The Moving Clock Problem
In our first article, we focused on the Doppler effect and mitigation strategies required in NTN systems. In this second article of the series we move on to discuss the key challenges in synchronization issue and how to tackle synchronization errors that undermine NTN functioning.
In terrestrial networks, synchronization is largely treated as infrastructure. In LEO non-terrestrial networks, it becomes a continuously controlled service function. When that function degrades, consequences can include reducing usable capacity, destabilizing access and handover, weakening beamforming gain, and making SLA failures difficult to attribute. The commercial risk is therefore not merely a failed timing specification; it is lost network efficiency, longer integration cycles, and an inability to prove where service degradation originated.
Delay variation, Doppler shift, and carrier phase are different expressions of the same orbital dynamics: the first derivative of range is directly proportional to Doppler shift, the second derivative is proportional to Doppler rate. They share a root cause and must be modeled together.
For a LEO satellite at ~550 km altitude, one-way service-link delay is about 1.8 ms near zenith and rises to roughly 6–8 ms at low operating elevation angles. End-to-end latency is larger, incorporating feeder link, processing, and terrestrial transport, but on the service link, it is the rate of change that makes synchronization mechanisms harder, not the absolute value. Slant range changes continuously and non-linearly throughout a pass. The network tracks a moving time reference, not a fixed offset.
This matters for testing as much as for design. A system that tracks steady Doppler while ignoring Doppler rate, or corrects nominal delay while allowing timing and frequency estimates to diverge, can pass a static lab benchmark and still fail during a real orbital pass.
In terrestrial 5G NR, the gNB uses Timing Advance to align UE uplink transmissions at the receiver. The mechanism is largely reactive: the network observes arrival timing and commands a correction. In NR-NTN, the much larger and continuously changing path makes that approach insufficient by itself.
The NTN timing process is better understood as a hybrid of geometry-based prediction and measurement-based correction. The network can broadcast satellite ephemeris and common Timing Advance parameters. An NTN-capable UE can use its position, satellite geometry, and common timing information to calculate much of the required uplink timing and frequency pre-compensation before transmission. Closed-loop network commands then correct residual error caused by imperfect ephemeris, position uncertainty, oscillator behavior, atmospheric variation, and implementation delay.
This architecture changes how handover must be evaluated. When the UE moves to a new satellite, beam, cell, or gateway path, it must establish timing for a new trajectory. If target ephemeris and timing parameters are available in advance, the transition can be prepared. If they are incomplete or inconsistent, handover can introduce one of the sharpest discontinuities in the synchronization chain.
Timing alignment, frequency synchronization, carrier tracking, and relative phase calibration are related, but they solve different problems. Timing alignment controls slot and symbol arrival; frequency synchronization controls carrier offset and Doppler; carrier tracking follows phase rotation over the propagation path, while relative phase calibration maintains a stable relationship among the RF channels participating in coherent transmission or reception.
At Ka-band and mmWave frequencies, a carrier period is measured in tens of picoseconds. The engineering requirement, however, is not absolute time traceability at every node to that precision: in many phased-array and MIMO systems, the critical requirement is sufficiently stable relative phase and channel-to-channel skew across the participating RF paths.
For a single phased array, temperature, cable variation, and converter clocking all can shift the calibration state. For future distributed MIMO or multi-satellite architectures is even more complex as each platform adds its own oscillator behavior, propagation trajectory, reference distribution, processing latency, and residual phase error. The consequence is concrete: phase errors distort the channel estimate, reduce precoding accuracy, and produce throughput losses that link SNR alone does not predict and that are difficult to diagnose without direct visibility into synchronization and the calibration layer.
Architecture insight: A timing reference can be highly accurate at the gateway and still produce poor air-interface synchronization if the moving satellite path is modeled or corrected incorrectly. Reference accuracy and air-interface performance are not the same specification.
Synchronization compliance is necessary but not sufficient. Synchronization compliance confirms that specified limits were met at a test point, while synchronization observability explains where the remaining margin resides, what is consuming it, and how a timing, frequency, or phase error propagates into service performance
When an NTN service degrades, the fault could lie in satellite-geometry error, stale ephemeris, UE position uncertainty, oscillator instability, gateway reference distribution, TA residual, or beam switching. Without visibility into the synchronization layer, all of these look identical — they surface as access failures, throughput loss, or unexplained SLA violations, and they stay unattributed.
A useful NTN test environment must do more than insert a fixed delay and a constant frequency offset. It must translate a complete orbital state into physically consistent delay, Doppler, Doppler rate, phase evolution, and path loss — with phase continuity across parameter updates and support for separate service-link and feeder-link contributions. Static testing validates only isolated operating points. A complete test plan must traverse full passes, including high Doppler magnitude at low elevation, high Doppler-rate regions near closest approach or maximum elevation, beam transitions, gateway changes, reference loss and reacquisition, handover preparation, and holdover recovery.
The objective is to expose how the control loops interact over time, not to establish one worst-case number.
- CalIQ-Runtime provides the adaptive calibration layer for the TruSystems Platform, including SAT100T, by continuously calibrating the FlexSDR transmit and receive channels across frequency, gain, and temperature. It corrects channel-dependent amplitude, timing, and phase variations and maintains alignment across all TX and RX channels within each FlexSDR module. This calibration ensures that the measurement system itself remains accurate and coherent as operating conditions, frequency configuration, hardware temperature, and component behavior change.
- CalIQ-Interlink extends this calibrated relationship across multiple FlexSDR modules in a multi-module TruSystems configuration. It aligns timing and phase across the complete set of distributed TX and RX channels, allowing the system to operate as one coherent, calibrated RF measurement and monitoring platform. This calibration is essential for accurate real-time characterization of NTN RF signals. By characterizing and compensating for measurement-system errors and channel-to-channel inconsistencies, TruSystems can detect impairments originating in the monitored NTN signal
- This enables more reliable root-cause isolation and diagnosis, continuous RF performance monitoring, and real-time detection of signal errors, degradation, and anomalies.
LEO NTN changes the operating model of the terrestrial synchronization principles. Prediction becomes essential, residual measurement remains necessary, and timing, Doppler, phase, and calibration must be validated as integrated functions.
The most important design shift is to view synchronization as an observable lifecycle process. It must be modeled during architecture, reproduced with physical consistency in the laboratory, correlated with system behavior in field trials, and monitored during operation.
Final takeaway: The winning NTN architecture will not merely stay synchronized. It will know why it is synchronized, how much margin remains, and where to look when that margin begins to disappear.
| Challenge | Root Cause | Primary Mitigation | Residual Risk |
|---|---|---|---|
| Propagation delay variation | Non-linear slant range change throughout LEO pass | Geometry-based TA pre-compensation (GNSS + ephemeris) + closed-loop residual correction | Stale TA at beam edges; sharp discontinuity at handover |
| Doppler and Doppler rate | First and second derivatives of range; peaks do not coincide with delay peak | Coupled emulation from single orbital model; UE frequency pre-compensation | Residual CFO under high Doppler magnitude, rapid Doppler transition, or inaccurate frequency pre-compensation. |
| Phase coherence loss | Changing path length; per-channel skew; oscillator drift across elements | Operational system: adaptive relative phase and timing calibration across participating RF paths. Measurement system: CalIQ-Runtime and CalIQ-Interlink maintain calibrated TruSystems TX/RX channels so phase-coherence loss can be measured accurately. |
Precoding degradation; throughput below link-SNR prediction |
| Dynamic timing-path asymmetry in NTN | Moving, path-dependent delay breaks symmetric PTP assumptions | Hybrid GNSS/PTP with ephemeris-aware correction (ITU-T G.8275 series) | Systematic timing offset under ephemeris or model error |
| Synchronization observability gap | No single node sees the full error chain end-to-end | TruSystems, CalIQ-Runtime, and CalIQ-Interlink provide calibrated, timing- and phase-aligned RF generation, measurement, and monitoring | Unattributed SLA failures; extended integration and diagnosis cycles |