5G NTN: Network Procedures and Parameters

A detailed set of parameter and procedure adaptations in 5G NTN ensures successful attach, random access, AMF selection, Mobility and UE location verification under very long RTT, moving beams, and cross‑border coverage—conditions that do not exist in conventional terrestrial networks (TN).

Network attach changes

In NTN, the attach/registration flow remains 3GPP‑compliant (SSB/SIB acquisition → RACH → RRC connection → NAS to AMF), but the RAN and UE use satellite assistance to handle long propagation and moving footprints; SIB19 supplies ephemeris(Satellite Position, Velocity & Orbit Info Timing Parameters), Common Timing Advance (TA), and validity times so initial access and measurements proceed with feasible timing even for GEO/NGSO links, unlike TN where these aids are unnecessary. Attach may include early location verification and country compliance checks when cell footprints span borders; this additional post‑attach gating is specific to NTN to ensure selection of a lawful AMF domain and regulatory enforcement, while TN usually assumes cells map cleanly to a single country. The why: very long RTT breaks default timer assumptions, moving beams invalidate static geography assumptions, and regulators require correct jurisdiction routing at registration and service enablement.

 The UE can determine the network type (terrestrial or NTN) by looking at the information broadcasted in the SIB by each cell.

1. When a UE attaches to the network (registration procedures are the same as for 5G terrestrial networks), an appropriate AMF for that UE must be selected, to enable proper routing via the NG interface.

2. This function (Non-Access Stratum, NAS Node Selection Function, NNSF) resides in the NG-RAN node to determine the AMF association for the UE, based on, e.g., the temporary identifier for the UE, slicing information, onboarding information, serving cell, and estimated UE location

RACH parameter adaptations

NTN retains 4‑step and 2‑step RA, but enlarges timers and windows in TS 38.321 to account for satellite RTT; operators must configure ra-ResponseWindow, contention resolution timer, power ramping, and preamble repetitions for GEO/NGSO paths so a UE can still receive RAR within expected latency. PRACH formats and resources expand via 38.211 change requests for NTN/FR2‑NTN, with larger zero correlation zones and longer cyclic prefixes/preamble durations so the preamble remains detectable after long propagation and Doppler, unlike TN’s shorter designs; these tables are being consolidated in Rel‑17/18 texts and CRs. The why: GEO one‑way delays (~120–270 ms) and NGSO dynamics require longer reception windows and robust detection to prevent false timeouts and RA failures; TN parameters would cause systematic RA aborts in NTN.

RACH procedure modifications

Procedure logic in 38.321 is unchanged, but implementations rely on Common TA and offsets (e.g., koffset) signaled via SIB19 to pre‑bias uplink timing before Msg1, reducing mis‑timing that would otherwise exceed gNB receive windows; this is unique to NTN and not needed in TN. For mobility, NTN allows RACH‑less handover in defined intra‑satellite/feeder‑link switch cases to cut interruption caused by long RTT re‑access, while DAPS HO is explicitly not supported in NTN releases; TN commonly supports DAPS and rarely uses RACH‑less HO. The why: minimizing interruption across moving beams or feeder changes avoids repeated RA cycles over long RTT; DAPS assumptions on simultaneous connectivity are impractical with satellite payload constraints, hence disabled in NTN.

SIB19 and assistance signaling

Rel‑17 introduced SIB19 carrying NTN assistance: ephemeris of serving/neighbor payloads, Common TA parameters, koffset and validity duration for UL sync, stop time (t-Service) for quasi‑earth‑fixed beams, and referenceLocation with distanceThresh to trigger idle/inactive measurements; TN does not broadcast these NTN‑specific IE sets. This information lets the UE predict beam movement, pre‑align uplink timing, and know when a quasi‑fixed footprint will end, enabling proactive measurements and orderly reselection—capabilities that TN achieves with conventional, largely static cell planning. The why: ephemeris and timing validity compensate for dynamic coverage and drift; without it, UEs would repeatedly miss windows, mis‑measure neighbors, and drop service at beam edge.

SIB19 sample structure from 3GPP 38.331

systemInformationBlockType19 {

ntn-Config {

ephemerisInfo // Satellite position and velocity (state vector or orbital parameters)

epochTime // Epoch time for assistance information

ntn-UlSyncValidityDuration// Validity duration for uplink sync info

t-Service // Indicates when NTN cell stops serving current area (service time in 10 ms steps from Jan 1, 1900)

referenceLocation // Reference geo-location of serving cell

distanceThresh // Used for location-based measurement initiation (each step typically 50m)

commonTimingAdvance // Common TA value for all UEs as initial pre-compensation

koffset // Offset for UL-HARQ timing, scheduling

ntn-PolarizationUL // Uplink polarization info (optional)

ntn-PolarizationDL // Downlink polarization info (optional)

}

ntn-NeighCellConfigList { // List of neighbor satellite cells/gateways

... // Each entry provisioned with same structure as serving cell

}

}

This block is broadcast by the gNB over PDSCH and read by UEs accessing the satellite NTN service. For field and log examples, see. For full ASN.1 syntax, refer to 3GPP TS 38.331 Annex section for SIB19.

• ephemerisInfo: Satellite’s real-time coordinates or orbital data used by UE for delay/Doppler calculation.

• epochTime: Reference time for ephemeris and sync assistance.

• ntn-UlSyncValidityDuration: Duration for which UL sync parameters are valid.

• t-Service: Service stop time for beam/coverage, useful for handover prediction.

• referenceLocation: Latitude/longitude reference for cell (used in measurements).

• distanceThresh: UE triggers location-based measurements when moving away from reference location.

• commonTimingAdvance & koffset: Used for initial TA compensation and scheduling, unique to NTN.

  • K_offset is a cell-specific timing parameter used in 5G NTN to handle the large and variable propagation delays introduced by satellite communication, especially in GEO orbit

  • When the UE receives scheduling information for uplink transmission, it calculates the actual transmission slot as:

    UL tx slot=n+k2+Koffset

    where:

    • n+k2 is the usual uplink scheduling offset slot in terrestrial 5G,

    • Koffset is the cell-specific offset accounting for satellite delay.

  • The parameter k2 in 5G NR defines the offset in terms of number of slots between when the UE receives the Downlink Control Information (DCI) on the PDCCH that grants uplink resources and when the uplink transmission (PUSCH) must occur.

  • The range of k2 is typically from 0 to 32 slots.

• ntn-PolarizationUL/DL: Defines satellite link polarization.

Uplink time alignment maintenance

After RA, NTN relies on Common TA and drift models from RAN O&M ephemeris to keep UL aligned over long, slowly varying path lengths; 38.321 provides maintenance mechanisms while 38.300 NTN clauses highlight the need to model drift and update TA across session time, unlike the small TA range and slow variation in TN. The why: satellite motion continuously changes propagation delay by milliseconds, so periodic TA updates are mandatory; TN’s microsecond‑scale changes do not require this level of assistance. Take a look at my article on this. 5G-NTN: DL/UL Timing Synchronization

AMF selection and redirection

Baseline AMF selection follows 23.502/GUAMI rules, but NTN adds a country‑aware enforcement step in the 5G‑AN: if the UE is determined (via GNSS and beam mapping) to be in a country outside the serving AMF’s domain, the gNB can trigger NG handover or release to ensure the next registration hits the correct AMF, an NTN‑specific measure; in TN, fixed cells rarely necessitate such redirection. RAN3 discussions codify preconditions for 5G‑AN enforced AMF selection by country, while SA2 and RAN2 interactions clarify when network‑verified location should be used to back AMF selection correctness, especially when footprints cross borders. The why: cross‑border beams create jurisdiction and lawful intercept issues; correct AMF jurisdiction must be guaranteed at or right after registration, which TN can assume from static site geography.

UE location reporting and verification(optional)

3GPP agreed that, upon request and after AS security in connected mode, UEs should report coarse location; for NTN, the core may initiate LMF‑based verification using network methods tailored to NTN such as Multi‑RTT augmented by ephemeris/TA/drift assistance, delivering 5–10 km consistency checks rather than full GNSS accuracy; this is framed in TR 38.882 and TS 38.305 Rel‑18. RAN2 Rel‑18 study results define that UE location is considered verified if GNSS‑reported position matches network assessment within a few to ~10 km and recommend reusing LCS/LMF to operationalize verification; TN uses LMF mainly for services, not typically to police country compliance at registration. The why: GNSS may be spoofed or unavailable, and regulatory routing (e.g., emergency, lawful service restrictions) hinges on trustworthy country/region; NTN must verify coarse position because beams can span or rapidly traverse borders, a risk TN avoids with static cells.

Tracking area and CGI/VCID mapping

RAN2 addressed that CGIs used by the core should map to fixed geographic areas comparable to TN cells; for earth‑moving cells, NTN introduces virtual cell identifiers (VCIDs) or mapped TAIs anchored to geographic tiles rather than moving footprints, allowing AMF to treat them like stable TN cells for routing and policy; in TN, physical CGIs already map stably to land sites. Operators configure TAI/CGI mapping in the RAN using ephemeris and beam plans so the gNB reports ULI and TAI consistent with the UE’s actual geographic tile, enabling core policy enforcement independent of instantaneous beam index; this mapping complexity is unnecessary in TN. The why: without geography‑anchored identities, the same cell ID could roam across countries as the beam moves, breaking policy and emergency routing; VCID/TAI mapping restores fixed‑area semantics.

Handover and cell reselection

Handover and cell reselection in 5G NTN reuse NR procedures but introduce targeted changes for long RTT, moving beams, and cross‑border coverage: DAPS handover is not supported, RACH‑less handover is added for specific satellite/feeder‑link cases, and conditional handover gets new time/location triggers, while reselection gains NTN‑specific assistance and exceptions tied to beam availability and geography.

• What’s unique in NTN mobility

  NTN mobility must contend with earth‑moving or quasi‑fixed beams, so triggers and timings are adapted to predictable beam motion, unlike TN where sites are stationary and propagation is short and stable. This drives three headline differences: no DAPS handover, support for RACH‑less handover in defined cases, and enhanced conditional handover criteria that can include time‑ and location‑based triggers alongside measurement events.

• DAPS handover status

  DAPS handover (simultaneous dual connectivity to source and target during HO) is explicitly not supported for NTN in current releases, reflecting satellite payload and feeder resource constraints; by contrast, TN commonly uses DAPS to reduce interruption. The absence of DAPS means NTN must minimize interruption via other means, such as RACH‑less HO or CHO tuned to beam ephemeris and pre‑allocation of initial UL resources.

• RACH‑less handover addition

  RAN2 agreed NTN supports RACH‑less HO for intra‑satellite handover with the same feeder link, and can also support it for inter‑satellite mobility with the same feeder and for feeder‑link switchover, reducing interruption otherwise amplified by long RTT during a fresh RA. The generic UE procedure includes receiving a RACH‑less HO command possibly with a pre‑allocated grant (type‑1 CG), performing DL/UL sync, starting the NTN‑specific timers, monitoring PDCCH for dynamic grant if needed, sending the initial UL transmission including RRCReconfigurationComplete, and considering HO complete upon network confirmation; these steps are being refined across RAN2/RAN1 liaison items for beam association and power control details.

• Pre‑allocated vs dynamic grant

  For RACH‑less HO, the command may include a pre‑allocated UL grant tied to the target beam/SSB to enable immediate UL without PRACH; alternatively, the UE monitors PDCCH for a dynamic grant, with open points on beam indication and initial power control being coordinated with RAN1. Using pre‑allocated grants reduces reliance on downlink grant search under Doppler and RTT, while dynamic grants can avoid wasting reserved resources when timing is uncertain due to beam motion; both modes are considered for NTN to balance reliability and efficiency.

• Timers and completion criteria

  NTN RACH‑less HO introduces the use or adaptation of timers such as T304 (stopped upon completion) and an NTN‑specific T430/time‑alignment maintenance during HO, alongside contention resolution identity reuse for completion confirmation similar to LTE; these details are captured in RAN2 running CRs and liaison with RAN1 for PHY/MAC specifics. These changes are necessary because the round‑trip can delay HO acknowledgements, so explicit completion and alignment handling prevent premature aborts that would be rare in TN.

• Conditional handover extensions

  NTN keeps CHO but adds time‑based and location‑based trigger conditions that must be configured together with measurement‑based events (A3/A4/A5), enabling the network to schedule HO at a predicted beam boundary or geographic threshold rather than waiting purely on RSRP/RSRQ crossings; TN typically relies on measurement events alone. This is crucial when beams are quasi‑earth‑fixed with known service end time or moving footprints crossing a border, letting CHO execute exactly when coverage quality or jurisdiction will change.

• NTN↔TN mobility

  The specifications clarify that NTN↔TN mobility does not require the UE to keep simultaneous connections to both domains, further underlining why DAPS is excluded and why pre‑timed CHO or RACH‑less HO are emphasized for continuity across domain boundaries with high RTT. This avoids impractical dual connectivity given disparate link budgets and access schemes, while still allowing planned, low‑interruption transitions using time/location triggers.

• Cell reselection: baseline and NTN changes

  Cell reselection fundamentals remain per TS 38.304/38.331 with priority‑, threshold‑, and timer‑based rules, but NTN introduces assistance and exceptions that reflect frequency/beam availability by area and time, unlike TN where frequency availability is spatially static. RAN2 notes include that an RRC_IDLE/RRC_INACTIVE UE is not required to perform neighbor cell measurements for a TN frequency in an area configured with no TN coverage, even if that frequency has high priority, preventing futile scans under moving satellite footprints or remote regions.

• SIB assistance for reselection

  NTN deployments use system information (e.g., SIB19) to provide ephemeris and service time of beams, enabling UEs to anticipate when a serving quasi‑earth‑fixed beam will end and to prepare measurements and reselection toward a suitable target before loss, which is not needed in TN. Such assistance can include reference location and distance thresholds to start measuring only when approaching the beam edge, saving UE power and avoiding pointless measurements when the beam remains stable.

• Treselection and threshold tuning

  Operators tune Treselection values and threshX parameters to account for long measurement/decision cycles and paging reachability under RTT, ensuring that reselection happens early enough before beam edge while avoiding ping‑pong caused by Doppler or transient fades, whereas TN can use shorter, more reactive timers. The 38.304 framework supports per‑frequency Treselection and per‑priority thresholds; NTN simply applies them with beam‑aware assistance and possibly longer guard times to absorb scheduling and grant delays.

• Multi‑beam implications

  In multi‑beam cells, reselection rankings and time‑to‑trigger must consider beam switching events and known future degradation, so the UE is guided to a beam or cell with stable service window rather than the highest instantaneous RSRP; this differs from TN where the highest‑ranked neighbor is usually also the most stable. Predictive triggers (time/location) integrated through CHO and assistance information let NTN avoid last‑second reselections that would fail due to PRACH and security setup delays under high RTT.

Parameters to set

• RACH windows and timers: tune ra-ResponseWindow, contentionResolutionTimer, PreambleTransMax, powerRampingStep for GEO/NGSO RTT; validate against TS 38.321 Rel‑17/18 tables to avoid timeouts.

• PRACH formats/resources: select NTN/FR2‑NTN capable formats with extended ZCZ/CP and frequency planning per latest 38.211 CRs; confirm device support matrix for NGSO Doppler.

• SIB19 assistance: broadcast ephemeris, Common TA, koffset, validity duration, t-Service, referenceLocation, distanceThresh to enable pre‑bias timing and idle/inactive measurement triggers.

• Common TA and drift model: provision ephemeris and timing model via O&M so gNB computes TA updates and country mapping robustly over session time.

• AMF country enforcement: implement 5G‑AN rules to redirect or release when UE country differs from serving AMF domain; align with RAN3 guidance on country‑specific routing.

• Location verification: integrate LMF Multi‑RTT verification with 5–10 km consistency target post‑attach when needed for jurisdictional gating; reuse LCS flows per TS 38.305/TR 38.882.

Summary: Why NTN procedures differ from TN

• Propagation: RTT and drift mandate longer RA windows, new PRACH formats, and continuous TA maintenance; TN defaults fail under satellite delays.

• Mobility: moving beams require ephemeris‑driven assistance (SIB19), RACH‑less HO options, and CHO triggers tied to beam motion; TN cells are static.

• Jurisdiction: cross‑border coverage forces AMF country selection enforcement and network‑verified location; TN cell geography inherently enforces country.

• Identity mapping: VCID/TAI anchored to geography preserve fixed‑area semantics despite moving radio footprints; TN CGIs are already fixed.

Procedure	TN baseline	NTN behavior
Random Access	Standard 4‑step/2‑step RA with terrestrial TA ranges	Same RA types, but RA/TA windows, PRACH configs, and drift handling enlarged; FR2‑NTN PRACH tables added; UL alignment uses Common TA and offsets
Attach & NAS routing	Initial NAS to AMF per 38.300/23.502; minimal geography gating	Same flow, plus location checks after attach for country compliance; gNB/5GC may verify UE location and redirect to country‑appropriate AMF
AMF selection	GUAMI/AMF‑Set based; stable cell geography	Added rule: if UE detected in different country than serving AMF, trigger NG HO or release; note on location verification for AMF correctness
UE location	LMF used mainly for services/positioning	LMF reused to verify UE‑reported location via multi‑RTT with ephemeris/TA/drift assistance; CN checks consistency within km bounds
Mobility/Handover	DAPS, CHO widely available; RACH‑less HO uncommon	DAPS HO not supported; RACH‑less HO supported for intra‑satellite and some feeder‑link switches; additional CHO triggers for moving beams

References

1. 3GPP TS 38.300 Rel‑17/18 NTN clauses;

2. 3GPP TS 38.321 Rel‑17/18 MAC RA;

3. 3GPP TS 38.331

4. 3GPP TS 38.211 for NTN/FR2‑NTN PRACH; SIB19 assistance and mobility constraints; SA2/RAN2/RAN3 notes on country‑aware AMF selection and network‑verified UE location via LMF.