Rel-17/18/19 features, NWDAF intelligence, NTN satellite, 5G RedCap, AI-native network, 6G preview, what GCC operators should prioritise
1. Where 5GC Is Going — The Simple Version
3GPP does not stop. Release 15 was the first SA specification. Release 16 added URLLC, V2X, and advanced slicing. Release 17 introduced Reduced Capability (RedCap) for IoT, Non-Terrestrial Networks (NTN) for satellite, and enhanced NWDAF analytics. Release 18 (the first “5G-Advanced” release) adds AI/ML-native network features, enhanced edge computing, and the foundations for ambient IoT. Release 19 will carry the first 6G concept studies.
For GCC operators who have just launched SA or are planning to: the immediate operational priorities are Rel-16 features (which are in production software today). Rel-17 and 18 are on the roadmap of every major vendor. 6G is a research item. This article covers what matters at each horizon and what you should actually be speccing for your next platform upgrade.
| 3GPP Reference |
| 3GPP TR 21.917 — Summary of Rel-17 work items |
| 3GPP TR 21.918 — Summary of Rel-18 work items (5G-Advanced) |
| 3GPP TS 23.288 — Architecture enhancement for 5G System supporting AI/ML-based Network Data Analytics (NWDAF) |
| 3GPP TS 23.737 — Study on architecture aspects for using satellite access in 5G |
2. Release-by-Release 5GC Evolution
| Release | Year | Key 5GC Features | GCC Relevance |
| Rel-15 | 2018–19 | First SA: AMF/SMF/UPF/PCF/UDM/AUSF/NRF/NSSF. Basic slicing, SBA, SUCI, 5G-AKA. | Foundation. All SA deployments run Rel-15 minimum. |
| Rel-16 | 2020 | URLLC (IIoT), V2X, enhanced UE Power Saving (eDRX, WUS), UPF redundancy, ATSSS (multi-access), 5G LAN. | Production now. URLLC for enterprise, V2X studies. Most commercial SA deployments. |
| Rel-17 | 2022 | RedCap (IoT NR), NTN (satellite 5G), enhanced NWDAF (AI analytics), network-controlled interactive services (NCIS), enhanced IIoT. | On roadmap. RedCap for asset tracking. NTN for maritime/oil rig. NWDAF for anomaly detection. |
| Rel-18 (5G-Advanced) | 2024 | AI/ML-native: RAN intelligence, NWDAF data collection for RAN. ISAC (Integrated Sensing and Communication). Ambient IoT (battery-free). Expanded satellite. Network energy efficiency. | 2025–2026 vendor roadmap. AI-driven optimisation. Energy savings (opex). ISAC for radar use cases. |
| Rel-19 | 2025+ | 6G concept studies, enhanced AI integration, advanced positioning, network as a sensor. | Research horizon. Inform 6G strategy but not specifiable yet. |
Table 1 — 3GPP release timeline. Rel-16 is production. Rel-17 is on vendor roadmap. Rel-18 (5G-Advanced) is 2025–2026. 6G is research.
3. NWDAF — The Intelligence Layer
The NWDAF (Network Data Analytics Function) is the AI/ML engine of 5GC. It collects telemetry from AMF (mobility events), SMF (session data), PCF (policy decisions), UPF (traffic statistics), and gNB (RAN measurements), runs ML models, and exposes analytics to consumer NFs via the Nnwdaf service.
| NWDAF Use Case | Data Sources | Output | Consumer NF |
| UE mobility prediction | AMF handover history, UE trajectory | Predicted next cell/TA in 5–30 seconds | AMF: pre-register to next cell; SMF: pre-insert UL CL at predicted DNAI |
| Slice load balancing | SMF per-slice session counts, UPF throughput per slice | Overloaded slice instance detected, rebalancing recommendation | NSSF: redirect new sessions to underloaded NSI; NSMF: scale-out NSI |
| Anomaly detection | AMF registration events, SMF session failures, UPF drop rates | Anomalous UE behaviour or NF behaviour flagged | PCF: apply restrictive policy; security function: investigate |
| QoS prediction | UPF per-session throughput history, RAN congestion | Predicted session throughput for next 30 seconds | PCF: adjust QoS class proactively; NEF: notify AF of predicted QoS change |
| Network energy optimisation | gNB cell load, UE distribution, time patterns | Cell sleep/wake schedule for low-load periods | OAM: automate cell shutdown during deep night; RAN: configure energy savings |
Table 2 — NWDAF use cases. Mobility prediction and anomaly detection are the two most near-term commercial deployments. Energy optimisation is growing as power costs rise.
| Field Note: NWDAF Mobility Prediction — 40% Reduction in Handover Latency |
| Operator trialled NWDAF mobility prediction on Muscat commercial district gNB cluster (50 cells). |
| NWDAF trained on 6-month historical handover trace. Predicted next cell for walking UEs with 78% accuracy at 10s horizon. |
| AMF pre-positioned UE context to predicted target AMF. X2 handover latency P95: reduced from 45ms to 27ms. |
| PDU session continuity during handover improved: 0.3% session drop rate vs 0.8% baseline. |
| Trial result: NWDAF ML model required 3 months to train on production data before accuracy was production-worthy. |
4. RedCap — 5G for Cost-Sensitive IoT
RedCap (Reduced Capability, 3GPP Rel-17) is 5G NR for IoT devices that do not need full 5G capability. Standard 5G NR requires MIMO antennas, wide bandwidth processing, and significant UE complexity. RedCap devices are allowed: 1–2 antennas (vs 4–8 for full 5G), 20 MHz bandwidth (vs 100 MHz), half-duplex FDD option, and relaxed processing power. Result: 5G device at IoT-grade cost and power consumption.
| Device Category | 3GPP Standard | Bandwidth | Complexity | Use Case |
| Full 5G NR | Rel-15+ | Up to 100 MHz UL, 400 MHz DL | High (4–8 antennas, complex UE) | Smartphones, fixed wireless, industrial broadband |
| RedCap (Rel-17) | Rel-17 | 20 MHz (or 5 MHz for FDD HD-FDD mode) | Medium (1–2 antennas) | Industrial sensors, wearables, smart home, basic video surveillance |
| Ambient IoT (Rel-18) | Rel-18 | Passive — backscatter | Ultra-low (no battery) | Battery-free tags, supply chain tracking, smart packaging |
Table 3 — 5G device categories. RedCap fills the gap between expensive full 5G and low-capability NB-IoT. Ambient IoT (Rel-18) targets battery-free mass deployment.
5. NTN — Satellite 5G
Non-Terrestrial Networks (NTN, Rel-17) extends 5G NR to satellite access. The same 5GC AMF, SMF, and UPF serve both terrestrial gNBs and satellite gateway nodes. From the 5GC perspective, a satellite node appears as a regular gNB on N2 (NGAP) and N3 (GTP-U). The difference is propagation delay: GEO satellite = ~600ms RTT, LEO satellite = 20–40ms RTT.
For GCC operators with maritime coverage requirements (Oman coastline, offshore oil platforms, Red Sea shipping lanes), NTN satellite integration into 5GC is the path to a single subscription offering seamless terrestrial and satellite coverage. Etisalat/e& and Ooredoo have publicly stated NTN roadmaps for GCC maritime and desert coverage.
| NTN Orbit | RTT | Bandwidth | 5GC Impact | Use Case GCC |
| GEO (35,786 km) | 600ms | High (established tech) | T3512/PFCP timers must be extended beyond default | Broadband fallback for remote sites |
| MEO (8,000–20,000 km) | 100–200ms | Moderate | Timer extension needed | Maritime, aviation |
| LEO (500–2,000 km) | 20–40ms | Growing with mega-constellations (Starlink, AST SpaceMobile) | Minimal timer changes needed | Maritime, desert coverage, offshore O&G |
Table 4 — NTN orbit types. LEO is the most commercially promising for GCC operators due to manageable latency. Oman’s extended coastline and interior desert make NTN commercially relevant.
6. AI-Native Network — Rel-18 and Beyond
Release 18 introduces AI/ML as a first-class design consideration, not just a bolt-on analytics function. Key concepts:
AI-native RAN: gNB components (beam management, scheduling, interference coordination) run ML models natively, not just reporting data to NWDAF. The gNB itself is an inference engine.
Split inference: ML model inference split between UE and network (for applications like video analysis where the UE captures but the network classifies). Requires dedicated QoS flows and NEF APIs for model distribution.
Digital twin: real-time virtual replica of the network used for “what-if” testing of configuration changes before applying them to production. NWDAF feeds the digital twin with live telemetry. Configuration candidates are tested on the digital twin before Helm upgrade.
Energy efficiency: AI-driven cell sleep/wake decisions based on predicted load. NWDAF models historical traffic patterns and generates sleep schedules. AMF and SMF coordinate idle UE migration to neighbouring cells before cell sleep.
7. GCC Operator Prioritisation — What to Focus On
| Horizon | Priority | Why | Investment Level |
| Now (Rel-15/16) | PFCP and N4 operational excellence | Most production incidents are Rel-15/16 configuration issues. Fix these first. | High — operational maturity |
| Now (Rel-16) | URLLC for enterprise (IIoT, port, logistics) | GCC government infrastructure and enterprise spending growing. First-mover URLLC wins 3–5 year contracts. | High — revenue |
| 12–18 months (Rel-17) | NWDAF anomaly detection and load balancing | Reduces ops cost, improves KPIs without additional hardware. | Medium — spec in next platform upgrade |
| 12–18 months (Rel-17) | RedCap for industrial IoT and smart city | GCC smart city programmes (NEOM, Masdar, MZDC). RedCap devices cheaper than full 5G for sensor deployments. | Medium — spec in RAN upgrade |
| 18–36 months (Rel-17/18) | NTN satellite integration for maritime/desert | Oman coastline, offshore platforms, Saudi desert coverage. Single subscription = higher ARPU. | Medium — partnership with LEO provider |
| 36+ months (Rel-18/19) | AI-native network, ambient IoT, ISAC | Research investment. Inform architecture decisions now so platform choices are future-compatible. | Low — research and early vendor engagement |
Table 5 — GCC operator 5GC evolution prioritisation. Operational excellence on Rel-15/16 is the highest-ROI investment. URLLC and RedCap are the near-term revenue opportunities.
8. Summary — Key Takeaways
| Topic | Key Takeaway |
| Release roadmap | Rel-15/16 = production. Rel-17 = vendor roadmap 2024-25. Rel-18 = 2025–26. 6G = research. Do not skip Rel-16 operational excellence for Rel-18 features. |
| NWDAF | AI analytics layer. 3-month minimum training period on production data before accuracy is useful. Start data collection now even if analytics not deployed yet. |
| RedCap | 5G for IoT. Lower cost, lower power, 5G security and QoS. Bridges gap between NB-IoT and full 5G. Key for smart city, industrial IoT, wearables. |
| NTN | Satellite 5G on standard 5GC. LEO (Starlink/AST) is most practical for GCC. Extends coverage to maritime and desert without separate satellite subscription. |
| 5G-Advanced (Rel-18) | AI-native RAN, split inference, digital twin, energy efficiency. Spec platform compatibility in next hardware procurement. These features require Rel-18 software on Rel-18-capable hardware. |
| GCC priority | Fix Rel-16 ops first. Win enterprise URLLC contracts now. Plan RedCap for smart city IoT. Engage LEO satellite partner for NTN roadmap. Let Rel-18 mature before committing capex. |
Table 6 — Post 20 summary. 5G is a 10-year platform. Understand the release roadmap and invest appropriately at each horizon.
5G Core Expert Series — Complete
All 20 posts in the 5G Core Expert Series are now complete. This series has covered every major dimension of 5GC design, deployment, and operations — from NSA Option 3x all the way to 5G-Advanced AI-native networks. The thread running through every post is the same: architecture decisions made at the design phase determine operational outcomes for the next decade. Get the fundamentals right — PFCP, N4, NSSAI, SEPP, certificate management, and GitOps — before chasing the next release’s features.