas). Airfields, taxiways, remote stands, and cargo areas --- precisely where autonomous GSE operates --- are often underserved. Private networks deploy coverage exactly where operational needs exist.
Interference protection: Private 5G on CBRS operates in coordinated spectrum with SAS-managed interference protection. The network is shielded from the consumer traffic spikes that cause public 5G degradation during peak travel hours.
Edge computing co-location: Private networks allow MEC servers to be deployed on-premise, keeping all vehicle control data within the airport perimeter. Public networks route data through carrier data centers, adding latency.
Regulatory and security: Airports are security-sensitive environments. Private networks ensure that autonomous vehicle telemetry, video feeds, and control commands never traverse public internet infrastructure.

Real-World Validation

DFW Airport: Deployed private CBRS specifically because public cellular couldn't deliver the latency and reliability needed for operational applications --- achieving 50--70% faster transaction speeds than public cellular.
Lufthansa LAX Cargo: Private 5G at their cargo facility delivered a 60% reduction in processing time per item by eliminating the latency spikes and dropped connections from WiFi and public cellular.
Stanley Robotics (Lyon Airport): Private 5G enables management of up to 100 autonomous parking robots simultaneously --- impossible on shared public infrastructure.
Port of Liverpool: Private 5G achieved a tenfold performance boost compared to legacy WiFi systems across 100 acres of port operations.

Sources

5. URLLC --- Ultra-Reliable Low-Latency Communication

Specification

URLLC is a 5G NR service category defined by 3GPP, targeting mission-critical communications:

Parameter	Target	Notes
User-plane latency	1 ms (one-way)	Radio interface only; end-to-end is higher
Reliability	99.999% (five nines)	Packet delivery within latency bound
Availability	99.999%	Network uptime
Packet size	32 bytes typical	Small control packets
Jitter	< 1 ms	Consistent latency critical for control loops

How URLLC is Achieved

URLLC relies on several enabling technologies:

Mini-slot scheduling: Shorter transmission time intervals (as low as 2 OFDM symbols vs. 14 for standard slots) reduce waiting time
Grant-free transmission: Devices transmit immediately without waiting for scheduling grants
Packet duplication: Same data sent over multiple paths for redundancy
Network slicing: Dedicated virtual network with guaranteed resources isolated from other traffic
Edge computing (MEC): Processing at the network edge eliminates backhaul latency
Robust error correction: Advanced coding schemes minimize retransmissions

Practical Reality vs. Theory

The 1ms target is an air-interface specification, not an end-to-end guarantee. In practice:

Measurement	Achievable Today	Context
Radio access RTT	< 4 ms	5G NR air interface
End-to-end (with MEC)	5--10 ms	Edge server co-located with base station
End-to-end (cloud)	20--50 ms	Data traverses backhaul to remote server
Application-level	40--100 ms	Includes processing, encoding, rendering

For autonomous airport GSE operating at 15--25 km/h (typical airside speeds), end-to-end latency of 10--20ms is sufficient for:

Teleoperation control commands (target: < 20 ms downlink)
Collision avoidance alerts (target: < 10 ms)
Fleet coordination updates (target: < 50 ms)
Video streaming for remote monitoring (target: < 100 ms glass-to-glass)

URLLC vs. eMBB Trade-offs

Factor	URLLC	eMBB
Latency	1 ms target	4--10 ms typical
Throughput	Lower (focused on small, time-critical packets)	Higher (optimized for data volume)
Reliability	99.999%	99--99.9%
Use case	Vehicle control, emergency stops	Video streaming, sensor data upload

Practical approach for autonomous GSE: Use URLLC slice for control commands and safety-critical messaging, and eMBB slice for video streaming and bulk sensor data upload. Network slicing makes this dual-slice architecture possible on a single physical network.

Sources

6. Network Architecture for Autonomous GSE

Reference Architecture

                                    CLOUD TIER
                        (Model training, analytics, reporting)
                                      |
                                  WAN/Internet
                                      |
                              ================
                              |  AIRPORT CORE |
                              |  DATA CENTER  |
                              ================
                              |              |
                    +---------+--------+     |
                    |                  |     |
              +-----------+    +-----------+ |
              | MEC Node  |    | MEC Node  | |
              | (Ramp A)  |    | (Ramp B)  | |
              +-----------+    +-----------+ |
                    |                |       |
              +-----+-----+   +-----+-----+ |
              |     |     |   |     |     |  |
             gNB  gNB   gNB gNB  gNB   gNB  |
              |     |     |   |     |     |  |
           [AV1] [AV2] [AV3] [AV4] [AV5] [AV6]
                                                |
                                         +------------+
                                         | Teleop     |
                                         | Control    |
                                         | Center     |
                                         +------------+

Tier 1: Vehicle Edge (On-Vehicle)

Each autonomous GSE vehicle contains:

Primary compute: Onboard AI inference for perception, planning, and control (operates independently of network)
5G modem/router: Dual-modem cellular router (e.g., Peplink MBX) with multi-SIM capability
Sensor suite: LiDAR, cameras, radar, ultrasonic, GNSS/RTK
V2X module (optional): C-V2X PC5 sidelink for direct vehicle-to-vehicle communication
Local buffer: Stores sensor data and telemetry during connectivity gaps

Tier 2: Multi-Access Edge Computing (MEC)

MEC nodes are deployed at the airport, co-located with or adjacent to 5G base stations:

Placement: Typically in equipment rooms at each ramp area or terminal zone, within 1--2 hops of the gNBs (5G base stations)
Functions:
- User Plane Function (UPF): Dedicated UPF per network slice, processing vehicle data locally without backhauling to cloud
- Teleoperation video processing: Receives and re-encodes vehicle camera streams for remote operators
- Fleet orchestration: Real-time coordination of vehicle routes, assignments, and conflict resolution
- Object detection offload: Optional computation offload for complex perception tasks
- Geofence enforcement: Monitors vehicle positions against operational boundaries
Latency benefit: Processing at MEC reduces round-trip to < 10 ms vs. 50+ ms for cloud routing

Tier 3: Airport Core Data Center

5G Core Network: Centralized control plane functions (AMF, SMF, PCF, UDM)
Network management: SAS integration for CBRS spectrum management, network monitoring
Data aggregation: Collects telemetry from all MEC nodes for fleet analytics
Model deployment: Distributes updated ML models to vehicles via MEC nodes
Integration hub: Connects to airport operational databases (A-CDM, AODB)

Tier 4: Cloud

Model training: Large-scale ML training on collected operational data
Long-term analytics: Historical trend analysis, performance reporting
Disaster recovery: Backup for airport core systems

Data Flow for Typical Operations

Data Type	Direction	Latency Req.	Path
Control commands (teleop)	Downlink	< 20 ms	Teleop Center -> MEC -> gNB -> Vehicle
Vehicle telemetry	Uplink	< 50 ms	Vehicle -> gNB -> MEC -> Fleet Manager
Camera streams (teleop)	Uplink	< 100 ms	Vehicle -> gNB -> MEC -> Teleop Center
Collision avoidance	Both	< 10 ms	Vehicle <-> gNB <-> MEC
Sensor data (recording)	Uplink	Best effort	Vehicle -> gNB -> MEC -> Core -> Cloud
Model updates	Downlink	Best effort	Cloud -> Core -> MEC -> gNB -> Vehicle
Fleet coordination	Both	< 50 ms	Vehicle <-> MEC <-> Fleet Manager

Sources

7. WiFi 6/6E as Alternative

Technical Comparison: WiFi 6/6E vs. Private 5G for Airside Operations

Parameter	WiFi 6/6E	Private 5G (CBRS)
Spectrum	Unlicensed (2.4, 5, 6 GHz)	Licensed/shared (3.5 GHz CBRS)
Max throughput (theoretical)	9.6 Gbps (WiFi 6)	Up to 20 Gbps
Latency	5--20 ms typical	1--10 ms with URLLC
Reliability	Best-effort; no guaranteed QoS	Deterministic; 99.999% URLLC
Indoor range per AP	30--50 meters	100--200 meters
Outdoor range per AP	100--150 meters	300--1,000 meters (macro)
Coverage efficiency	1/5th to 1/30th of 5G coverage per AP	3--4x indoor, up to 10x outdoor vs. WiFi
Handoff	Client-driven; 50--200 ms roaming gaps	Network-controlled; seamless < 0ms
Interference	Unlicensed band; contention from neighboring networks	Coordinated spectrum; SAS-managed protection
QoS	Best-effort; clients contend for access	Centralized scheduling; dedicated access
Security	WPA3; password/certificate-based	SIM-based authentication
Device density	Degrades above ~50 clients per AP	Up to 1 million devices per km^2
Power per AP	15--30 watts	Up to 50 watts per 10 MHz

Where WiFi 6/6E Falls Short for Autonomous Vehicles

Roaming gaps: WiFi handoff between access points is client-driven, causing 50--200 ms connectivity gaps. For a vehicle traveling at 25 km/h, this means potential loss of control signal every time the vehicle crosses an AP boundary. 5G handoffs are network-controlled and seamless.
No QoS guarantees: WiFi is a contention-based protocol. During peak airport operations (airline shift changes, heavy passenger traffic), WiFi performance degrades unpredictably. 5G provides scheduled, deterministic access.
Outdoor coverage: Covering a 27-square-mile airport campus with WiFi would require thousands of access points. Private 5G achieves the same coverage with far fewer base stations due to superior range (a single outdoor macrocell covers up to 1 square mile).
Interference susceptibility: The unlicensed WiFi bands (especially 2.4 and 5 GHz) face interference from passenger devices, airline operations systems, and neighboring networks. The 6 GHz band (WiFi 6E) requires AFC coordination and has restrictions on outdoor use.

Where WiFi 6/6E Makes Sense

Terminal and indoor operations: Passenger WiFi, staff devices, IoT sensors in controlled indoor environments
Supplementary connectivity: As a backup/secondary network for autonomous vehicles (not primary)
Low-mobility applications: Fixed or slow-moving equipment in hangars, workshops, and maintenance areas
Cost-sensitive deployments: When 5G infrastructure CapEx is not yet justified

WiFi 6E Specific Considerations

WiFi 6E adds the 6 GHz band (U-NII 5, 6, 7, 8), providing up to 1,200 MHz of additional spectrum. However:

Outdoor use requires AFC: Standard-power outdoor APs must operate under Automated Frequency Coordination, adding complexity
Shorter range at 6 GHz: Higher frequency = shorter propagation distance and more susceptibility to obstacles
Coverage continuity: Best practice is to build contiguous 6 GHz coverage rather than islands, which is challenging across a sprawling airfield

Recommendation for Airport AV Deployments

Primary network: Private 5G (CBRS or licensed spectrum) for all autonomous vehicle operations. Secondary/backup: WiFi 6E in areas with dense AP coverage (terminals, ramp areas near buildings). Tertiary: Public cellular as emergency fallback.

Sources

8. Mesh Networking Between Vehicles

V2X Communication Technologies

Vehicle-to-Everything (V2X) communication enables autonomous vehicles to interact with other vehicles, infrastructure, pedestrians, and the network. Two competing technologies exist:

DSRC (Dedicated Short-Range Communication)

Parameter	Specification
Standard	IEEE 802.11p
Frequency	5.9 GHz
Range	300--1,000 meters
Latency	< 5 ms
Data rate	3--27 Mbps
Status	Mature but declining adoption
Limitation	Small packet sizes; not suited for high-volume data exchange needed by autonomous vehicles

C-V2X (Cellular Vehicle-to-Everything)

Parameter	Specification
Standard	3GPP Release 14+ (LTE-V2X), Release 16+ (NR-V2X)
Frequency	5.9 GHz (PC5 sidelink), plus cellular bands (Uu)
Range	Up to 1,500+ meters (PC5 direct)
Latency	< 5 ms (PC5 direct)
Data rate	Up to 1 Gbps (NR-V2X)
Communication modes	PC5 (direct, no network needed) + Uu (via cellular network)
Status	Actively deployed; FCC approved for connected vehicles in 2023

C-V2X Communication Modes

PC5 Sidelink (Direct): Vehicles communicate directly without base station involvement. Works even without cellular coverage. Two sub-modes:
- Mode 1: Base station allocates radio resources for sidelink communication
- Mode 2: Vehicles autonomously select resources (works out-of-coverage)
Uu Interface (Network): Communication routed through the cellular network for longer range and cloud integration.

When is V2V/Mesh Networking Needed for Airport GSE?

Scenario	V2V Needed?	Why
Collision avoidance at intersections	Yes	Sub-10ms vehicle-to-vehicle alerts for crossing paths on taxiways
Platooning (convoy of baggage tractors)	Yes	Tight formation requires < 5ms inter-vehicle latency
Coverage gaps (remote stands, construction zones)	Yes	PC5 sidelink operates without network infrastructure
Cooperative perception	Optional	Sharing sensor data between vehicles to extend situational awareness
Fleet coordination	No	Better handled by centralized MEC-based fleet manager
Teleoperation	No	Requires network backhaul to control center

Airport-Specific Considerations

Controlled environment advantage: Unlike public roads, airports control all vehicles on the airfield. This makes centralized fleet management more practical than distributed mesh networking.
Low speeds: At 15--25 km/h, the urgency of V2V collision avoidance is reduced compared to highway scenarios, though still valuable.
Complement, not replace: V2V/C-V2X is best deployed as a supplement to the private 5G network, providing a safety-critical backup communication path that works independently of network infrastructure.

Recommendation

Deploy C-V2X with PC5 sidelink on all autonomous GSE for:

Safety-critical collision avoidance (independent of network)
Operations in temporary coverage gaps
Cooperative awareness between nearby vehicles

Use the private 5G network (Uu interface) for:

All non-safety-critical communication
Teleoperation
Fleet management
Data upload

Sources

9. Bandwidth Requirements

Per-Vehicle Bandwidth Budget

Based on research from teleoperation field trials and academic studies:

Uplink (Vehicle to Network)

Data Stream	Bandwidth	Notes
Single camera (1080p/30fps, H.265)	8 Mbps	Per camera; H.264 requires ~12 Mbps
4-camera teleoperation suite	24--32 Mbps	Front, rear, left, right cameras
6-camera panoramic coverage	36--48 Mbps	Full surround view for teleop
LiDAR (64-beam, raw)	277 Mbps	Impractical to stream raw
LiDAR (64-beam, voxel downsampled)	~51 Mbps	With 0.5m^3 voxel downsampling
LiDAR (128-beam, raw)	307 Mbps	Even less practical
Vehicle telemetry (position, speed, status)	0.1--0.5 Mbps	Low bandwidth
Radar data	1--5 Mbps	Compact data format
Total per vehicle (practical teleop)	30--50 Mbps uplink	4 cameras + compressed LiDAR + telemetry
Total per vehicle (full sensor upload)	100--350 Mbps uplink	All raw sensor data; for logging, not real-time

Downlink (Network to Vehicle)

Data Stream	Bandwidth	Notes
Control commands (steering, throttle, brake)	0.1--0.3 Mbps	Very small packets
Fleet coordination messages	0.1--0.5 Mbps	Route updates, assignments
Map/model updates	1--10 Mbps (burst)	Periodic, not continuous
OTA software updates	10--100 Mbps (burst)	Scheduled during idle periods
Total per vehicle (operational)	0.5--1 Mbps downlink	Continuous operation

Daily Data Volume Estimates

Metric	Estimate
Raw sensor data per hour (all sensors)	~4 TB
Compressed operational data per hour	15--25 GB
Teleop-mode data per hour (4 cameras)	10--15 GB
Autonomous-mode uplink per hour	1--5 GB (telemetry + periodic snapshots)
8-hour shift per vehicle	8--120 GB depending on mode

Fleet-Level Bandwidth Planning

Fleet Size	Teleop Uplink (worst case)	Autonomous Uplink (typical)	Notes
10 vehicles	300--500 Mbps	10--50 Mbps	Small initial fleet
25 vehicles	750 Mbps -- 1.25 Gbps	25--125 Mbps	Medium fleet (e.g., Changi 2027 target)
50 vehicles	1.5--2.5 Gbps	50--250 Mbps	Large airport fleet
100 vehicles	3--5 Gbps	100--500 Mbps	Full-scale deployment

Critical note: Not all vehicles require simultaneous teleoperation. A typical ratio is 1 teleoperator per 10--100 vehicles, with most vehicles operating fully autonomously. Realistic simultaneous teleop demand is 5--15% of fleet at any time.

5G Network Capacity Considerations

Standard 5G TDD frame structures require more than 60 MHz of spectrum to support teleoperation of even a single vehicle per cell
The TDD uplink/downlink asymmetry in commercial 5G is a key challenge: typical configurations allocate 70--80% of time slots to downlink, but teleoperation needs heavy uplink
Private 5G allows custom TDD configurations optimized for uplink-heavy autonomous vehicle traffic
Practical 5G uplink throughput in field trials: ~77.7 Mbps per cell (commercial network)

Practical Research Data (2025 Field Trial)

A 6-month field trial teleoperating autonomous vehicles over commercial 5G (1,748 km of driving):

Single camera (1080p/30fps/H.265): Median per-frame network delay of 73.5 ms (exceeds 45 ms target)
Only 0.487% of frames exceeded the 100 ms application-level deadline (acceptable)
Multiple cameras: 45--48% of frames violated the 45 ms network-level deadline (problematic)
Command/control: 64.29% of messages met application requirements; median delay 17.29 ms
Raw LiDAR streaming: Median delay of 2--6 seconds (impractical without heavy compression)

Key takeaway: Commercial 5G can support single-camera teleoperation but struggles with multi-camera and LiDAR streaming. Private 5G with custom uplink-optimized TDD configurations and MEC is essential for production-grade autonomous vehicle teleoperation.

Sources

10. Redundancy and Failover

The Fundamental Requirement

For autonomous vehicles, even one second of downtime can lead to disastrous results. The connectivity architecture must ensure that no single point of failure can compromise vehicle safety.

Multi-Layer Redundancy Architecture

Layer 1: Onboard Autonomy (no network required)
    |
Layer 2: C-V2X PC5 Sidelink (direct vehicle-to-vehicle/infrastructure)
    |
Layer 3: Primary Private 5G (CBRS / licensed spectrum)
    |
Layer 4: Secondary Cellular (public LTE/5G via different carrier)
    |
Layer 5: Tertiary Backup (WiFi / satellite in extreme cases)

What Happens When Connectivity Drops

Behavior Hierarchy

Connectivity State	Vehicle Behavior
Full connectivity	Normal autonomous operation with teleoperation available
Degraded connectivity (high latency, packet loss)	Continue autonomous operation; alert fleet manager; buffer telemetry
Teleoperation link lost	Continue autonomous operation if safe; queue for teleop reconnection
All network lost, V2X available	Continue with V2X-based cooperative awareness; reduced speed
All external comms lost	Execute safe stop or continue on pre-planned route (depending on ODD)
Emergency	Immediate safe stop at current position or nearest safe harbor

Safe Stop Protocol

The vehicle must be capable of independently:

Detecting connectivity loss within 1--2 seconds
Assessing whether continued operation is safe based on onboard perception
If continuing is safe: proceeding at reduced speed to the nearest designated safe stop point
If continuing is not safe: executing an immediate controlled stop, activating hazard indicators
Maintaining position awareness via GNSS and broadcasting location via any available channel

Hardware Redundancy Approaches

Dual-Modem Router (Recommended for Airport AV)

Feature	Specification
Architecture	Two independent radio modules, active simultaneously
Failover speed	Sub-second ("hot standby")
SIM configuration	Dual SIM (different carriers/spectrum bands)
Bandwidth bonding	Combines both connections for aggregate throughput
Technology	SpeedFusion or similar (Peplink, Cradlepoint)
Cost	$1,000--$3,000 per vehicle

Comparison of Failover Approaches

Approach	Failover Time	Pros	Cons
Dual-Modem (hot standby)	< 1 second	Seamless; backup always connected	Higher cost; more power draw
Dual-SIM, single modem	30--90 seconds	Lower cost	Unacceptable gap for AV operations
Network bonding (active-active)	0 seconds	Both links always active; aggregate bandwidth	Highest cost; most complex
WiFi + cellular	2--5 seconds	Low incremental cost	WiFi coverage gaps; inconsistent

Network Bonding for Teleoperation

Leading AV companies implement triple-redundant communication networks:

Primary: High-bandwidth 5G connection for normal operations
Secondary: LTE/4G fallback with optimized compression
Tertiary: Satellite link for emergency connectivity in cellular dead zones

SpeedFusion technology (Peplink) enables:

Hot Failover: Seamless transfer to alternative WAN; users may not realize a failover occurred
Bandwidth Bonding: Combines multiple connections into one, using the combined bandwidth of multiple WANs
Forward Error Correction: Cross-channel FEC recovers lost packets without retransmission
Adaptive Buffering: Smooths out connectivity variations

Buffering Strategy

Data Type	Buffer Duration	Behavior During Connectivity Loss
Control commands	0 (real-time only)	Vehicle reverts to onboard autonomy
Vehicle telemetry	30--60 seconds	Cached locally; transmitted when connection restores
Camera streams	5--10 seconds	Circular buffer; teleoperator sees frozen frame then reconnects
Sensor recordings	Hours	Written to onboard storage; uploaded later
Fleet coordination	10--30 seconds	Vehicle continues last known route plan

Sources

11. Airport RF Environment

The Challenge

Airports are among the most complex RF environments in existence. Autonomous vehicle connectivity must coexist with safety-critical aviation systems while dealing with extreme device density and diverse interference sources.

Aviation Systems and Frequency Bands

System	Frequency Band	Function	Proximity to CBRS?
Radar Altimeters	4.2--4.4 GHz	Aircraft altitude measurement during approach/landing	Close to C-Band 5G (3.7--3.98 GHz); NOT close to CBRS (3.5 GHz)
ILS (Instrument Landing System)	108--112 MHz (localizer), 329--335 MHz (glideslope)	Precision approach guidance	No conflict
Airport Surveillance Radar (ASR)	2.7--2.9 GHz	Aircraft tracking in terminal area	Below CBRS; minimal concern
Surface Movement Radar	9.0--9.5 GHz (X-band)	Ground vehicle/aircraft tracking on surface	No conflict
ADS-B	1090 MHz / 978 MHz	Aircraft position broadcasting	No conflict
VHF Communications	118--137 MHz	Air traffic control voice	No conflict
DME (Distance Measuring Equipment)	960--1215 MHz	Navigation distance measurement	No conflict
GNSS (GPS)	1176, 1227, 1575 MHz	Satellite navigation	No conflict with CBRS; susceptible to broadband interference
Military Radar (Navy)	3550--3700 MHz	Ship-based radar systems	Direct overlap with CBRS --- managed by SAS/DPA

CBRS (3.5 GHz) vs. C-Band (3.7 GHz) Safety Distinction

Critical clarification: The widely publicized 5G-aviation interference controversy involves the C-Band (3.7--3.98 GHz), not CBRS (3.55--3.7 GHz). The C-Band operates much closer to radar altimeter frequencies (4.2--4.4 GHz) and at much higher power levels.

CBRS operates at lower frequencies and lower power levels than C-Band, providing greater spectral separation from radar altimeters:

Parameter	CBRS	C-Band 5G
Frequency	3550--3700 MHz	3700--3980 MHz
Gap to radar altimeters	500+ MHz	220 MHz (minimum)
Max power (outdoor)	~50 dBm/10 MHz EIRP (Category B)	Much higher (carrier macro)
Interference risk to altimeters	Very low	Documented concern; FAA mitigations required

Airport-Specific Interference Considerations

GNSS/RTK vulnerability: Autonomous vehicles relying on RTK-GNSS for centimeter-level positioning can be disrupted by broadband RF emissions. Mitigation: Use filtered GNSS receivers and multi-constellation (GPS + GLONASS + Galileo + BeiDou) redundancy.
Radar side-lobe interference: Airport surveillance radar operates at 2.7--2.9 GHz. While not directly overlapping with CBRS, high-power radar side lobes can desensitize nearby receivers. Mitigation: Physical separation and RF filtering.
Passenger device density: During peak hours, thousands of active cellular and WiFi devices create a high noise floor. Private 5G on dedicated CBRS spectrum is isolated from this by design.
Jet engine EMI: Jet engines and APUs generate broadband electromagnetic interference. Vehicles operating near running aircraft need hardened radio equipment.
Reflections and multipath: Large metal aircraft surfaces, hangars, and terminal buildings create severe multipath propagation. 5G NR's beam management and MIMO capabilities handle multipath better than WiFi.

Regulatory Protections

CAAS (Singapore): Established restrictions on transmission power and antenna tilt angle for 5G at Changi Airport to ensure flight operation safety
FAA (US): Developed interference tolerance requirements for radio altimeters; phased retrofit of RF filters on susceptible aircraft
FCC/NTIA: SAS manages CBRS to protect federal incumbents including military radar near airports
ICAO: Working on global guidance for 5G/aviation coexistence

Sources

12. Cost Model

Infrastructure Cost Breakdown

Capital Expenditure (CapEx)

Component	Cost Range	Notes
5G Base Station (macro)	$100,000--$200,000 per site	Outdoor, high-power; covers up to 1 sq mi
Small cells (indoor/outdoor)	$10,000--$50,000 per unit	Depends on location and mounting infrastructure
5G Core Network	$100,000--$500,000	On-premise core; Nokia DAC, Ericsson EP5G, or equivalent
MEC servers	$50,000--$150,000 per node	Edge compute for each ramp zone
Backhaul (fiber)	$50,000+ per endpoint	Or leverage existing airport fiber plant
Field routers (vehicles)	$1,000--$3,000 per vehicle	Dual-modem cellular routers (Peplink, Cradlepoint)
SAS subscription	$5,000--$15,000/year	Spectrum Access System for CBRS coordination
CBRS spectrum (GAA)	Free	No licensing fee for General Authorized Access
CBRS spectrum (PAL)	Varies by county	Priority Access License via auction; provides interference protection
Site surveys and RF design	$25,000--$100,000	One-time; scales with campus size
Installation and integration	$100,000--$500,000	Professional services; depends on complexity

Deployment Scenarios

Scenario	Coverage	Infrastructure	Estimated CapEx
Pilot (single ramp area)	~1 sq mi	3--5 small cells + 1 core	$200K--$500K
Medium (terminal complex)	~5 sq mi	10--15 small cells + 2--3 MEC nodes	$1M--$3M
Full airport (DFW-scale)	~27 sq mi	30+ sites + full core + MEC	$5M--$15M

Reference: DFW's contract with AT&T was $10 million over 5 years, covering both private 5G CBRS and WiFi infrastructure across 27 square miles.

Operating Expenditure (OpEx)

Component	Annual Cost	Notes
Network management	15--20% of CapEx	24/7 monitoring, troubleshooting
Software licenses	$50,000--$200,000	Core network, management platform, SAS
SIM management	$5--$15 per device/month	Provisioning, lifecycle management
Power and cooling	$20,000--$100,000	For MEC nodes and base stations
Backhaul/internet	$20,000--$100,000	Dedicated fiber or leased lines
Staff or managed services	$150,000--$500,000	Network engineers or outsourced to provider
Spectrum fees (if PAL)	Varies	Recurring if using Priority Access Licenses

Cost Comparison: Private 5G vs. Alternatives

Solution	CapEx per sq mi	Annual OpEx	Suitability for AV
Private 5G (CBRS)	$150K--$400K	$50K--$150K	Excellent
WiFi 6/6E	$300K--$800K (more APs needed)	$100K--$300K	Poor for outdoor AV
Public cellular	Near zero	$50--$100/device/month	Insufficient for AV operations
Fiber to each location	$50K+ per endpoint	$10K--$50K	Not feasible for moving vehicles

Key insight from DFW: CBRS field routers cost under $1,000 each vs. $50,000+ per fiber endpoint for remote connectivity --- a 50x cost advantage for connecting mobile and remote assets.

ROI Considerations

Most organizations achieve ROI within 12--24 months of private 5G deployment
Stanley Robotics (Lyon Airport): Private 5G enabled scaling from initial robots to management of 100 robots simultaneously, with 50% improvement in parking efficiency
Port of Liverpool: Tenfold performance boost vs. legacy WiFi, reducing operational costs by up to 50%
Automation paired with private 5G connectivity can boost ROI by up to 178% (Firecell analysis)

Sources

13. Vendors

Nokia

Product: Nokia Digital Automation Cloud (DAC) / Modular Private Wireless (MPW)

Attribute	Details
Offerings	4G/LTE and 5G private wireless with edge computing
Architecture	Radios + MX Industrial Edge (core + applications) + cloud management
Spectrum support	Licensed, CBRS, unlicensed
Edge platform	MX Industrial Edge --- runs core network and industrial applications
Deployment model	Plug-and-play; compact option for smaller sites
Airport deployments	DFW Airport (33 CBRS sites via Nokia radios), MCA Aviation partnership for US airports
Strengths	Deep industry expertise in mission-critical sectors (ports, airports, mining, utilities); carrier-grade reliability
Key feature	Deterministic QoS; centimeter-level indoor positioning; GNSS-enhanced outdoor tracking
Note (2025)	Nokia announced plans to restructure its private networking business, potentially selling its Enterprise Campus Edge unit. Mission-critical focus remains.

Ericsson

Product: Ericsson Private 5G (EP5G)

Attribute	Details
Offerings	4G and 5G private cellular networks as managed service
Architecture	Ericsson Radio System portfolio + enterprise-grade network controllers
Management	Cloud-based Ericsson Network Manager (ENM) + self-service portal
Deployment sizes	Small, Medium, Large, Extra Large
Airport deployments	Paris-Charles de Gaulle / Paris-Orly / Paris-Le Bourget (with Hub One / Air France); Purdue University Airport (with Saab, CBRS GAA)
2025-2026 expansion	Added 33 new radios to EP5G portfolio in 2025; hundreds of enterprise customers across multiple countries
Strengths	Largest global cellular infrastructure company; massive R&D; open APIs for IT/OT integration
Aviation partnership	Ericsson + Streamwide for MCx (mission-critical) communications at airports
Key feature	Connected Aviation vision: private mobile networks for baggage handling automation, connected workers, autonomous systems

Samsung

Product: Samsung Private 5G Network Solutions

Attribute	Details
Offerings	Compact, Standard, and Premium configurations
Architecture	Virtualized platform consolidating DU, CU, UPF, and virtual switch in a single server ("Network in a Server")
Compact option	All-in-one-box with RAN and compact core --- simplest deployment
Standard option	Multi-site configuration for mid-sized businesses
Premium option	Large-scale with high scalability
Key hardware	Compact Macro radio --- designed for dense environments including stadiums and airports
CBRS	Received FCC conditional waiver (May 2025) for multiband devices operating in both 3.5 GHz CBRS and 3.7 GHz C-Band
Strengths	End-to-end solution including baseband, radio, core, and management; URLLC support; carrier-grade capabilities in compact form factors

Qualcomm

Product: Qualcomm 5G RAN Platform for Small Cells (FSM series)

Attribute	Details
Role	Chipset provider --- powers small cells from multiple OEMs
Key platform	FSM200xx --- first 3GPP Release 16 5G Open RAN platform
Performance	Up to 8 Gbps with 1 GHz mmWave bandwidth; 200 MHz carrier bandwidth support
Process node	4nm --- energy-efficient enough for Power over Ethernet (PoE) deployment
Ecosystem	5G Private Network Partner Ecosystem Program with documented "blueprints"
Airport relevance	Designed for seamless connectivity in crowded environments (airports, venues, hospitals)
Partners using FSM	Airspan, Radisys, Globalstar, and multiple other small cell OEMs
Strengths	Open RAN leadership; power efficiency; broad ecosystem of partners; competitive cost for small cells

Other Notable Vendors

Vendor	Role	Airport Relevance
Celona	Private 5G LAN solutions	CBRS-based enterprise 5G with self-service management
Betacom	Private 5G managed service	Airport-focused private 5G deployments
Peplink	Vehicle connectivity routers	MBX 5G routers with SpeedFusion bonding/failover for AV teleoperation
Cradlepoint (Ericsson)	Enterprise routers	5G/LTE vehicle routers with NetCloud management
Federated Wireless	SAS provider	Spectrum Access System for CBRS coordination
Cisco	WiFi + converged management	WiFi APs + unified management (as deployed at DFW)
ORAXIO Telecom	Private 5G deployment	Partnered with Stanley Robotics for autonomous parking at Lyon Airport
Firecell	Private 5G platform	Network slicing, edge computing for ports and airports
HPE	Private 5G + edge compute	End-to-end private 5G with Aruba and edge infrastructure

Vendor Selection Framework for Airport AV Deployments

Criterion	Weight	Considerations
Airport/aviation experience	High	Proven deployments in regulated airport environments
CBRS expertise	High	SAS integration, GAA/PAL management, DPA handling
URLLC support	High	Network slicing, deterministic QoS for vehicle control
Edge computing integration	High	MEC platform for local processing of AV data
Managed service option	Medium	Airport IT teams may lack cellular expertise
Open RAN / interoperability	Medium	Avoid vendor lock-in; enable best-of-breed components
Vehicle router ecosystem	Medium	Compatibility with dual-modem vehicle connectivity hardware
Scalability	Medium	Path from pilot (5 vehicles) to full fleet (100+ vehicles)
Cost model	Medium	CapEx vs. OpEx; subscription vs. perpetual licensing

Sources

Summary: Connectivity Architecture Recommendation for Airport Autonomous Vehicles

Primary Network: Private 5G on CBRS

Deploy CBRS-based private 5G across the operational airfield
Use GAA tier initially (no spectrum licensing cost); acquire PAL licenses for critical areas if interference protection is needed
Place MEC nodes at each major ramp area for sub-10ms latency
Configure custom TDD frame structure optimized for uplink-heavy AV traffic
Implement URLLC network slice for control commands, eMBB slice for video/data

Vehicle Connectivity: Dual-Modem with Network Bonding

Each vehicle: dual-modem 5G router (e.g., Peplink MBX) with primary on private 5G, secondary on public LTE/5G
SpeedFusion or equivalent for hot failover and bandwidth bonding
C-V2X PC5 sidelink module for direct vehicle-to-vehicle safety communication

Failover Hierarchy

Onboard autonomy (always available, no network needed)
C-V2X PC5 sidelink (direct, network-independent)
Private 5G (CBRS)
Public cellular (secondary SIM)
WiFi (where available)
Safe stop (last resort)

Bandwidth Planning

Budget 30--50 Mbps uplink per vehicle in teleoperation mode
Plan for 5--15% simultaneous teleoperation ratio
For 25-vehicle fleet: provision ~200--375 Mbps aggregate uplink capacity
Ensure custom TDD ratio favoring uplink (e.g., 60/40 DL/UL instead of standard 80/20)

Cost Estimate for Mid-Size Airport Deployment

Item	Estimated Cost
Infrastructure (20 small cells + core + 3 MEC nodes)	$2M--$5M
Vehicle equipment (25 vehicles x $3K)	$75K
Installation and integration	$200K--$500K
Annual OpEx (managed service)	$400K--$800K
Total Year 1	$2.7M--$6.4M
5-Year TCO	$4.3M--$9.6M

Report compiled March 2026. Sources verified as of publication date.

SLAM Methods

Methods

Airport Connectivity Infrastructure for Autonomous Vehicles ​

Research Report: 5G, CBRS, and Wireless Networking for Airside Autonomous Operations ​

Table of Contents ​

1. DFW Airport 5G --- CBRS Deployment at Scale ​

Overview ​

Infrastructure Deployed ​

Network Architecture ​

What It Enables ​

Performance Metrics ​

Key Lessons ​

Sources ​

2. Changi Airport --- Private 5G for Autonomous Tractors ​

5G Aviation Testbed ​

Autonomous Tractor Fleet ​

5G's Role in Autonomous Operations ​

Future Expansion ​

Sources ​

3. CBRS 2.0 --- Regulatory Evolution and Airport Implications ​

What is CBRS? ​

Key CBRS 2.0 Changes (2024--2025) ​

Dynamic Protection Area (DPA) Reduction ​

NPRM for Further Changes (August 2024) ​

Controversy: Spectrum Reallocation Proposal ​

Implications for Airport Deployments ​

Sources ​

4. Private 5G vs Public 5G --- Why Airports Need Dedicated Networks ​

The Fundamental Problem with Public 5G for Autonomous Operations ​

Comparison Matrix ​

Why Private Networks Win for Autonomous Airside Operations ​

Airport Connectivity Infrastructure for Autonomous Vehicles

Research Report: 5G, CBRS, and Wireless Networking for Airside Autonomous Operations

Table of Contents

1. DFW Airport 5G --- CBRS Deployment at Scale

Overview

Infrastructure Deployed

Network Architecture

What It Enables

Performance Metrics

Key Lessons

Sources

2. Changi Airport --- Private 5G for Autonomous Tractors

5G Aviation Testbed

Autonomous Tractor Fleet

5G's Role in Autonomous Operations

Future Expansion

Sources

3. CBRS 2.0 --- Regulatory Evolution and Airport Implications

What is CBRS?

Key CBRS 2.0 Changes (2024--2025)

Dynamic Protection Area (DPA) Reduction

NPRM for Further Changes (August 2024)

Controversy: Spectrum Reallocation Proposal

Implications for Airport Deployments

Sources

4. Private 5G vs Public 5G --- Why Airports Need Dedicated Networks

The Fundamental Problem with Public 5G for Autonomous Operations

Comparison Matrix

Why Private Networks Win for Autonomous Airside Operations