Why is it difficult to extend a cellular network into space? Distance is the obvious answer, but distance alone does not explain the access, mobility and capacity problems. The difficult part is making all these constraints work together while delivering a useful service.

This page maps the challenges to the mechanisms and design choices that address them. It distinguishes physical limits from protocol adaptations and deployment decisions. Start with Architecture if the service link, feeder link and payload types are unfamiliar.

• Why Is NTN More Than a Longer Radio Link ?
• Distance, Delay and Doppler in Numbers
• Propagation Delay Is Not Delay Spread
• Doppler Depends on Radial Velocity
• Why Can Coverage Fail Even With a Satellite in View ?
• What Can the Radio Procedures Compensate ?
• Moving Coverage, Handover and Battery Life
• Capacity, Ground Infrastructure and Cost
• How to Evaluate a Proposed NTN System
• References

Why Is NTN More Than a Longer Radio Link ?

A satellite link changes several things together. The signal travels farther, the satellite moves, and many users share a large footprint. Increasing transmit power helps only one part of this problem. The network still has to acquire the UE, schedule its uplink, maintain timing and keep a usable route to the core.

This page concentrates on satellite access, particularly LEO and GEO. Airborne platforms face related constraints, but their distances and motion are different. Also, NR NTN and IoT NTN do not use identical procedures. A mechanism supported by one radio technology or release should not be assumed for every terminal.

Figure 1 connects the physical cause to the protocol problem and then to the service impact. Each step needs its own measurements. A strong received signal alone does not prove successful access or acceptable application latency.

Figure 1. Physical constraints become protocol constraints, and both affect the delivered service.

Distance, Delay and Doppler in Numbers

Let us put numbers on the geometry first. The examples here are calculations with stated assumptions, not maximum values from a deployment. Use slant range, which is the actual distance along the radio path. Satellite altitude equals slant range only for a satellite directly overhead.

Propagation Delay Is Not Delay Spread

For distance d, one-way propagation time is d/c, where c is approximately 300,000 km/s. At a 600 km slant range, one service-link traversal takes about 2 ms. A transparent payload also adds the satellite-to-gateway path before the signal reaches the ground gNB.

The GEO row is an idealized minimum-distance calculation with both ground endpoints near the subsatellite point. Real geometry, processing, scheduling, terrestrial transport and queueing increase the total. An application ping is therefore not the same measurement as a service-link propagation round trip.

There are three different quantities to keep separate. Absolute delay is the travel time. Differential delay is the difference between two UEs: paths of 600 km and 900 km differ by 1 ms one way. Multipath delay spread describes copies of one signal arriving along different paths. A long satellite path does not by itself require a millisecond cyclic prefix.

Doppler Depends on Radial Velocity

The approximate frequency shift is fD = vr × fc / c. Here vr is radial velocity, not the full orbital speed. With vr = 7 km/s and fc = 2 GHz, the magnitude is about 46.7 kHz. The sign changes between approaching and receding motion.

A correction for the beam center leaves residual errors at other locations. Oscillator errors add another contribution. The practical test therefore needs frequency offset, rate of change and residual error after compensation. Testing only one constant offset misses part of the tracking problem. See Frequency Compensation.

Why Can Coverage Fail Even With a Satellite in View ?

Visibility is only the first condition. The receiver needs enough energy relative to noise and interference, and the antenna must provide useful gain in the satellite direction. Trees, buildings, terrain and the terminal installation can block or weaken the path even inside a predicted footprint.

Free-space path loss in dB is 92.45 + 20 log10(fGHz) + 20 log10(dkm). At 2 GHz, it is about 154.0 dB for 600 km and 160.0 dB for 1,200 km. Doubling distance adds about 6 dB of loss. This calculation excludes atmosphere, polarization mismatch, body loss and interference.

Lower elevation usually increases range and exposes the path to more local obstructions. Rain attenuation also depends strongly on frequency: it deserves particular attention for higher-frequency feeder links and broadband terminal links. Do not apply a Ka-band rain-loss assumption directly to an L- or S-band handset link.

A larger antenna can improve gain, but narrower beams need accurate pointing and tracking. Mechanical steering and electronic beamforming have different installation, power and cost implications. A small IoT terminal, a handheld UE and a directional VSAT therefore need separate budgets. A nominal 0 dBi assumption is not a promise of equal gain in every direction.

Check both directions. Satellite EIRP can make the downlink detectable while the UE uplink remains too weak. More robust coding or repetitions may improve reliability, but they consume airtime and energy. The worked calculations in Link Budget show how these terms combine.

What Can the Radio Procedures Compensate ?

Compensation makes a long link usable; it does not shorten the physical path. In the Release 17 baseline, NR NTN uses UE position and satellite ephemeris for timing and service-link Doppler pre-compensation. The assistance must remain valid as the geometry changes. See TR 21.917, clause 5 for the baseline scope.

Long feedback loops also affect scheduling and HARQ. NTN timing adaptations and additional HARQ options address those loops. They do not create extra bandwidth or guarantee a particular application response time. NR and IoT NTN must be checked separately against the supported release and UE capabilities.

For a simple pipeline example, assume one transport block is started every 1 ms and its process cannot be reused for 20 ms. Roughly 20 independent processes would be needed to keep that simplified pipeline full. This is a teaching calculation, not an NR scheduling rule: actual feedback timing, processing and configuration determine reuse.

The troubleshooting sequence matters. First establish whether the signal is detected. Then check timing and frequency alignment, followed by response timing and retransmissions. A timeout may be caused by a weak uplink or a timing error; increasing a timer without finding the cause can hide the failure and increase waiting time.

Moving Coverage, Handover and Battery Life

A stationary UE can still experience changing range, changing Doppler and the end of a serving satellite pass. Beam steering can hold a footprint over an area for a period, but the satellite continues moving. Beam movement and cell movement should therefore be described separately.

A satellite change does not always mean an RRC handover. The outcome depends on cell identity, payload architecture and network coordination. Where handover is needed, regular measurement-based procedures remain relevant; conditional handover can add prepared execution conditions. Time or location assistance does not replace every measurement-triggered handover. See Handover.

If no next satellite can provide service, handover cannot solve the gap. The device needs suitable search and retry behavior, and the application must tolerate delayed delivery or unavailability. A coverage map should state whether it represents an instantaneous footprint, availability over time or continuous service.

Power saving introduces a second timing problem. A sleeping UE saves energy but may miss a service opportunity or wait longer for a downlink. Repeated scanning during a known gap also wastes power. NR paging and DRX should not be treated as interchangeable with the eDRX and PSM behavior available in applicable IoT systems.

For design review, record pass duration, gap duration, time to acquire a cell, retry energy and the probability of completing the transaction before coverage ends. Coverage and Beam Movement, Discontinuous Coverage and Paging and Power Saving develop these questions.

Capacity, Ground Infrastructure and Cost

A large footprint is useful for geographic reach, but it can also collect traffic from many users into limited radio resources. Capacity depends on bandwidth, frequency reuse, interference, beam scheduling and available payload power. Satellite count alone is not a sufficient capacity measure.

The ground path matters as well. A satellite needs an available route toward the service network. Inter-satellite links can reach a distant gateway, but they introduce routing and link-management requirements. A regenerative payload can move processing onboard; it does not remove the need for a route to the relevant network services.

Satellite power, thermal limits, antenna size and processing capability constrain how much traffic can be served at once. Ground gateways need suitable sites, transport connectivity and resilience. A feeder-link outage can interrupt service even when the UE has a clear view of the satellite. Gateway diversity helps only if the alternate route has usable capacity.

Spectrum coordination, national authorizations and interference management constrain the usable footprint. Borders are not physical radio barriers. Operational planning also includes collision avoidance, debris mitigation, satellite replacement and the effects on astronomy. These are continuing responsibilities, not tasks completed when the first satellite reaches orbit.

The business comparison should use cost per delivered service at the required availability. Include terminals, gateway operations, replenishment and support, as well as launch. Different traffic types justify different designs: an occasional sensor report and a continuous broadband session place very different demands on the same coverage area.

How to Evaluate a Proposed NTN System

A useful review starts with a service requirement and then tests the weakest conditions. Specify the terminal, radio technology, orbit, minimum elevation, traffic load and availability target. Without these assumptions, two impressive demonstrations may be measuring completely different things.

Measure distributions as well as averages. A good average delay can hide long outages, and a high peak rate can hide poor loaded performance. Repeat tests across terminal orientations and representative locations. Keep the radio measurements separate from transport and application measurements so that a failure can be assigned to the correct part of the path.

The physical constraints remain, but their severity depends on design choices. Lower altitude reduces distance while increasing the number of satellites needed for broad continuous coverage. Narrower beams can improve reuse and control residual errors, while increasing beam-management demands. More redundancy improves resilience while consuming equipment and operating budget.

For conventional cellular-phone satellite access, use D2C Challenges. That page focuses on keeping the handset radio behavior compatible while the satellite and ground network accommodate the unusual link.

References

The protocol discussion uses the Release 17 baseline summary in clause 5; it is not a claim that every later feature is mandatory. Numerical examples on this page are illustrative calculations from the stated geometry.

• 3GPP TR 21.917 V17.0.1 : Release 17 Description, clause 5, integration of satellite components and NR / IoT NTN adaptations.