Release 18 is the first formal 5G-Advanced specification package. It builds on the foundational 5G work of Releases 15–17 while adding intelligence, efficiency, broader device support, and deeper integration of non-terrestrial and industrial capabilities. It is now frozen and serves as the baseline for subsequent 5G-Advanced releases (Rel-19 and beyond) on the path toward 6G.

Timeline and Status

3GPP Release 18 content was largely agreed at the December 2021 TSG meetings. Specification work ran through 2022–2024. Functional freeze (Stage 3) occurred around SA#103 in March 2024, with protocol stability and ASN.1 freeze completed at SA#104 on 21 June 2024. The release is officially frozen.

The authoritative high-level summary is contained in 3GPP TR 21.918 (ETSI TR 121 918 V18.0.0), which catalogues every major Feature and Work Item in its initial post-freeze state. Maintenance and corrections continue, as is normal for several years after freeze.

Aspect	Detail
Position in roadmap	First 5G-Advanced release
Start of content decisions	December 2021 (TSG #94-e)
Functional freeze	March 2024 (SA#103)
Protocol / ASN.1 freeze	June 2024 (SA#104)
Current status	Frozen
Summary document	TR 21.918

Positioning Within 5G Evolution

• Release 15: Initial 5G NR and 5GC baseline (eMBB focus).
• Release 16: URLLC, industrial IoT, V2X, unlicensed spectrum, positioning foundations.
• Release 17: NTN (satellite), RedCap, multicast/broadcast (MBS), further IoT and coverage improvements.
• Release 18: Explicitly branded 5G-Advanced. It strengthens the radio and system foundation while expanding support for lower-complexity devices, immersive media, AI/ML-driven optimization, energy efficiency, and deeper vertical integration.

3GPP formally adopted the “5G-Advanced” designation for Rel-18 and later specifications. The release is designed both for near-term commercial value and as a bridge toward 6G studies.

Two Overarching Pillars of Release 18

Industry analyses commonly organize Rel-18 work into two complementary directions:

1. Strengthen the end-to-end 5G system foundation — higher spectral efficiency, better mobility, lower energy consumption, and intelligent network behaviour.
2. Proliferate 5G to virtually all devices and use cases — cost- and power-optimized devices, XR, sidelink, precise positioning, non-terrestrial networks, drones, and industrial verticals.

Major Feature Areas

Radio Access Network (RAN) Enhancements

MIMO Evolution

Continued refinement of multi-antenna techniques remains central. Rel-18 improves CSI feedback for medium- and high-velocity UEs, extends the unified TCI framework to multi-TRP scenarios, supports higher-rank uplink transmission (up to rank 8 in certain configurations), and increases the number of orthogonal DMRS ports for multi-user MIMO. These changes improve both capacity and coverage, especially at cell edges and in multi-TRP deployments.

AI/ML for the NR Air Interface and NG-RAN

This is the first release in which AI/ML appears as a formal standards topic in RAN. Studies and normative work focused on three representative air-interface use cases (CSI feedback compression, beam management, and positioning) plus system-level use cases in NG-RAN: network energy saving, load balancing, and mobility optimization. Data-collection frameworks, signalling support, and lifecycle management for models were specified. Full AI/ML-native air-interface designs remain for later releases, but Rel-18 establishes the necessary frameworks and interfaces.

Energy Efficiency and Network Energy Savings

New techniques allow cells to enter deeper sleep states (cell DTX/DRX), adapt antenna ports and transmission power more dynamically, support SSB-less secondary cells, and coordinate energy-saving modes with mobility procedures. Device-side power saving is also improved.

Duplex Evolution, Coverage, and Deployment Flexibility

Work included sub-band non-overlapping full-duplex studies, dynamic TDD enhancements, network-controlled repeaters (NCR), and mobile Integrated Access and Backhaul (IAB). These improve spectrum utilization and allow more flexible, cost-effective densification.

Mobility Improvements

Layer-1/Layer-2 triggered mobility (LTM), conditional handover enhancements, and better multi-connectivity handling reduce interruption times and improve reliability, particularly for high-mobility and multi-TRP scenarios.

Device and IoT Evolution

Enhanced Reduced Capability (eRedCap)

Building on Rel-17 RedCap, Rel-18 further reduces complexity: narrower baseband bandwidth (down to 5 MHz in some configurations), lower peak data-rate targets (around 10 Mbps), FR1-only operation in the lowest tier, and extended eDRX cycles in RRC_Inactive (well beyond 10.24 s). These changes target lower-cost industrial sensors, smart-grid devices, and other mid-tier IoT endpoints while remaining distinct from NB-IoT/eMTC.

XR (Extended Reality) Enhancements

Specific awareness of XR traffic patterns (buffer status, latency hints, PDU-set handling), improved configured-grant efficiency, power-saving DRX adaptations matched to frame rates, and capacity optimizations allow more simultaneous XR users with acceptable quality and battery life.

Sidelink Evolution

Carrier aggregation, operation in unlicensed spectrum, improved relay (UE-to-UE and UE-to-Network) with service continuity and multi-path support, and coexistence mechanisms with LTE sidelink expand device-to-device and public-safety use cases.

Positioning and Location Services

Significant accuracy and applicability improvements:

• Bandwidth aggregation of positioning reference signals.
• Carrier-phase positioning measurements aiming at centimetre-level accuracy.
• Support for RedCap devices (including frequency hopping beyond their maximum channel bandwidth).
• Low-power high-accuracy positioning (LPHAP) suitable for industrial IoT with long battery life.
• Sidelink-based positioning and ranging.
• Integrity signalling and continuity across terrestrial/non-terrestrial access.

Non-Terrestrial Networks (NTN) and Aerial Platforms

Further integration of satellite access into the 5G System:

• Uplink coverage improvements for smartphone-class devices.
• Mobility and service continuity between terrestrial and non-terrestrial networks (and between different satellite systems).
• Support for higher-frequency bands (including Ka-band scenarios with VSAT terminals).
• IoT-NTN enhancements and discontinuous coverage handling.
• Uncrewed Aerial Vehicle (UAV) support: altitude reporting, flight-path information, detect-and-avoid (DAA) via sidelink, and related regulatory signalling.

Core Network, Services, and Vertical Enablement

• Network Slicing Phase 3: Temporary slices, partial coverage, improved admission control (NSAC), slice-specific authentication enhancements, and better roaming support.
• Multicast/Broadcast Services (MBS) Phase 2: Reception in RRC_Inactive, resource efficiency in shared RAN deployments, and improved simultaneous unicast/multicast handling.
• Edge Computing Phase 2, northbound API enhancements (including resource-owner-aware access), and Service Enabler Architecture Layer (SEAL) expansions for vertical applications.
• Media and Real-Time Communication: IMS Data Channel support enabling richer “New Calling” services, XR media profiles, and tactile/multi-modality communication foundations.
• Security: Vertical-specific enablers, EDGEAPP security, slicing security, and continued SCAS (Security Assurance Specification) work.
• Management and Orchestration: AI/ML model lifecycle management, intent-driven management, enhanced measurements/KPIs for energy and performance, and charging adaptations for the new features.

Industry and Commercial Implications

Release 18 enables operators to extract more value from existing 5G spectrum and infrastructure through higher spectral efficiency, lower energy costs, and support for new device classes and services. Vertical industries (factories, energy, public safety, automotive, media) gain more mature APIs, precise positioning, deterministic communication tools, and satellite coverage options. Device vendors can produce lower-cost 5G IoT modules and XR-optimized terminals. The introduction of structured AI/ML frameworks begins the shift from purely rule-based to data-driven network operation—an essential step toward autonomous networks and eventual 6G architectures.

Because the release is frozen, commercial implementations and chipset support are progressing. Many features are already appearing in operator trials and early commercial offerings labelled “5G-Advanced” or “5G-A.”

Looking Ahead

Release 18 is the foundation. Release 19 continues the 5G-Advanced trajectory with further AI/ML integration, additional energy and spectrum-efficiency gains, and expanded support for emerging use cases. Parallel 6G study items are already under way, but Rel-18 ensures that the 5G system remains commercially relevant and technologically progressive for the remainder of the decade.

In summary, 3GPP Release 18 marks the formal start of 5G-Advanced by delivering measurable improvements in radio performance, energy efficiency, device accessibility, positioning accuracy, non-terrestrial integration, and intelligent network behaviour. It is both an evolutionary refinement of the 5G system defined in earlier releases and a deliberate platform for the next wave of commercial and industrial applications.

---

1) Strengthen the End-to-End 5G System Foundation

Release 18 treats the 5G system not merely as a connectivity layer but as a robust, evolvable platform. Enhancements target spectral efficiency, reduced interruption times, lower energy consumption, data-driven optimization, and flexible densification. These improvements deliver immediate gains in existing networks while preparing the architecture for denser, more intelligent, and greener operations.

Advanced Downlink and Uplink MIMO Evolution

Massive MIMO has been central to 5G since Release 15. Release 18 continues its refinement with a focus on four areas: channel state information (CSI), multi-transmission/reception point (multi-TRP) operation, uplink capabilities, and reference signals.

Key technical advances include:

• Enhanced CSI and CSI reference signals that better exploit time-domain correlation for medium- and high-velocity user equipment (UEs) in sub-7 GHz bands. This mitigates channel aging and improves feedback accuracy under mobility.
• Extension of the unified Transmission Configuration Indicator (TCI) framework (introduced in Rel-17 for single-TRP) to multi-TRP scenarios, supporting multiple simultaneous TCI states for downlink and uplink along with dual timing advance (TA) values.
• Support for a larger number of orthogonal demodulation reference signal (DMRS) ports—up to 24 for multi-user MIMO in CP-OFDM—enabling higher multi-user capacity.
• Uplink enhancements targeting customer premises equipment (CPE), fixed wireless access (FWA), vehicles, and industrial devices: multi-panel simultaneous uplink transmission, support for 4+ transmit antennas and up to rank-8 transmission, and improved sounding reference signal (SRS) designs for better interference randomization across TRPs.

These changes raise both downlink and uplink spectral efficiency, improve cell-edge performance, and make multi-TRP deployments more practical. Uplink MIMO gains are particularly valuable in mid-band spectrum where uplink coverage has historically been a bottleneck.

Enhanced Mobility: Layer-1/Layer-2 Triggered Mobility (LTM)

Legacy Layer-3 handovers incur measurable interruption and signalling overhead. Release 18 introduces L1/L2-triggered mobility (also called L1/L2 inter-cell mobility) to reduce latency, overhead, and service interruption.

Core mechanisms:

• Pre-configuration of multiple candidate cells by the source node.
• Dynamic switching among candidates based on Layer-1 measurements and beam management (including L1 measurement reporting and beam indication).
• Support limited to intra-central-unit (intra-CU) scenarios in Rel-18, with non-serving-cell beam management and timing advance acquisition.
• Lightweight MAC-level cell-switch commands rather than full RRC reconfiguration in many cases.

System-level evaluations show shorter interruption times and improved mobility robustness because decisions leverage faster, more sensitive L1 measurements, reducing the risk of radio-link failure and subsequent re-establishment. This is especially beneficial for high-mobility users and for maintaining quality in multi-connectivity or multi-TRP environments.

Flexible Deployment: Mobile IAB and Network-Controlled Repeaters

Coverage densification and backhaul challenges are addressed through two complementary tools.

Mobile Integrated Access and Backhaul (IAB) extends stationary IAB by allowing the IAB node itself to move. Procedures cover inter-donor migration, interference mitigation, radio-link failure recovery, and support for legacy UEs. Mobile IAB enables temporary or vehicle-mounted coverage extension while reusing the NR access and backhaul framework.

Network-Controlled Repeaters (NCR) evolve traditional RF repeaters into network-managed devices. An NCR exchanges side-control information with its serving gNB, enabling beamforming control, awareness of TDD configuration, and more intelligent amplification/forwarding. This provides cost-effective coverage extension on both FR1 and FR2 without the full complexity of an IAB node.

Together, these features expand the operator’s toolbox for rapid, low-cost densification and temporary coverage solutions.

AI/ML Data-Driven Designs

Release 18 marks the first formal introduction of AI/ML into the NG-RAN standards framework. Work focused on three system-level use cases—network energy saving, load balancing, and mobility optimization—plus foundational air-interface studies (CSI feedback, beam management, and positioning).

Normative elements include:

• Enhanced data collection and reporting over existing interfaces (notably Xn).
• Signalling to support model inference and prediction exchange between neighbouring nodes.
• A management framework covering model lifecycle aspects (training, deployment, inference) that is further elaborated in SA5.

AI/ML does not yet replace traditional algorithms; instead it augments them with predictive insights (for example, future load or energy-cost estimates). This lays the groundwork for more autonomous, proactive network behaviour and is a deliberate stepping stone toward deeper AI-native designs in later releases.

Green Networks and Devices: Network Energy Savings

Energy efficiency is elevated to a first-class concern. Release 18 specifies techniques in the spatial, power, time, and frequency domains after a dedicated study that defined a base-station energy-consumption model and evaluation methodology.

Notable mechanisms:

• Dynamic adaptation of the number of active antenna ports and transmission power, guided by enhanced UE feedback on channel quality at different adaptation levels.
• Cell discontinuous transmission/reception (cell DTX/DRX) with improved alignment and coordination.
• Support for SSB-less secondary cells and other techniques that create more opportunities for deep sleep states under low-to-medium load.
• Spatial adaptation and power-domain optimizations that can yield double-digit percentage energy savings while protecting user throughput and latency.

Device-side power-saving foundations are also strengthened, including studies on low-power wake-up signals/receivers. These features directly address operator operating-expense pressures and sustainability targets.

Evolved Duplexing

Release 18 includes a study on the evolution of NR duplex operation, with particular attention to sub-band non-overlapping full duplex (SBFD) at the gNB and enhancements for dynamic/flexible TDD.

SBFD allows simultaneous downlink and uplink in different frequency sub-bands within a conventional TDD carrier, improving uplink coverage and perceived latency. Cross-link interference (CLI) modelling and mitigation techniques were evaluated. While full single-frequency full-duplex remains a longer-term goal, Rel-18 establishes the regulatory, interference-management, and performance foundations needed for progressive duplex evolution.

Summary of Foundation-Strengthening Features

Feature Area	Primary Technical Gains	Key Benefits
Advanced DL/UL MIMO	Higher-rank UL, multi-TRP TCI, more DMRS ports, velocity-aware CSI	Higher capacity, better cell-edge & mobility performance
L1/L2 Mobility (LTM)	Pre-configured candidates, L1-triggered switch	Lower interruption, greater robustness
Mobile IAB & NCR	Mobile nodes, network-controlled beamforming	Flexible, cost-effective coverage
AI/ML for NG-RAN	Data collection, prediction for energy/load/mobility	Proactive optimization, path to autonomy
Network Energy Savings	Antenna/power adaptation, cell DTX/DRX	Significant energy reduction at low-medium load
Evolved Duplexing	Sub-band full duplex foundations, CLI handling	Improved UL coverage & spectrum flexibility

Broader System Impact and Commercial Relevance

Collectively these enhancements raise the performance ceiling of existing mid-band and high-band deployments, reduce the cost of densification, lower energy consumption, and introduce the first standardized AI/ML hooks into the RAN. Operators gain tools that improve spectral efficiency and total cost of ownership without requiring entirely new spectrum or site builds. Device and infrastructure vendors obtain clearer targets for chipset and base-station implementations that began appearing in commercial 5G-Advanced offerings after the June 2024 protocol freeze.

The foundation-strengthening work of Release 18 is deliberately evolutionary rather than revolutionary. It refines the mature 5G system so that it can absorb higher traffic volumes, support denser and more dynamic topologies, operate more sustainably, and host the AI-driven and immersive services that define the next phase of 5G-Advanced. Subsequent releases build directly on these capabilities, confirming Release 18 as the essential platform upgrade that keeps 5G competitive through the remainder of the decade.

---

2) AI/ML Data-Driven Designs

Release 18 marks a pivotal shift: AI/ML moves from high-level principles and studies into concrete, interoperable signalling, data-collection mechanisms, and lifecycle management support. It does not replace traditional algorithms with fully AI-native air interfaces; instead it establishes the necessary frameworks, evaluation methodologies, and interfaces so that data-driven techniques can augment existing 5G functions in a multi-vendor environment.

AI/ML for the NR Air Interface (RAN1 Study)

The Release 18 study item on AI/ML for the NR air interface (documented primarily in TR 38.843) was the first of its kind in 3GPP. It defined a general functional framework and evaluated three representative use cases.

Functional Framework

The framework includes data collection, model training, model inference, and an actor function that applies the inference results. It supports both one-sided models (inference at either the UE or the network) and two-sided models (encoder at the UE, decoder at the gNB, or vice versa). Considerations include lifecycle management signalling, capability indication, model identification, data generation/transfer, generalization testing, and interoperability.

Three Primary Use Cases

1. CSI Feedback Enhancement
   • Spatial-frequency domain compression using two-sided models (UE encoder + gNB decoder) to reduce uplink overhead while maintaining reconstruction accuracy.
   • Time-domain CSI prediction using one-sided models to combat channel aging, especially under medium-to-high UE velocity. Evaluations showed overhead reductions in the range of roughly 8–79 % accompanied by modest throughput gains (0–17 % in reported cases), depending on scenario, model complexity, and generalization conditions.
2. Beam Management
   • Spatial-domain downlink beam prediction (predicting the best beam from a reduced measurement set).
   • Temporal-domain beam prediction (forecasting future best beams to reduce measurement and reporting latency). These approaches aim to lower the overhead and latency of exhaustive beam sweeping, particularly valuable in FR2 (mmWave) deployments that rely on analog beamforming.
3. Positioning Accuracy Enhancements
   • Direct AI/ML positioning (model outputs location estimates directly from measurements).
   • AI/ML-assisted positioning (models refine conventional measurements such as ToA, AoA, or multi-RTT). Simulations indicated potential for horizontal accuracy better than 1 m at the 90th percentile in certain indoor/outdoor scenarios, compared with significantly worse results from purely conventional methods under the same conditions.

The study also assessed model complexity, generalization across environments and vendors, data requirements, and potential specification impacts on the physical layer and higher-layer protocols. Full normative work for many of these air-interface techniques continues into Release 19 and later.

AI/ML for NG-RAN (RAN3 Normative Work)

Building on the Release 17 study (TR 37.817), Release 18 delivered the first normative specifications for AI/ML support inside the NG-RAN. The work is limited primarily to the non-split architecture and focuses on three high-priority use cases that align with self-organizing network (SON) goals.

Core Enablers

• Enhanced data-collection and reporting procedures over the Xn interface (Data Collection Reporting Initiation and Reporting procedures).
• Exchange of measurements, predictions, and feedback between neighbouring NG-RAN nodes.
• Introduction of an “Energy Cost” metric for energy-related decisions.
• Support for two main deployment scenarios: – Model training in OAM with inference at the NG-RAN node. – Both training and inference located at the NG-RAN node.

Three Standardized Use Cases

• Network Energy Saving
  • AI/ML models use collected and predicted energy-cost and load information to decide cell or node activation/deactivation. The goal is to reduce overall energy consumption in a geographic area while protecting coverage and quality of service. High-level actions (rather than fine-grained DTX/DRX control) were prioritized in Rel-18.
• Load Balancing
  • Models generate load predictions that are exchanged between nodes. These predictions, combined with UE performance feedback after offloading actions, enable more proactive traffic steering and resource distribution across cells, carriers, or areas.
• Mobility Optimization
  • UE trajectory predictions are generated and can be included in handover-request messages. The receiving node can use the predicted path to select better target cells or prepare resources, improving handover success rates and reducing unnecessary handovers or radio-link failures. Actual trajectory feedback can later be used to monitor prediction accuracy.

These mechanisms remain use-case and data-type agnostic at the signalling level, giving implementations flexibility while ensuring interoperability for the essential information exchange.

AI/ML Management and Lifecycle Framework (SA5)

Parallel work in SA5 produced a domain-agnostic AI/ML management framework (TS 28.105) covering the full operational workflow of ML models and inference functions.

Key lifecycle phases supported include:

• Model training (including validation and testing).
• Emulation.
• Deployment.
• Inference execution and monitoring.

Management capabilities are exposed via Network Resource Model (NRM)-based services and OpenAPIs. Operators or management consumers can request, control, monitor, and report performance of training, re-training, and inference processes. This framework is intentionally independent of specific RAN or core-network features so it can serve multiple AI/ML-enabled functions across the 5G System.

Benefits, Constraints, and Multi-Vendor Considerations

Potential Benefits

• Reduced signalling overhead and improved spectral efficiency (CSI and beam management).
• Higher positioning accuracy in challenging environments.
• More proactive and accurate network energy, load, and mobility decisions.
• Foundation for progressive automation of RAN operations.

Practical Constraints and Open Issues

• Generalization of models across different vendors, environments, and UE types remains a key challenge.
• Data collection, privacy, and labelling quality affect model performance.
• Computational and energy cost of running inference on UEs or base stations must be balanced against gains.
• Multi-vendor model transfer and trustworthiness are only partially addressed; full inter-vendor collaboration models continue to be refined.
• Rel-18 air-interface work was largely a study; normative impact on the physical layer is limited and expands in later releases.

Evolution Beyond Release 18

Release 19 builds directly on these foundations. For NG-RAN it adds support for network slicing and coverage/capacity optimization (CCO), continuous MDT collection across RRC states, mobility optimization for NR dual connectivity, and initial split-architecture considerations. Air-interface work progresses toward normative two-sided models and additional use cases. Management frameworks are further refined for scalability and consistency across the 5G System.

In summary, the AI/ML data-driven designs of Release 18 establish the first standardized scaffolding—functional frameworks, data-exchange procedures, lifecycle management, and carefully scoped use cases—upon which more ambitious AI-native capabilities can be built. They transform AI/ML from a research topic into an interoperable system feature that strengthens the end-to-end 5G foundation while preserving the multi-vendor, standards-based nature of the mobile ecosystem.

---

3) Green Networks and Devices: Network Energy Savings

Energy efficiency became a first-class priority in Release 18. Operators face rising energy costs and sustainability targets, while the radio access network typically accounts for the majority of a mobile network’s energy use. Release 18 introduced the first dedicated normative work on network energy savings for NR after a comprehensive study, providing standardized tools that allow dynamic, finer-granularity adaptation of transmissions and receptions while protecting user experience and supporting legacy devices.

Study Foundations

The Release 18 work began with a study item that defined:

• A base-station energy consumption model (the first common model of its kind in 3GPP for this purpose).
• Evaluation methodology and key performance indicators (KPIs) that balance energy savings against user-perceived throughput (UPT), latency, capacity, spectral efficiency, and UE power consumption.
• Priority scenarios of idle/empty and low-to-medium load, while allowing different loads across carriers and neighbouring cells.
• Requirements that legacy UEs must generally continue to access networks implementing the new techniques (with limited exceptions for greenfield-only features).

These foundations enabled rigorous comparison of candidate techniques across time, frequency, spatial, and power domains.

Key Normative Techniques in Release 18

The normative work specified enhancements primarily in the spatial, power, and time domains, with supporting frequency-domain and mobility features.

Spatial and Power Domain Adaptations

These deliver the largest energy-saving gains under low-to-medium load. The gNB can dynamically reduce the number of active antenna ports/elements or lower transmission power according to actual traffic and channel conditions.

• Spatial-domain (SD) adaptation includes two types: reducing the number of antenna ports based on a common CSI-RS resource, or reducing the number of antenna elements associated with ports without reducing the port count.
• Power-domain (PD) adaptation adjusts the power offset between PDSCH and CSI-RS (or absolute power levels) on a more dynamic basis.

To make these adaptations efficient and responsive, Release 18 introduced a new CSI reporting framework. The network configures multiple sub-configurations within a single CSI report setting. The UE can then report multiple CSI sub-reports in one instance, each corresponding to a different spatial or power adaptation pattern. This gives the gNB accurate, multi-hypothesis feedback so it can select the optimal reduced configuration without excessive measurement overhead.

Industry evaluations (using the 3GPP model) show that dynamic antenna and power adaptations alone can achieve 15–30 % energy savings at low-to-medium cell loads, with joint use of both techniques providing the best daily-average trade-off between savings and throughput impact. Higher gains (up to 39–56 % in specific configurations relative to a 64-Tx baseline) are reported in detailed vendor analyses when load conditions are favourable.

Time Domain: Cell DTX/DRX

Cell discontinuous transmission (DTX) and discontinuous reception (DRX) allow the gNB to enter sleep states during periods of inactivity for UEs in RRC_CONNECTED mode.

• A UE can be configured with periodic cell DTX/DRX patterns defining active and non-active periods of the gNB.
• Patterns can be activated or deactivated independently or jointly, with support for up to two patterns per MAC entity for different serving cells.
• Careful alignment with UE connected-mode DRX (C-DRX) is essential to avoid situations where the network is awake while the UE is sleeping (or vice versa).

This technique creates additional opportunities for deep sleep of radio hardware when traffic is sparse.

Frequency Domain and Supporting Features

• SSB-less secondary cell (SCell) operation is extended to inter-band carrier aggregation (under conditions of acceptable timing and power difference). This removes the need for continuous synchronization signal block transmission on the SCell, reducing always-on overhead.
• Enhancements to conditional handover (CHO) improve robustness when cells are switched off or energy-saving modes are applied, enabling reliable traffic offloading.
• Mechanisms help prevent legacy UEs from camping on energy-saving cells in ways that would undermine the savings, while still allowing access.
• Inter-node beam activation and paging enhancements support coordinated energy-saving operation across cells.

Device-Side Considerations and Low-Power Wake-Up Signal Study

While the primary focus of the network energy savings work item is the gNB, Release 18 also advanced device power efficiency. A dedicated study examined low-power wake-up signals (LP-WUS) and wake-up receivers (LP-WUR). The concept allows the main radio to remain in deep sleep (or powered off) while a simple, ultra-low-power receiver monitors a dedicated wake-up signal. Upon detection, the main radio is activated. Evaluations indicated potential UE power-saving gains approaching 90 % in idle mode under favourable conditions, though network overhead from the wake-up signals themselves must be carefully managed. Normative work on LP-WUS continues or expands in later releases.

Existing UE DRX and other power-saving features remain foundational and are designed to coexist with the new network-side techniques.

Integration with AI/ML

Network energy saving is one of the three priority use cases for the normative AI/ML support in NG-RAN. AI/ML models can generate energy-cost predictions and load forecasts that are exchanged between neighbouring nodes. These predictions enable more proactive decisions on cell activation/deactivation or resource adaptation, reducing the risk of ping-pong effects or quality degradation. The standardized data-collection procedures over Xn provide the necessary inputs.

Performance Gains, Trade-offs, and Operational Considerations

Technique	Primary Domain	Typical Energy-Saving Gain (Low–Medium Load)	Main Trade-off Considerations
Antenna + Power adaptation	Spatial / Power	15–30 % (higher in specific configs)	Potential UPT impact if adaptation is aggressive
Cell DTX/DRX	Time	Significant additional sleep opportunities	Alignment with UE DRX; latency for bursty traffic
SSB-less SCell	Frequency	Reduction of always-on signals	Timing/power difference constraints
Combined techniques	Multi-domain	Best daily-average balance	Requires careful load-aware control

Gains are highest at low-to-medium loads and diminish as cells approach busy-hour traffic. Proper configuration—guided by UE feedback, load predictions, and AI/ML assistance—is essential to keep user-perceived throughput and latency impacts minimal. Legacy-UE support and multi-vendor interoperability were explicit design requirements.

Broader Impact

These Release 18 features give operators standardized, interoperable tools to lower energy costs and carbon footprint without waiting for entirely new spectrum or site architectures. They complement earlier lean-carrier designs and vendor-specific sleep modes, while providing a clearer path for multi-vendor deployments. Combined with the AI/ML energy-saving use case, they move networks toward more autonomous, load-aware green operation. Subsequent releases continue to refine and expand these capabilities, confirming Release 18 as the foundational step for greener 5G-Advanced networks.

In essence, the green-networks and network-energy-savings work of Release 18 transforms energy efficiency from an implementation choice into a standardized, multi-domain, feedback-driven system capability that strengthens the overall 5G foundation for sustainable, high-performance operation.

---

4) Evolved Duplexing

Traditional TDD, widely deployed in both sub-6 GHz and mmWave 5G bands, allocates entire time slots (or symbols) exclusively to either downlink or uplink. This creates an inherent uplink coverage and latency bottleneck because uplink transmission opportunities are limited, especially at the cell edge or under asymmetric traffic. Release 18 addressed this by studying evolutionary steps that allow simultaneous downlink and uplink transmission within the same TDD carrier while remaining compatible with existing spectrum allocations and equipment.

Scope of the Release 18 Study

RAN1 completed the study item on evolution of NR duplex operation, documented in TR 38.858. The primary focus was Sub-band non-overlapping Full Duplex (SBFD) operated at the gNB side within a conventional unpaired (TDD) band.

In SBFD, the available frequency bandwidth of a TDD carrier is divided into non-overlapping sub-bands (separated by a guard band). One or more sub-bands are used for downlink while others are used for uplink in the same time resources. This creates “SBFD symbols/slots” in which the gNB transmits and receives simultaneously, but on different frequencies. Configurations can be semi-static or more dynamic.

The study also examined enhancements relevant to flexible and dynamic TDD operation and the associated interference challenges.

New Interference Challenges and CLI Handling

Simultaneous opposite-direction transmissions introduce several new interference types that do not exist (or are far less severe) in conventional half-duplex TDD:

• gNB self-interference (from its own downlink transmission into its uplink receiver).
• gNB-to-gNB cross-link interference (one base station’s downlink interfering with a neighbouring base station’s uplink reception).
• UE-to-UE cross-link interference (one UE’s uplink transmission interfering with a nearby UE’s downlink reception).

Proper modelling of these interferers was a major part of the study. CLI handling techniques evaluated included:

• Inter-gNB and inter-UE co-channel CLI measurement and reporting.
• Coordinated scheduling of time/frequency resources between neighbouring gNBs.
• Spatial-domain coordination (beam or nulling techniques).
• Power-control adjustments at both gNB and UE.
• Transmission and reception timing alignment considerations.

These mechanisms aim to keep residual interference manageable so that the coverage and capacity benefits of SBFD are not cancelled out.

Performance Evaluation Results

Evaluations covered system-level and link-level aspects under various deployment scenarios (urban macro, dense urban, indoor), frequency ranges (FR1 and FR2-1), and coexistence cases (single-operator non-coexistence, co-channel multi-layer, and adjacent-channel multi-operator).

Uplink Coverage Gains Link-level results for semi-static SBFD (with PUSCH repetition Type A) relative to a conventional TDD pattern (DDDSU) showed notable uplink coverage improvements, assuming realistic self-interference suppression (1 dB desense) and co-site inter-sector isolation:

• Approximately 5.41 dB gain in FR1 Urban Macro.
• Approximately 6.92 dB gain in FR2-1 Dense Urban Macro.

These gains arise primarily from the increased number of uplink transmission opportunities available to cell-edge users.

User-Perceived Throughput (UPT) System-level results for semi-static SBFD indicated UPT gains in specific scenarios. Configurations that allocate more resources to uplink (e.g., patterns with extended uplink sub-bands) deliver clear uplink UPT improvements stemming from both more uplink resources and more transmission opportunities. Downlink UPT can experience some loss depending on the exact sub-band split, residual CLI, and noise-figure effects from higher blocker power. Overall conclusions confirmed that semi-static SBFD can provide net benefits when interference is adequately controlled.

Additional academic and industry evaluations aligned with the 3GPP assumptions have reported up to fourfold uplink throughput improvements for cell-edge users under low-to-medium load when sufficiently high self-interference cancellation (around 145–149 dB in high-power macro scenarios) is achieved.

Practical Considerations and Limitations

• Self-interference cancellation at the gNB remains the most critical implementation requirement. High isolation (spatial, analog, and digital) is needed, especially in high-power macro deployments.
• Guard-band sizing and sub-band configuration flexibility must balance spectral efficiency against residual interference.
• Legacy UE compatibility and multi-operator coexistence (especially adjacent-channel) require careful RF and scheduling design.
• Dynamic versus semi-static operation offers different trade-offs between adaptability to traffic and signalling/complexity overhead.
• The study prioritized feasibility assessment; full normative specification of many SBFD features progressed into Release 19.

Path Toward More Advanced Full-Duplex Operation

Release 18 deliberately positioned SBFD as an evolutionary stepping stone rather than a complete in-band full-duplex solution. By solving simultaneous transmission and reception on non-overlapping sub-bands within existing TDD spectrum, it reduces the technical and regulatory barriers to true single-frequency full duplex in later releases and toward 6G. The interference modelling, measurement frameworks, and performance baselines established in TR 38.858 provide essential input for subsequent work on in-band full duplex, more advanced CLI mitigation, and AI-assisted duplex adaptation.

Summary of Key Aspects

Aspect	Key Points in Release 18
Core Concept	Sub-band non-overlapping full duplex (SBFD) at gNB
Primary Benefit	Improved uplink coverage and more flexible UL/DL allocation
Main Challenges	Self-interference, gNB-gNB CLI, UE-UE CLI
Coverage Gain Example	~5.4 dB (FR1 UMa), ~6.9 dB (FR2-1 Dense UMa)
Configuration Options	Semi-static and dynamic/flexible patterns
Status	Study completed (TR 38.858); normative work continued later

In summary, the evolved duplexing work of Release 18 provides a practical, standards-based pathway to overcome long-standing TDD uplink limitations while remaining compatible with deployed spectrum and equipment. By introducing SBFD, rigorously evaluating interference, and quantifying coverage and throughput benefits, it strengthens the radio foundation of 5G-Advanced and creates a clear technical trajectory toward more capable full-duplex systems in the future.

---

5) MIMO Evolution

Massive MIMO has been a cornerstone of 5G New Radio since Release 15. Release 18 continues its evolution with targeted refinements that improve spectral efficiency, multi-user capacity, high-mobility performance, and support for advanced device types such as fixed wireless access (FWA), customer premises equipment (CPE), vehicles, and industrial terminals. The work concentrates on four interrelated areas: CSI, multi-TRP, uplink transmission, and reference signals.

CSI Enhancements

Accurate and timely CSI remains essential for effective beamforming and multi-user pairing. Release 18 addresses two persistent challenges: performance under medium-to-high UE velocity and support for coherent joint transmission (CJT).

• High- and medium-velocity scenarios (primarily sub-7 GHz)
  • Conventional CSI assumes relatively static channels within the reporting period. Mobility causes channel aging that degrades precoding quality. Rel-18 introduces mechanisms that exploit time-domain correlation, including CSI prediction and Doppler/time-domain compression within the enhanced Type-II codebook framework. An enhanced Type-II codebook allows the UE to report both current and predicted precoder information, improving downlink throughput for UEs moving at typical urban speeds (roughly 30–60 km/h).
• Coherent joint transmission support
  • Enhancements improve the quality and overhead of CSI feedback needed when multiple TRPs transmit coherently to the same UE. This enables better phase and amplitude alignment across TRPs under ideal backhaul and synchronization assumptions.

These CSI improvements reduce feedback overhead relative to the performance gain and make multi-user MIMO more robust in realistic mobility conditions.

Multi-TRP Enhancements

Multi-TRP operation improves both capacity (through spatial multiplexing) and reliability (through diversity or joint transmission). Release 18 significantly expands the unified Transmission Configuration Indicator (TCI) framework that was introduced in Release 17 for single-TRP cases.

Key advances include:

• Extension of the unified TCI framework to support multiple simultaneous downlink and uplink TCI states, enabling more flexible multi-TRP beam indication.
• Support for two timing advances (dual TA) to handle asynchronous multi-TRP deployments where TRPs have different propagation delays.
• Coherent joint transmission for up to four TRPs in sub-7 GHz, assuming ideal backhaul and synchronization.
• Asynchronous multi-TRP support with corresponding power-control enhancements for both single-DCI and multi-DCI scheduling.
• Simultaneous multi-panel uplink transmission, particularly valuable for mmWave and multi-TRP scenarios involving CPE, FWA, vehicle, or industrial devices.

These features allow networks to exploit denser TRP deployments more effectively while maintaining manageable signalling complexity.

Uplink MIMO Evolution

Uplink has historically lagged downlink in MIMO order. Release 18 closes much of that gap for capable devices.

• Support for up to rank-8 / 8-layer PUSCH transmission.
• Multi-panel simultaneous uplink transmission.
• New and enhanced codebook types tailored for 8 Tx operation, covering different coherence assumptions across antenna ports or panels (fully coherent, partial coherent within groups, and non-coherent).
• Enhancements to full-power transmission modes (Modes 0/1/2) so that maximum power can still be achieved under the new higher-rank and multi-panel configurations.
• SRS resource and port expansions to support the higher-rank operation and better channel sounding.

These capabilities primarily target devices with more flexible antenna configurations (CPE, FWA, vehicles, industrial sensors) rather than typical smartphones, delivering substantial uplink throughput and reliability gains in those segments.

Reference Signal Enhancements

Reference signals underpin both CSI acquisition and data demodulation.

• Demodulation Reference Signal (DMRS) The number of orthogonal DMRS ports for multi-user MIMO is increased to 24 for both downlink and uplink in CP-OFDM. This directly enables higher multi-user pairing capacity without a proportional increase in DMRS overhead (achieved through techniques such as longer orthogonal cover codes).
• Sounding Reference Signal (SRS) Improvements help randomize interference among different TRPs and support the expanded uplink MIMO configurations.

Higher DMRS port counts translate into greater multi-user MIMO capacity, while improved SRS designs enhance uplink channel quality estimation in multi-TRP environments.

Summary of Key MIMO Evolution Features

Area	Main Enhancements	Primary Benefits	Target Scenarios
CSI	Time-domain prediction/compression, enhanced Type-II	Better performance under mobility, lower aging	Medium/high-velocity UEs, sub-7 GHz
Multi-TRP	Unified TCI multi-state, dual TA, CJT up to 4 TRPs	Higher capacity & reliability, flexible beams	Dense multi-TRP, asynchronous deployments
Uplink MIMO	Rank-8 / 8 Tx, multi-panel simultaneous Tx	Higher UL throughput & coverage	CPE, FWA, vehicles, industrial devices
Reference Signals	Up to 24 orthogonal DMRS ports, improved SRS	Higher MU-MIMO capacity, better interference randomization	Multi-user and multi-TRP cells

System Impact and Strengthening of the 5G Foundation

Collectively these enhancements raise spectral efficiency, improve cell-edge and high-mobility user experience, and expand the practical applicability of advanced MIMO to a wider range of devices and deployment topologies. They build directly on the multi-TRP and CSI frameworks of Releases 16 and 17 while removing key limitations (rank, port count, mobility robustness, and multi-TRP signalling flexibility).

The result is a more capable radio foundation that can support higher traffic densities, more demanding industrial and fixed-access use cases, and smoother evolution toward further AI-assisted or even more advanced multi-TRP techniques in later releases. MIMO evolution in Release 18 therefore remains one of the most tangible performance-oriented contributions to the overall strengthening of the end-to-end 5G system.

---

6) AI/ML for the NR Air Interface and NG-RAN

Release 18 is the first 3GPP release in which AI/ML transitions from high-level principles into concrete study outcomes for the air interface and normative data-collection and signalling support inside the NG-RAN. The work deliberately separates the physical-layer/air-interface domain from the system-level RAN domain while establishing common functional concepts that can later converge.

AI/ML for the NR Air Interface (RAN1 Study)

The Release 18 study item on AI/ML for the NR air interface (primarily documented in TR 38.843) defined a general functional framework and evaluated three representative use cases. It examined both one-sided models (inference performed entirely at the UE or entirely at the gNB) and two-sided models (for example, an encoder at the UE and a decoder at the gNB).

Functional Framework

The framework comprises data collection, model training, model inference, and an actor that applies the inference results. Supporting aspects include lifecycle management signalling, UE capability indication, model identification and meta-information, data generation and transfer mechanisms, generalization testing across environments and vendors, and interoperability considerations.

Three Primary Use Cases

1. CSI Feedback Enhancement
   • Spatial-frequency domain compression via two-sided models to reduce uplink feedback overhead while preserving reconstruction accuracy.
   • Time-domain CSI prediction via one-sided models to mitigate channel aging, particularly under medium-to-high UE velocity. Evaluations demonstrated substantial overhead reduction (commonly reported in the range of 8–79 %) accompanied by modest throughput gains, depending on scenario, model complexity, and generalization conditions.
2. Beam Management
   • Spatial-domain downlink beam prediction (selecting the best beam from a reduced measurement set).
   • Temporal-domain beam prediction (forecasting future best beams to lower measurement latency and overhead). These techniques are especially relevant for FR2 (mmWave) systems that rely on analog beamforming and exhaustive beam sweeping.
3. Positioning Accuracy Enhancements
   • Direct AI/ML positioning, in which a model outputs location estimates directly from measurements.
   • AI/ML-assisted positioning, in which models refine conventional measurements such as time-of-arrival, angle-of-arrival, or multi-RTT. Simulations indicated potential for significantly improved horizontal accuracy (for example, better than 1 m at the 90th percentile in certain indoor scenarios) compared with purely conventional methods under the same conditions.

The study also assessed model complexity, data requirements, generalization performance, and the potential specification impact on the physical layer and higher-layer protocols. Full normative work for many of these air-interface techniques continues into later releases.

AI/ML for NG-RAN (RAN3 Normative Work)

Building on the Release 17 study (TR 37.817), Release 18 delivered the first normative specifications for AI/ML support inside the NG-RAN. The scope is limited primarily to the non-split architecture and focuses on three high-priority use cases that align with self-organizing network objectives.

Core Enablers

• Enhanced data-collection and reporting procedures over the existing Xn interface (Data Collection Reporting Initiation and Reporting procedures).
• Exchange of measurements, predictions, and feedback between neighbouring NG-RAN nodes.
• Introduction of an “Energy Cost” metric for energy-related decisions.
• Support for two principal deployment scenarios: model training located in OAM with inference at the NG-RAN node, or both training and inference located at the NG-RAN node.

Three Standardized Use Cases

• Network Energy Saving
  • AI/ML models leverage collected and predicted energy-cost and load information to decide cell or node activation/deactivation. The objective is to reduce overall energy consumption across a geographic area while safeguarding coverage and quality of service. High-level actions (rather than fine-grained discontinuous transmission/reception control) were prioritized.
• Load Balancing
  • Models generate load predictions that are exchanged between nodes. Combined with UE performance feedback after offloading actions, these predictions enable more proactive traffic steering and resource distribution across cells, carriers, or areas.
• Mobility Optimization
  • UE trajectory predictions are generated and can be included in handover-request messages. The target node uses the predicted path to select better handover targets or prepare resources in advance, improving handover success rates and reducing unnecessary handovers or radio-link failures. Actual trajectory feedback can later be used to monitor prediction accuracy.

Signalling procedures are deliberately use-case and data-type agnostic, providing implementation flexibility while ensuring interoperable exchange of the essential information.

Common Themes, Benefits, and Constraints

Both the air-interface study and the NG-RAN normative work share a common philosophy: AI/ML is introduced as an augmenting capability rather than a wholesale replacement of existing algorithms. Standardized frameworks for data collection, model lifecycle aspects, and inter-node or UE-network signalling create the necessary interoperability foundation for multi-vendor deployments.

Benefits

• Air interface: lower feedback overhead, improved robustness under mobility, higher positioning accuracy, and reduced beam-management latency.
• NG-RAN: more proactive and accurate decisions for energy efficiency, traffic distribution, and mobility, reducing ping-pong effects and improving overall network efficiency.
• System-level: a clear path toward progressive automation of RAN operations while preserving the standards-based, multi-vendor nature of 5G.

Practical Constraints

• Model generalization across vendors, environments, and UE types remains challenging.
• Data quality, labelling, privacy, and collection overhead must be carefully managed.
• Computational and energy cost of inference (especially on UEs) must be balanced against performance gains.
• Rel-18 air-interface work was largely a study; normative physical-layer impact is limited and expands later. NG-RAN support is currently restricted mainly to non-split architecture.

Summary of Release 18 AI/ML Scope

Domain	Nature of Work	Primary Use Cases	Key Enablers
NR Air Interface	Study (TR 38.843)	CSI feedback, beam management, positioning	Functional framework, one-/two-sided models, evaluation methodology
NG-RAN	Normative	Network energy saving, load balancing, mobility optimization	Xn data-collection procedures, Energy Cost metric, prediction exchange

Evolution Beyond Release 18

Release 19 builds directly on these foundations. For the air interface, normative work progresses on two-sided models and additional use cases. For NG-RAN, support is extended to network slicing, coverage and capacity optimization, continuous minimization of drive tests across RRC states, mobility optimization for NR dual connectivity, and initial considerations for split architecture. Management frameworks (developed in SA5) continue to mature to support consistent lifecycle handling across the 5G System.

In summary, the AI/ML work for the NR air interface and NG-RAN in Release 18 establishes the first standardized scaffolding—functional frameworks, carefully scoped use cases, data-exchange procedures, and evaluation methodologies—upon which more ambitious data-driven capabilities can be built. It transforms AI/ML from a research topic into an interoperable system feature that strengthens both the radio interface and the broader RAN foundation of 5G-Advanced.

---

7) Energy Efficiency and Network Energy Savings

Energy efficiency moved from an implementation concern to a standardized system capability in Release 18. Rising operational costs, sustainability targets, and the recognition that the RAN accounts for the majority of network energy consumption drove both radio-layer innovations and management-plane enhancements. The work spans RAN1/RAN2/RAN3 (network energy savings for NR), SA5 (EE KPIs, measurements, and energy-saving use cases), and related studies on new aspects of EE for 5G Phase 2.

RAN Network Energy Savings for NR

The primary normative effort focused on enabling the gNB to adapt transmissions and receptions more dynamically across multiple domains while protecting user experience and supporting legacy devices.

Study Foundations

A dedicated study defined a common base-station energy consumption model, evaluation methodology, and KPIs that balance energy savings against user-perceived throughput, latency, capacity, spectral efficiency, and UE power consumption. Priority scenarios were idle/empty and low-to-medium load, with the requirement that legacy UEs generally remain able to access networks using the new techniques.

Key Techniques

• Spatial and power domain adaptations
  • The gNB can dynamically reduce the number of active antenna ports or elements, or lower transmission power, according to traffic and channel conditions. A new CSI reporting framework allows the UE to report multiple CSI sub-reports in one instance, each corresponding to different spatial or power adaptation hypotheses. This enables the network to select optimal reduced configurations efficiently. These techniques deliver the largest gains—typically 15–30 % energy savings under low-to-medium load, with higher figures reported in specific configurations when antenna and power adaptation are combined.
• Time domain: Cell DTX/DRX
  • Cell discontinuous transmission and reception allow the gNB to enter sleep states during inactivity periods for UEs in RRC_CONNECTED mode. Configurable active/non-active patterns (up to two per MAC entity) can be activated independently or jointly and should be aligned with UE connected-mode DRX to avoid mismatched wake periods.
• Frequency domain and supporting features
  • SSB-less secondary cell operation is extended to inter-band carrier aggregation (under acceptable timing and power difference conditions). Conditional handover enhancements improve robustness when cells enter energy-saving modes. Mechanisms help manage legacy UE camping behaviour, and inter-node coordination supports beam activation and paging in energy-saving contexts.

These multi-domain tools give operators standardized means to create additional sleep opportunities and reduce always-on overhead without requiring new spectrum or wholesale hardware replacement.

Management and KPI Frameworks (SA5)

Parallel work in SA5 extended energy efficiency beyond the radio layer into measurable, optimizable system properties.

• EE KPIs and measurements
  • KPIs are defined for NG-RAN, 5GC, and network slices, typically expressed as performance (data volume or coverage area) divided by energy consumption. New or refined measurements cover physical network function (PNF) power and energy consumption, virtualized network function (VNF) energy consumption (including improved accuracy for virtual CPUs and containerized functions), and slice-level energy efficiency (with variants for eMBB, URLLC, and other slice types).
• Energy-saving use cases and solutions
  • Specifications address energy-saving scenarios applying to NG-RAN, 5GC, and/or network slicing. Operators can express energy-efficiency requirements for slices (aligned with external frameworks such as GSMA NG.116) and receive periodic reports of actual efficiency. Management services support monitoring, optimization, and energy-saving actions.
• Study on new aspects of EE for 5G Phase 2
  • A dedicated study examined additional key issues related to energy efficiency, energy saving, and digital sobriety, with conclusions feeding into normative enhancements.

These management capabilities turn energy efficiency into an observable and controllable attribute of the 5G System rather than a purely local radio implementation choice.

Device-Side Power Savings

While the network energy savings work item prioritizes the gNB, Release 18 also advanced device efficiency. Existing UE discontinuous reception mechanisms remain foundational. A study on low-power wake-up signals (LP-WUS) and wake-up receivers evaluated the potential for the main radio to remain in deep sleep while a simple ultra-low-power receiver monitors a dedicated wake-up signal. Evaluations indicated substantial UE power-saving potential (approaching 90 % in favourable idle-mode conditions), although network overhead from the wake-up signals themselves must be managed. Normative follow-up continues in subsequent releases.

Integration with AI/ML

Network energy saving is one of the three priority use cases for normative AI/ML support in NG-RAN. Models can generate energy-cost predictions and load forecasts exchanged between neighbouring nodes via standardized Xn data-collection procedures. These predictions enable more proactive activation/deactivation decisions and help avoid quality degradation or ping-pong effects. Management frameworks further support AI/ML-assisted energy optimization using analytics from functions such as the Management Data Analytics Function or NWDAF.

Performance Gains, Trade-offs, and Operational Considerations

Domain / Technique	Typical Benefit (Low–Medium Load)	Main Considerations
Antenna + power adaptation	15–30 % (higher in joint use)	Potential throughput impact if aggressive
Cell DTX/DRX	Additional deep-sleep opportunities	Alignment with UE DRX; latency for bursty traffic
SSB-less SCell	Reduced always-on signalling	Timing/power difference constraints
Combined multi-domain techniques	Best daily-average trade-off	Load-aware control required
Management KPIs & slice EE	Measurable, optimizable EE	Accuracy of VNF/container measurements

Gains are highest at low-to-medium loads and require careful, load-aware configuration to keep user experience impacts minimal. Multi-vendor interoperability and legacy-UE support were explicit design goals.

Broader System Impact

Release 18 elevates energy efficiency from a vendor-specific optimization to a standardized, multi-domain, measurable system capability. RAN techniques reduce the dominant energy consumer (the radio access network), while management frameworks provide the visibility and control needed for end-to-end optimization across RAN, core, and slices. Combined with AI/ML assistance, these features support more autonomous, sustainable network operation and lower total cost of ownership. Subsequent releases continue refining both the radio techniques and the management plane, confirming Release 18 as the foundational step for greener 5G-Advanced networks.

In essence, the energy efficiency and network energy savings work of Release 18 delivers practical, interoperable tools that strengthen the overall 5G foundation for high-performance, sustainable operation.

---

8) Duplex Evolution, Coverage, and Deployment Flexibility

Release 18 treats spectrum utilization, coverage robustness, and flexible topology as complementary levers. Evolving duplexing creates more uplink opportunities within existing TDD bands; coverage enhancements improve reach and reliability; and new deployment nodes allow operators to extend or densify coverage cost-effectively without always requiring new fibre backhaul or full base stations.

Duplex Evolution

Traditional TDD allocates entire time resources exclusively to downlink or uplink, creating an inherent uplink coverage and latency bottleneck. The Release 18 study on evolution of NR duplex operation (TR 38.858) focused primarily on Sub-band non-overlapping Full Duplex (SBFD) operated at the gNB within a conventional unpaired TDD carrier.

In SBFD, the carrier bandwidth is divided into non-overlapping sub-bands separated by a guard band. Downlink and uplink can then occur simultaneously on different sub-bands within the same time resources. Configurations may be semi-static or more dynamic. The study also examined enhancements relevant to flexible and dynamic TDD.

Interference Challenges and Mitigation

Simultaneous opposite-direction transmissions introduce self-interference at the gNB, gNB-to-gNB cross-link interference (CLI), and UE-to-UE CLI. The study evaluated modelling of these interferers and a range of handling techniques, including CLI measurement and reporting, coordinated scheduling, spatial-domain coordination, power control, and timing considerations.

Performance Outcomes

Link-level evaluations of semi-static SBFD (with PUSCH repetition) relative to a conventional TDD pattern showed uplink coverage gains of approximately 5.4 dB in FR1 Urban Macro and 6.9 dB in FR2-1 Dense Urban Macro under realistic self-interference suppression and isolation assumptions. System-level results indicated user-perceived throughput gains in specific scenarios, driven by increased uplink transmission opportunities, with some downlink trade-offs depending on the exact sub-band split and residual CLI.

SBFD is positioned as an evolutionary step rather than full in-band full duplex. It improves uplink coverage and resource flexibility within existing TDD spectrum while establishing interference frameworks and performance baselines for more advanced full-duplex solutions in later releases.

Coverage Enhancements

Uplink coverage has been a persistent challenge, particularly at the cell edge and in mid-band or higher-frequency deployments. Release 18 addresses this through multiple complementary means.

• Duplex evolution (SBFD) directly increases the number of uplink transmission opportunities, delivering the multi-dB coverage gains noted above.
• MIMO evolution contributes through higher-rank uplink transmission (up to rank 8), multi-panel simultaneous uplink, improved CSI for medium-to-high velocity, and expanded DMRS ports that support more robust multi-user and multi-TRP operation.
• Additional coverage-related work includes further uplink coverage enhancements (for example, in the context of NTN and commercial smartphone-class devices), random access enhancements supporting multiple PRACH transmissions with the same beam, and refinements that improve reliability under challenging channel conditions.

Collectively these improvements reduce the uplink bottleneck, improve cell-edge user experience, and support more consistent service quality across diverse deployment scenarios.

Deployment Flexibility

Two major tools expand the operator’s ability to extend or densify coverage cost-effectively.

Mobile Integrated Access and Backhaul (IAB)

Mobile IAB introduces a mobile IAB-node that provides NR access links to UEs while maintaining an NR backhaul link to a parent node and that can physically move across the RAN area. The node comprises a mobile IAB-MT and a mobile IAB-DU. Procedures cover inter-donor migration, radio-link failure recovery, interference mitigation, and support for legacy UEs. Mobile IAB enables temporary or vehicle-mounted coverage extension (for example, for events, disaster recovery, or moving platforms) while reusing the existing NR access and backhaul framework.

Network-Controlled Repeaters (NCR)

NCR evolves traditional RF repeaters into network-managed devices. An NCR-node consists of an NCR-MT (partial UE functionality that receives side-control information from the gNB over a control link) and an NCR-Fwd (the amplifying-and-forwarding function). Side-control information enables the network to manage beamforming, TDD configuration awareness, and on/off behaviour. This allows more intelligent, interference-aware coverage extension on both FR1 and FR2 without the full protocol complexity of an IAB node. NCR is specified primarily for stationary, single-hop operation.

These tools expand the densification toolbox: NCR for low-complexity, network-controlled amplification and beamforming; mobile IAB for movable, decode-and-forward coverage with backhaul reuse. Both reduce reliance on new fibre or full gNB deployments in challenging locations.

Interrelationships and System Impact

Aspect	Primary Contribution	Key Benefit
Duplex Evolution (SBFD)	Simultaneous DL/UL on non-overlapping sub-bands	Improved UL coverage and resource flexibility within TDD spectrum
Coverage Enhancements	More UL opportunities + stronger MIMO/RA tools	Better cell-edge performance and reliability
Deployment Flexibility	Mobile IAB + Network-Controlled Repeaters	Cost-effective, rapid, or temporary coverage extension

Duplex evolution creates more uplink capacity within existing carriers; coverage enhancements make that capacity usable at the cell edge and under mobility; and flexible deployment nodes allow operators to place capacity and coverage where they are most needed without traditional infrastructure constraints. Together they strengthen the radio foundation by improving spectral utilization, reach, and topological adaptability while remaining compatible with deployed spectrum and multi-vendor environments.

Release 18 therefore delivers practical, standardized levers that address three of the most common operational challenges—uplink limitations of TDD, coverage consistency, and densification cost—laying a more robust platform for both near-term commercial use and further evolution toward advanced full-duplex and denser 5G-Advanced deployments.

---

9) Mobility Improvements

Traditional L3 handovers involve substantial signalling, measurement reporting delays, and interruption times that can reach tens of milliseconds (sometimes reported around 50–90 ms in challenging cases). This can degrade experience for latency-sensitive services such as extended reality (XR), ultra-reliable low-latency communications, and high-mobility scenarios. Release 18 introduces LTM to address these limitations by shifting key triggering and execution steps to lower layers while retaining higher-layer preparation.

Core Concept of L1/L2 Triggered Mobility (LTM)

LTM enables the network to change a UE’s serving cell using a lightweight MAC Control Element (MAC CE) cell-switch command, based primarily on Layer 1 measurements (and, in some cases, L3 reports). The candidate cell configurations are prepared and provided to the UE in advance via RRC signalling.

Key principles include:

• Pre-configuration of one or more LTM candidate cells by the source gNB.
• Early downlink and uplink synchronization (including timing advance acquisition) with candidate cells before the switch command.
• Triggering via L1 measurement reports (periodic, semi-persistent, or aperiodic) that are faster and more sensitive to channel changes than conventional L3 reports.
• Execution through a MAC CE that indicates the target LTM candidate configuration.
• Maintenance of security keys across the cell switch in many cases, avoiding full re-keying overhead.
• Support for subsequent LTM procedures.

In Release 18, LTM is limited to intra-gNB (intra-CU) mobility, covering both intra-DU and inter-DU cases within the same central unit. It supports intra-frequency and inter-frequency mobility, including cases where the target is not a current serving cell, and works with carrier aggregation and dual connectivity scenarios (for example, PCell change, SCell changes, and certain SN changes without MN involvement). It is specified for licensed spectrum.

Enabling Mechanisms

To make L1/L2-based mobility practical, Release 18 specifies:

• Non-serving cell beam management — Activation and management of Transmission Configuration Indicator (TCI) states for candidate cells in advance, allowing the UE to maintain beam readiness before becoming the serving cell.
• Non-serving cell timing advance (TA) acquisition — Early uplink synchronization so that random access can often be avoided (RACH-less handover) when TA is already known.
• Simplified handover execution that reduces the need for full random-access procedures to the target cell in many situations.
• Compatibility with existing connectivity techniques such as carrier aggregation and dual connectivity.

These mechanisms significantly cut the latency associated with measurement reporting and the interruption associated with reconfiguration and re-synchronization.

Performance Benefits

Evaluations and early field results highlight two primary improvements:

• Reduced interruption time — Pre-configuration, early synchronization (both DL and UL), early decoding of configurations, and lightweight MAC CE triggering typically bring interruption times down to the 20–30 ms range (or lower in optimized cases), compared with higher values for baseline L3 handover. Some analyses and trials report reductions of up to 25–40 % in data interruption during cell change.
• Improved mobility robustness — Triggering based on more sensitive L1 measurements allows earlier and more accurate decisions, reducing the likelihood of radio-link failure and subsequent re-establishment. This further contributes to lower effective interruption and higher success rates.

The combination makes LTM particularly valuable for services requiring consistent high bit rates and low latency during mobility, including XR, cloud applications, and critical IoT.

Supported Scenarios and Limitations in Release 18

Aspect	Release 18 Support
Scope	Intra-gNB (intra-CU) mobility only
Frequency	Intra- and inter-frequency
Architecture	Intra-DU and inter-DU within the same CU
Connectivity	Non-CA, CA, and certain dual-connectivity cases
Spectrum	Licensed spectrum
Measurement triggers	Periodic / semi-persistent / aperiodic L1 reports (plus L3 in some cases)
Future evolution (Rel-19+)	Inter-gNB / inter-CU LTM, enhanced event-triggered reporting, and further refinements

Broader Context and System Impact

LTM builds on earlier mobility robustness features (conditional handover, dual active protocol stack, beam-based mobility) while providing a more fundamental reduction in latency and overhead. It complements other Release 18 foundation enhancements such as advanced MIMO (better channel quality under mobility), AI/ML-assisted mobility optimization (trajectory prediction and smarter target selection), and energy-saving modes that must coordinate with mobility procedures.

By making handovers faster and more reliable, LTM strengthens the overall user experience and network resilience, particularly as 5G networks support denser deployments, higher frequencies, and more demanding applications. Early commercial trials (for example, involving major operators and vendors) have already demonstrated practical gains, confirming its readiness as a key 5G-Advanced capability.

In summary, the mobility improvements of Release 18—centred on L1/L2 Triggered Mobility—deliver a practical, standards-based reduction in handover interruption and overhead while improving robustness. They form an essential part of the stronger, more responsive radio foundation that Release 18 provides for the continued evolution of 5G.

---

10) Enhanced Reduced Capability (eRedCap)

eRedCap (also called enhanced RedCap or eNR RedCap) builds directly on the Reduced Capability (RedCap) framework standardized in Release 17. While RedCap already lowered complexity relative to full 5G NR (targeting roughly the performance range of LTE Categories 1–4), eRedCap applies additional constraints to reach an even lower cost and power profile suitable for high-volume, modest-throughput IoT use cases. The specifications were completed as part of the Release 18 freeze in 2024.

Motivation and Positioning

Full 5G NR devices support high peak rates, multiple antennas, wide bandwidths, and advanced features that drive up modem complexity, cost, and power consumption. Many IoT applications (sensors, meters, trackers, basic industrial devices) do not need those capabilities. Release 17 RedCap addressed mid-tier needs with peak rates typically in the tens to low hundreds of Mbps. eRedCap targets the next tier down—devices whose requirements align with LTE Category 1 or Category 1bis (around 10 Mbps downlink)—while running on modern 5G Standalone (SA) infrastructure. This enables operators and device makers to migrate large existing LTE IoT fleets to 5G without the cost of full NR or even full RedCap hardware, while gaining access to network slicing, improved positioning, and long-term spectrum support.

Key Technical Features and Reductions

eRedCap devices inherit many RedCap simplifications (reduced antenna configurations, optional half-duplex FDD, relaxed processing in some areas) and add further constraints:

• Peak data rate capping
  • All eRedCap devices are limited to a peak rate of 10 Mbps in both downlink and uplink, regardless of other supported features. This is a hard limit that enables simpler baseband processing, reduced memory, and lower-cost silicon compared with RedCap (where peak rates depend on antenna configuration and can reach well above 100 Mbps downlink in many designs).
• Baseband bandwidth reduction for data channels
  • The maximum RF bandwidth remains up to 20 MHz (FR1), allowing compatibility with network control signalling and system information. However, the baseband bandwidth used for data channels (PDSCH and PUSCH) can be limited to the equivalent of approximately 5 MHz. This is achieved by restricting the number of physical resource blocks (for example, 25 PRBs at 15 kHz subcarrier spacing or 12 PRBs at 30 kHz). Control channels and signals may still use the wider bandwidth. Some implementations support peak-rate limiting without the full 5 MHz baseband reduction, giving design flexibility.
• Antenna configuration
  • Primarily targets 1 transmit / 1 receive (1T1R). This is simpler and lower-cost than the 1T2R option commonly associated with higher-end RedCap designs.
• Frequency range
  • Focused on FR1 (sub-6 GHz). FR2 (mmWave) support is not a primary target for the lowest-complexity eRedCap tier.
• Duplex operation
  • Supports FDD, TDD, and half-duplex FDD (HD-FDD). HD-FDD further reduces hardware complexity by avoiding simultaneous transmit and receive.
• Modulation and other relaxations
  • Typical support includes up to 64QAM downlink and 16QAM uplink in many configurations, with relaxed processing timelines in certain procedures to accommodate simpler processors.
• Power-saving enhancements
  • Extended discontinuous reception (eDRX) in the RRC_INACTIVE state is expanded beyond the Release 17 limit of 10.24 seconds, up to the same long cycles available in RRC_IDLE (approximately 3 hours / up to 10485.76 seconds). This allows devices to remain registered in a low-power suspended state for much longer periods, improving battery life for intermittent traffic. Power Saving Mode (PSM) and related mechanisms continue to apply.

A new RAT type and capability signalling allow networks to identify eRedCap devices, apply access restrictions if needed (for example, based on number of receive branches), and manage roaming or slicing appropriately.

Comparison with RedCap and LTE Categories

Parameter	LTE Cat-1 / Cat-1bis	RedCap (Rel-17)	eRedCap (Rel-18)
Peak DL rate	~10 Mbps	Typically 50–150+ Mbps	≤ 10 Mbps
Peak UL rate	~5 Mbps	Tens of Mbps	≤ 10 Mbps
Typical bandwidth	20 MHz (LTE)	Up to 20 MHz (FR1)	RF up to 20 MHz; data ~5 MHz optional
Receive antennas	1 or 2	1 or 2	Primarily 1
Network generation	4G LTE	5G SA	5G SA
eDRX in RRC_INACTIVE	Limited	Up to 10.24 s	Up to ~3 hours
Network slicing / advanced 5G features	Limited or none	Supported	Supported
LTE sunset risk	High	None	None

eRedCap is designed to align hardware requirements with LTE Cat-1/1bis, facilitating dual-mode implementations that ease migration.

Target Use Cases and Benefits

eRedCap is aimed at high-volume, cost-sensitive IoT applications that need reliable connectivity in the low-to-moderate data-rate range but do not require the higher throughput of full RedCap or NR. Typical examples include:

• Smart metering and utility sensors
• Asset tracking and fleet telematics
• Industrial wireless sensors and smart-grid devices
• Basic wearables, point-of-sale terminals, and environmental monitors
• Other machine-type communications with intermittent or modest traffic

Benefits include:

• Lower device bill-of-materials and form-factor cost through simpler RF, baseband, and antenna designs.
• Improved battery life via extended eDRX in inactive state and overall reduced complexity.
• Future-proof operation on 5G SA networks, including access to network slicing, improved positioning, and long-term spectrum strategies, without exposure to LTE network sunsetting.
• Compatibility with existing 5G infrastructure once operators enable RedCap/eRedCap support (same core network as full NR and RedCap).

Network and Deployment Considerations

eRedCap requires a 5G Standalone core. Operators can control access and resource allocation for these devices. Because the RF bandwidth remains compatible with RedCap/NR control signalling, networks can support mixed populations of full NR, RedCap, and eRedCap devices. Module and chipset availability is expected to ramp from the mid-to-late 2020s as the ecosystem matures.

Summary

Enhanced Reduced Capability (eRedCap) in Release 18 represents a deliberate further step in the 5G device continuum. By capping peak rates at 10 Mbps, optionally narrowing data-channel baseband bandwidth to approximately 5 MHz, emphasising single-antenna designs, and extending power-saving cycles, it delivers a cost- and energy-efficient 5G option that closely matches the performance envelope of widely deployed LTE Cat-1/1bis devices. At the same time, it provides the benefits of modern 5G infrastructure. Together with Release 17 RedCap, eRedCap significantly expands the addressable 5G IoT market and strengthens the proliferation of 5G to virtually all device classes and use cases.

---

11) XR (Extended Reality) Enhancements

XR traffic is characterized by high data rates, strict low latency, periodic or quasi-periodic frame arrivals (often at non-integer millisecond intervals corresponding to frame rates such as 60, 90, or 120 fps), jitter, and the need to treat related packets as coherent application units (for example, a video frame or slice). Legacy 5G mechanisms were not fully optimized for these patterns, limiting the number of simultaneous XR users, increasing device power consumption, and risking quality degradation under congestion. Release 18 introduces the first set of targeted normative enhancements based on prior studies.

Key Concepts Enabling XR Support

Two foundational concepts defined or refined in the system architecture (primarily SA2) and leveraged by the RAN are central to many of the radio enhancements:

• PDU Set — One or more Protocol Data Units (PDUs) that together carry the payload of a single application-level information unit (for example, a video frame, audio frame, or slice). New QoS parameters such as PDU Set Delay Budget (PSDB) and PDU Set Error Rate allow the network to treat the entire set as an integrated unit rather than individual packets.
• Data Burst — A set of data PDUs generated and sent by the application in a short time window.

These concepts enable the network to identify application-meaningful units, apply differentiated handling (prioritization or discarding of incomplete sets), and optimize scheduling and resource use.

Main Categories of RAN Enhancements

Release 18 XR work (spanning RAN1, RAN2, and related groups) organizes enhancements into several complementary areas.

1. Application Awareness (XR Awareness)

The network gains better visibility into XR traffic characteristics so it can schedule more intelligently.

• Extended UE Assistance Information (UAI) and related reporting allow the UE to provide XR-specific details such as buffer status, expected arrival times, periodicity, jitter, and PDU Set Importance (PSI) or related indicators.
• In both uplink and downlink, awareness of PDU Sets and Data Bursts helps the gNB optimize radio resource allocation, prioritize important content, and discard remaining PDUs of a set once one is known to be lost or discarded (freeing resources).
• Congestion and delay feedback mechanisms (including support related to L4S—Low Latency, Low Loss, Scalable Throughput) help applications adapt rates in real time.

This awareness is particularly valuable under load, improving user experience by protecting critical application units.

2. Capacity Enhancements

These increase the number of simultaneous XR users a cell can support, especially in the uplink where AR applications can generate significant traffic.

• Configured Grant (CG) enhancements — Support for multiple PUSCH transmission occasions within a single CG period, providing periodic low-overhead uplink resources matched to XR traffic patterns.
• Unused Transmission Occasion indication — The UE can signal via Uplink Control Information (UCI, sometimes called UTO-UCI) which upcoming CG occasions will not be used. The gNB can then reallocate those resources to other users.
• PDU Set-based discarding and prioritization further improve efficiency by avoiding transmission of incomplete or low-priority application units.

System-level evaluations of these techniques have shown capacity gains (examples in the literature range from approximately 11 % to substantially higher figures depending on scenario and traffic model) compared with conventional dynamic grant scheduling.

3. Power Saving Enhancements

XR devices (especially wearables and glasses) are highly sensitive to battery life and thermal constraints.

• DRX adaptations for XR frame rates — Connected-mode Discontinuous Reception (DRX) is enhanced to better align with non-integer periodicities typical of XR (for example, 16.66 ms for 60 fps, 11.11 ms for 90 fps, 8.33 ms for 120 fps). This allows more efficient sleep periods without missing expected data.
• PDCCH monitoring adaptations, including search space set group switching and PDCCH skipping, can be coordinated with PDU Set information and traffic jitter to reduce unnecessary monitoring.
• Alignment of DRX with XR traffic periodicity is shown in studies to be critical for both power saving and capacity.

Combined power-saving techniques contribute to estimated gains on the order of 10–30 % in relevant evaluations.

4. Support for Adaptive Low-Latency Traffic

Release 18 strengthens handling of delay-sensitive adaptive traffic, including L4S support. L4S traffic can be isolated into dedicated QoS flows, enabling better congestion signalling and rate adaptation while protecting classic traffic.

System Architecture and Media Support

Beyond pure RAN features, Release 18 includes broader 5G System enhancements for XR and media services (XRM work):

• Improved QoS frameworks that incorporate PDU Set handling.
• Media capabilities and profiles suited to AR/VR, including aspects of split rendering, immersive audio, and real-time transport configurations.
• Support for conversational AR services and related IMS or WebRTC scenarios in some work items.

These ensure end-to-end consistency from application to radio.

Performance Benefits and Use Cases

Studies and system-level simulations associated with the Release 18 features indicate typical gains of 10–30 % in XR capacity and device power saving under representative traffic models. The improvements are most pronounced when multiple techniques (awareness + CG enhancements + DRX alignment) are used together.

Target applications include:

• Cloud XR and split-rendering scenarios
• AR navigation, remote assistance, and industrial visualization
• Immersive multi-party communication and gaming
• Tethered or standalone XR glasses and headsets

The features also complement other Release 18 capabilities such as improved mobility (lower interruption for moving users) and energy-efficient network operation.

Summary of Key XR Enhancements

Category	Main Features	Primary Benefit
Application Awareness	PDU Set / Data Burst handling, UE assistance info, PSI-based discarding	Better prioritization and resource efficiency under load
Capacity	Multi-occasion Configured Grant, unused occasion UCI indication	Higher number of simultaneous XR users
Power Saving	DRX alignment to non-integer XR frame rates, PDCCH monitoring adaptation	Longer battery life for XR devices
Low-Latency Adaptive Traffic	L4S support, dedicated QoS flows	Improved congestion control and latency

Looking Ahead

Release 18 establishes the foundational XR-aware mechanisms in 5G-Advanced. Subsequent releases continue refining these capabilities (further power saving, mobility interaction, AI-assisted optimization, and media handling) to support denser XR deployments and more demanding multi-modal services.

In summary, the XR enhancements of Release 18 make 5G significantly more capable of delivering high-quality, power-efficient immersive experiences by introducing application-level awareness, more efficient uplink resource allocation, and traffic-aligned power-saving mechanisms. These features form a key part of the platform that enables 5G to proliferate into consumer, enterprise, and industrial XR use cases.

---

12) Sidelink Evolution

NR sidelink was first introduced in Release 16 primarily for advanced V2X services as a complement to LTE-V2X. Release 17 added power-saving features and inter-UE coordination for resource allocation. Release 18 continues this trajectory with a dedicated work item on NR Sidelink Evolution, focusing on three major directions that address spectrum limitations, data-rate needs, and mixed-generation deployments.

1. Sidelink Operation in Unlicensed Spectrum (SL-U)

A major expansion is the support for NR sidelink in FR1 unlicensed bands, specifically Band n46 (5 GHz) and Bands n96/n102 (6 GHz). This opens access to larger bandwidths that can support higher data rates for commercial D2D applications beyond traditional ITS/V2X.

• Channel access mechanisms reuse the principles established for NR-U (unlicensed) in Release 16, adapted for both resource allocation Mode 1 (network-scheduled) and Mode 2 (UE-autonomous).
• UEs performing sidelink transmissions in these bands follow listen-before-talk and related regulatory requirements.
• The feature targets commercial use cases that benefit from abundant unlicensed spectrum while maintaining compatibility with existing NR sidelink frameworks.

This significantly broadens the applicability of sidelink to scenarios such as high-throughput proximity services, industrial D2D, and other non-V2X applications where licensed ITS spectrum may be limited or unavailable.

2. Sidelink Carrier Aggregation

Release 18 introduces carrier aggregation (including packet duplication) for NR sidelink specifically in the ITS band (Band n47, 5.9 GHz).

• Intra-band aggregation allows UEs to combine multiple carriers for higher peak data rates and improved reliability.
• Packet duplication across carriers further enhances robustness for critical traffic.
• Primary target applications include high-degree automated driving scenarios that require sharing of high-volume sensor data (for example, video or high-resolution sensor feeds) between vehicles.

This addresses the bandwidth constraints of single-carrier operation in the dedicated ITS spectrum and improves support for advanced V2X services that demand both capacity and reliability.

3. Co-Channel Coexistence of LTE Sidelink and NR Sidelink

In many regions the 5.9 GHz ITS band is shared by both LTE-V2X and NR-V2X deployments. Release 18 specifies mechanisms for co-channel coexistence so that the two technologies can operate on the same frequency channel with minimized mutual interference.

• Support includes both semi-static and dynamic resource pool sharing.
• Dynamic coexistence allows overlapping resource pools with coordination to reduce cross-RAT impacts.
• The goal is to enable gradual migration from LTE sidelink to NR sidelink and concurrent operation during transition periods without severe performance degradation for either technology.

This is particularly important for regions with limited dedicated ITS spectrum where both generations must coexist.

Related Sidelink Enhancements in Release 18

While the core Sidelink Evolution work item focuses on the three areas above, Release 18 also includes closely related improvements:

• Sidelink Relay enhancements — Improvements for device-to-device and device-to-network relays, including service continuity (for example, inter-gNB mobility of relays, path switching), multi-path relay (aggregation or switching between direct and relayed paths), and support for higher reliability and throughput.
• Sidelink positioning / ranging — Support for positioning or ranging between UEs (vehicles or smartphones) using sidelink measurements and reference signals. This is specified as part of the broader expanded NR positioning work and enables sidelink-only or hybrid network+sidelink positioning scenarios.
• Continued support for unicast, groupcast, and broadcast transmissions, with refinements to resource allocation and power control where needed.

Summary of Key Sidelink Evolution Features

Feature	Main Scope	Primary Benefit	Target Use Cases
Unlicensed spectrum (SL-U)	FR1 bands n46, n96/n102	Higher data rates via larger bandwidths	Commercial D2D, industrial, non-V2X
Carrier Aggregation (n47)	Intra-band CA + packet duplication	Higher rate and reliability in ITS band	Advanced V2X sensor/video sharing
LTE–NR co-channel coexistence	Semi-static & dynamic resource pool sharing	Interference management in shared spectrum	Mixed LTE/NR V2X deployments
Related: Relay & Positioning	Service continuity, multi-path, SL ranging	Coverage, reliability, proximity positioning	Public safety, IoT, V2X, commercial

System Impact and Broader Context

These enhancements expand NR sidelink from a primarily V2X-centric technology into a more versatile proximity communication platform. Unlicensed operation and carrier aggregation increase capacity and data rates; coexistence mechanisms protect investments in existing LTE-V2X deployments; and related relay and positioning features improve coverage, reliability, and location services.

Together they support a wider range of applications—advanced autonomous driving, public safety, industrial automation, commercial D2D, and emerging proximity services—while remaining compatible with the overall 5G system architecture. Subsequent releases continue refining sidelink capabilities (including further relay, multi-beam, and integration with other 5G-Advanced features).

In summary, Sidelink Evolution in Release 18 delivers practical spectrum expansion, higher performance in dedicated ITS bands, and smoother coexistence between generations. It strengthens the foundation for device-to-device connectivity across both vehicular and commercial domains as part of 5G-Advanced.

---

13) Positioning and Location Services

NR positioning was first introduced in Release 16 with methods such as downlink Time Difference of Arrival (DL-TDOA), uplink TDOA, multi-Round Trip Time (multi-RTT), Angle of Arrival/Departure, and enhanced Cell ID, supported by dedicated Positioning Reference Signals (PRS) and Sounding Reference Signals (SRS). Release 17 refined latency, integrity, and other aspects. Release 18 significantly expands both the scope and the performance of these capabilities.

Key Objectives of Release 18 NR Positioning

The work item focused on five main areas:

• Positioning support for Reduced Capability (RedCap) devices
• Bandwidth aggregation for positioning measurements
• Enhancements enabling low-power high-accuracy positioning (LPHAP)
• Carrier phase measurements for positioning
• Sidelink-based positioning and ranging

These address limitations in bandwidth, device complexity, power consumption, and coverage scenarios (in-coverage, partial coverage, and out-of-coverage).

Major Technical Enhancements

1. Positioning for RedCap Devices

RedCap UEs have restricted RF bandwidth (typically up to 20 MHz in FR1). To improve their positioning accuracy despite this constraint, Release 18 introduces frequency hopping for positioning measurements. A RedCap UE can measure successive portions of a wider-bandwidth PRS (or corresponding uplink signals) across different frequency hops and combine the results, achieving an effective measurement bandwidth larger than its instantaneous RF capability. This enables usable high-accuracy positioning for lower-complexity IoT and industrial devices.

2. Bandwidth Aggregation for Positioning Measurements

For regular NR UEs, accuracy of timing-based measurements (RSTD, Rx-Tx time difference, etc.) improves with larger measurement bandwidth. Release 18 allows aggregation of PRS (and corresponding SRS) resources from up to three Positioning Frequency Layers (PFLs) that are intra-band contiguous. The UE performs measurements across the aggregated bandwidth, yielding higher resolution and better accuracy without requiring a single continuous wideband allocation.

3. Carrier Phase Positioning (CPP)

For the first time in a cellular standard, Release 18 introduces carrier phase measurements for NR positioning. New measurements—Reference Signal Carrier Phase (RSCP) and Reference Signal Carrier Phase Difference (RSCPD)—can be reported together with existing time-difference measurements. Carrier phase information provides much finer resolution than pure time-of-arrival methods and is a key enabler for centimetre-level accuracy in favourable conditions, drawing inspiration from techniques long used in high-precision GNSS.

4. Low-Power High-Accuracy Positioning (LPHAP)

Targeted at industrial IoT and similar power-constrained use cases, LPHAP introduces mechanisms that reduce the energy cost of high-accuracy positioning. This includes optimized configurations, longer discontinuous reception cycles in certain states, alignment of measurement opportunities with power-saving patterns, and support for positioning while the UE remains in low-power modes (for example, extended support in RRC_INACTIVE). The goal is to maintain high accuracy while significantly lowering average power consumption.

5. Sidelink-Based Positioning and Ranging

A major expansion of scope is the introduction of sidelink positioning. A new Sidelink Positioning Reference Signal (SL-PRS) is defined, supporting unicast, groupcast, and broadcast transmission. Positioning and ranging methods operate in in-coverage, partial-coverage, and out-of-coverage scenarios. This enables UE-to-UE relative positioning or ranging without continuous reliance on the network infrastructure—valuable for V2X, public safety, industrial, and commercial proximity applications. Hybrid network + sidelink solutions are also supported.

Positioning Integrity and Other Improvements

Release 18 continues work on positioning integrity for RAT-dependent methods, providing better information about the reliability and potential error bounds of location estimates. This is important for safety-critical and regulatory applications.

5GC Location Services – Phase 3

On the core network side, the 5GC Location Services Phase 3 work enhances the overall location architecture and procedures. Notable improvements include more efficient transfer of positioning signalling (including support for a user-plane path between the UE and the Location Management Function to reduce latency), refinements to mobile-terminated and mobile-originated location request procedures, better support for deferred location events, and integration with new radio features such as sidelink positioning. These changes improve end-to-end latency, scalability, and flexibility of location services.

Summary of Key Features

Feature	Main Benefit	Primary Target Scenarios
RedCap positioning (frequency hopping)	High-accuracy positioning for limited-bandwidth devices	IoT, industrial sensors, cost-sensitive UEs
PRS/SRS bandwidth aggregation	Higher measurement resolution and accuracy	Commercial and industrial high-accuracy needs
Carrier phase measurements	Potential centimetre-level accuracy	Precision applications, outdoor/indoor
Low-Power High-Accuracy Positioning	Accurate location with reduced energy cost	Industrial IoT, battery-constrained devices
Sidelink positioning / ranging	UE-to-UE location without full network dependence	V2X, public safety, out-of-coverage, commercial D2D
5GC LCS Phase 3	Lower latency signalling, broader procedure support	End-to-end location service efficiency

System Impact and Use Cases

Collectively these enhancements raise the achievable accuracy, extend positioning to a wider range of devices (including RedCap), improve power efficiency, and enable new topology options (sidelink-only or hybrid). They support demanding verticals such as:

• Industrial IoT and factory automation (sub-metre or better accuracy with low power)
• Advanced V2X and autonomous systems
• Public safety and emergency services
• Commercial indoor/outdoor navigation and asset tracking
• Regulatory location services

By reducing dependence on GNSS in many scenarios and providing integrity information, Release 18 positioning strengthens the overall location capability of the 5G System. Subsequent releases continue refining accuracy, integrity, power efficiency, and integration with other 5G-Advanced features such as AI/ML-assisted positioning.

In summary, the Positioning and Location Services work in Release 18 delivers a substantial step forward in accuracy, device inclusivity, power efficiency, and topological flexibility, establishing a stronger foundation for high-performance location services across consumer, industrial, and vehicular applications in 5G-Advanced.

---

14) Non-Terrestrial Networks (NTN) and Aerial Platforms

Release 17 introduced the baseline for NR-NTN (transparent payload) and IoT-NTN (NB-IoT/eMTC over satellite), supporting GEO, MEO, and LEO orbits with GNSS-capable UEs in L/S bands. Release 18 focuses on practical enhancements for commercial viability, higher performance, broader spectrum, better mobility, and device inclusivity (including realistic handheld terminals).

NR NTN Enhancements

The primary work item “NR NTN enhancements” and related spectrum items deliver the following improvements.

Uplink Coverage Enhancements

Commercial smartphones typically have limited antenna gain (for example, –5.5 dBi with polarisation loss). To improve uplink performance under the challenging path losses and long delays of satellite links, Release 18 introduces:

• PUCCH repetition for Msg4 HARQ-ACK — Configurable repetitions (1, 2, 4, or 8) signalled via system information or DCI. This strengthens the reliability of the critical random-access response acknowledgement.
• PUSCH DMRS bundling enhancements — Support for DMRS bundling that maintains phase continuity in the presence of timing drift caused by satellite motion and long propagation delays. This improves channel estimation quality for uplink data.

These measures specifically target better coverage for handheld devices without requiring specialised high-gain antennas.

Spectrum and Bandwidth Support

• New frequency bands, including additional L/S-band combinations (for example, Band n254) and Ka-band options (approximately 17.7–20.2 GHz downlink / 27.5–30.0 GHz uplink, with bands such as n510–n512) for both GSO and NGSO deployments.
• Support for 30 MHz channel bandwidth in selected FR1 NTN bands (in addition to the previously supported 5/10/15/20 MHz options).
• Requirements for VSAT-class terminals (fixed and on moving platforms) and electronically or mechanically steered beams in higher bands.

These expansions enable higher data rates and more flexible deployments.

Network-Verified UE Location

To satisfy regulatory requirements (lawful intercept, emergency services, public warning, etc.), Release 18 supports network verification of the GNSS location reported by the UE. Assuming a single satellite in view, the network can use timing measurements (for example, enhancements related to multi-RTT) to cross-check the reported coordinates, with a target granularity on the order of 10 km. This is an optional UE feature triggered by the core network.

Mobility and Service Continuity

NTN environments feature long delays, large cells, moving cells (especially LEO), and frequent satellite switches. Release 18 strengthens mobility in several ways:

• RACH-less handover support to reduce latency and congestion during handovers or feeder/satellite switches.
• Conditional handover enhancements with time-based and location-based trigger conditions, applicable also to Earth-moving cells.
• Satellite switch with re-synchronisation procedures.
• Improved system information: TN cells can broadcast NTN neighbour ephemeris (via SIB19); NTN cells can broadcast TN coverage area information (for example, via new SIBs) so UEs can optimise measurements and reduce power consumption.
• Support for mobility between NTN and terrestrial networks (TN) in both directions, and between NTN payloads in different orbits, while noting that simultaneous dual connectivity to NTN and TN is not required in this release.
• Handling of discontinuous coverage scenarios.

These features improve service continuity for both fixed and mobile users under satellite dynamics.

IoT NTN Enhancements

Building on Release 17 IoT-NTN, Release 18 provides further optimisations for NB-IoT and eMTC over satellite. These include refinements related to HARQ behaviour (for example, options to mitigate stalling under long delays), improved GNSS handling when position information becomes temporarily invalid, additional band support (extended L-band and L+S combinations), and capacity or signalling efficiency improvements. The goal is more robust massive machine-type connectivity via satellite for sensors, tracking, and other low-data-rate applications in remote areas.

Aerial Platforms and UAV Support

In parallel with pure satellite NTN, Release 18 advances support for Uncrewed Aerial Vehicles (UAVs), Uncrewed Aircraft Systems (UAS), and Urban Air Mobility (UAM):

• Architecture enhancements for UAV and UAM (Phase 2), including improved identification, tracking, route authorisation, and group communication aspects.
• NR radio support for devices onboard aerial vehicles: refined measurement reporting (to manage interference from aerial UEs), subscription-based UAV identification (including multicast options), conditional handover triggers suited to aerial trajectories, and beam management considerations (including base-station uptilt beamforming for serving low-altitude UAVs).
• Related work addresses interference management and mobility for denser low-altitude aerial deployments.

High-Altitude Platforms (HAPs) and other airborne vehicles are generally considered within the broader NTN framework, benefiting from the same coverage, mobility, and spectrum enhancements.

System-Level and Related Aspects

Release 18 also includes system architecture work (for example, 5GSAT Phase 2) covering satellite backhaul, discontinuous coverage handling, management aspects, and guidelines for extra-territorial operation. These ensure end-to-end consistency between the radio, core, and management planes.

Summary of Key Features

Area	Main Enhancements	Primary Benefit
Uplink Coverage	PUCCH Msg4 HARQ-ACK repetition, PUSCH DMRS bundling	Better performance for commercial handhelds
Spectrum	New L/S and Ka bands, 30 MHz CBW	Higher rates and more deployment flexibility
Location Verification	Network check of UE-reported GNSS coordinates	Regulatory compliance
Mobility & Continuity	RACH-less HO, conditional HO (time/location), satellite switch with re-sync, improved SIBs	Robust service under satellite dynamics
IoT NTN	Further optimisations for NB-IoT/eMTC	Reliable massive IoT via satellite
UAV / Aerial	Architecture, identification, measurements, CHO, beam management	Support for drones and aerial platforms

Impact and Outlook

Release 18 moves NTN from a foundational capability to a more commercially practical solution by addressing real-world uplink limitations of smartphones, expanding spectrum options, strengthening mobility across hybrid TN–NTN deployments, and adding regulatory location tools. Combined with UAV enhancements, it enables broader use cases ranging from global IoT connectivity and maritime/aviation services to direct-to-device satellite broadband and low-altitude aerial operations.

Subsequent releases continue this trajectory with regenerative payloads, further capacity and latency improvements, and deeper TN–NTN integration. Overall, the NTN and aerial platform work in Release 18 significantly advances the goal of ubiquitous 5G coverage beyond traditional terrestrial footprints.

---

15) Network Slicing Phase 3

Network slicing allows operators to create multiple logical networks on a shared physical infrastructure, each tailored to specific service requirements (for example, eMBB, URLLC, mMTC, or vertical-specific needs). Releases 15–17 established the basic framework (S-NSSAI, NSSF, slice selection, isolation principles, and initial management). Phase 3 in Release 18 focuses on greater deployment flexibility, reduced operational constraints, improved continuity, and better support for localized or temporary slices.

Key Architectural and System Enhancements (Primarily SA2)

Partial Allowed NSSAI and Localized Slice Support

A major limitation in earlier releases was the expectation that a network slice would be uniformly available across an entire Tracking Area (or Registration Area). This constrained highly localized deployments such as enterprise campuses or industrial sites.

Release 18 introduces the concepts of partially allowed NSSAI and S-NSSAIs rejected partially within a Registration Area. A slice may now be supported only in a subset of the cells or Tracking Areas belonging to a UE’s Registration Area. The AMF signals this information to the NG-RAN, enabling more precise mobility and resource decisions. This change is particularly valuable for enterprise and private-network use cases where a slice needs to exist only in a limited geographic footprint.

Network Slice Replacement and Service Continuity

Operators can replace one S-NSSAI (or slice instance) with an alternative S-NSSAI. This supports:

• Congestion or load management
• Planned maintenance
• Dynamic redirection of traffic to a more suitable slice

The RAN is informed of the replacement so that service continuity can be maintained with minimal disruption. Combined with partial NSSAI handling, this provides more resilient and flexible slice lifecycle management.

Network Slice Admission Control (NSAC) Enhancements

NSAC is extended to control the maximum number of UEs that have at least one PDU session (or PDN connection) associated with a given slice. This gives operators finer quota management, including in roaming scenarios where the home operator can influence or enforce preferred limits in coordination with the visited network. Interworking with EPC is also improved for mixed 4G/5G environments.

Support for Dynamic and Temporary Slices

Signalling procedures are optimized for slices that are activated or deactivated frequently (for example, for events, emergency response, or short-term enterprise needs). The goal is to reduce registration and session-management overhead associated with repeated activation/deactivation cycles.

RAN Impacts

The NG-RAN receives and acts on the new partial-allowed and replacement information:

• Cells outside a slice’s intended service area can be configured with zero radio resources for that slice.
• Neighbouring NG-RAN nodes exchange information about zero-resource configurations via the Xn interface.
• Mobility decisions in RRC_CONNECTED mode can take partial NSSAI information into account.
• Overall resource management between slices becomes more dynamic while still respecting SLAs.

These mechanisms allow the radio network to enforce slice-specific service areas more accurately and avoid unnecessary resource reservation.

Management, Exposure, and Charging Aspects

• Network Slice Capability Exposure — Application-layer enablement (via SEAL frameworks) allows vertical applications or application functions to interact with slice capabilities, including remapping of application traffic to different slices and related lifecycle or optimization actions.
• Management enhancements — Improved modelling of isolation profiles, transport entry points, scheduled slice availability, and asynchronous lifecycle operations.
• Charging — Extensions cover Network Slice-Specific Authentication and Authorization (NSSAA), slice replacement scenarios, tenant-based charging, and wholesale/roaming slice charging. These enable more sophisticated monetization models for third-party or enterprise slices.

Security aspects were also studied (enhanced security for network slicing Phase 3) to address the additional flexibility introduced by partial and dynamic slice handling.

Benefits and Target Scenarios

Enhancement	Main Benefit	Typical Use Cases
Partial Allowed NSSAI	Localized slice deployment without TA-wide support	Enterprise campuses, industrial sites, private networks
Slice replacement / continuity	Resilience under congestion or maintenance	Dynamic load balancing, planned outages
Enhanced NSAC	Finer quota and admission control (including roaming)	Multi-tenant, tiered services, wholesale
Dynamic/temporary slice support	Lower signalling overhead for short-lived slices	Events, emergency, temporary services
RAN zero-resource & Xn exchange	Precise resource isolation and mobility	Efficient multi-slice RAN operation
Capability exposure & charging	Application interaction and flexible monetization	Vertical applications, B2B slicing

Overall Impact

Network Slicing Phase 3 in Release 18 removes several rigid constraints of earlier releases and makes slicing more practical for real-world enterprise, localized, and dynamic deployments. By supporting partial coverage within Registration Areas, enabling seamless replacement, refining admission control, and improving RAN awareness, operators gain greater operational flexibility while still providing differentiated quality of service and isolation.

These features complement other Release 18 capabilities (for example, non-public networks, edge computing, and AI/ML-assisted management) and strengthen the foundation for advanced multi-tenant and vertical-specific 5G services in 5G-Advanced networks. Subsequent releases continue refining automation, intent-driven management, and deeper integration with analytics and AI.

---

16) Multicast/Broadcast Services (MBS) Phase 2

Release 17 defined the core 5G MBS architecture, including Multicast and Broadcast MBS sessions, 5GC Shared and Individual delivery methods, PTM (Point-to-Multipoint) and PTP (Point-to-Point) transmission options in the RAN, and support for broadcast reception across RRC states. Multicast reception, however, was limited primarily to RRC_CONNECTED UEs. Phase 2 in Release 18 removes several practical limitations and improves deployment efficiency.

Key Architectural and System Enhancements

The architectural work (including the study on enhancements for 5G multicast-broadcast services Phase 2) evaluated and led to normative support for:

• More efficient resource utilisation for broadcast content in different network-sharing scenarios.
• Multicast reception by UEs in RRC_INACTIVE state.
• On-demand multicast MBS sessions that can be triggered by an Application Function (AF).
• Solutions addressing performance and scalability issues when a large number of public-safety or mission-critical devices receive the same service.

These changes improve both the core network handling of MBS sessions and the interaction with the NG-RAN.

RAN Enhancements (NR MBS Enhancements)

The corresponding radio work item focused on three primary practical improvements:

1. Multicast Reception in RRC_INACTIVE State

In Release 17, multicast service data was generally delivered only to UEs in RRC_CONNECTED. Release 18 extends multicast reception to UEs in RRC_INACTIVE. This is critical for scenarios with many users (for example, public safety groups or large-scale event audiences) because it avoids the need to keep every interested UE in the Connected state, thereby reducing signalling load, power consumption, and the number of active connections the network must maintain.

HARQ feedback is typically not supported for multicast data reception in the Inactive state (consistent with the power-saving and complexity goals of that state). Configuration and scheduling information are provided so that Inactive UEs can receive the relevant Multicast Traffic Channel (MTCH) transmissions.

2. Shared Processing for Broadcast and Unicast

UEs may need to receive MBS broadcast services simultaneously with unicast traffic. When hardware or processing resources are shared, this can affect the UE’s unicast capabilities. Release 18 introduces signalling enhancements that allow the UE to indicate changes in its resource/capability status related to concurrent MBS and unicast reception. The network can then reconfigure the unicast leg appropriately (for example, adjusting MIMO layers, bandwidth, or other parameters) so that overall service quality is maintained.

3. Resource Efficiency in RAN Sharing Scenarios

In multi-operator RAN-sharing deployments it is inefficient for each operator to transmit the same MBS content using separate radio resources. Release 18 specifies mechanisms that improve coordination and resource utilisation so that a single transmission can serve UEs of multiple operators where appropriate, reducing redundant air-interface usage and improving overall capacity.

Additional supporting features include refinements to MBS frequency-layer prioritisation, neighbour-cell information exchange (via Xn) for frequency prioritisation, and continued support for location-dependent content via Area Session IDs.

Mission-Critical and Public-Safety Support

A related work stream addresses Mission Critical services over 5G MBS (MC over 5MBS). This covers functional models, identification, session management, and mobility aspects needed to deliver group communications and other mission-critical traffic efficiently via the multicast/broadcast framework. The Inactive-state multicast support and scalability enhancements are particularly relevant here.

Delivery Methods and Session Handling (Recap with Phase 2 Context)

• Broadcast MBS sessions — Received by UEs in RRC_IDLE, RRC_INACTIVE, and RRC_CONNECTED. Configuration is provided via MCCH (Multicast Control Channel) and system information.
• Multicast MBS sessions — Now receivable in RRC_INACTIVE (in addition to Connected). Support both Shared (PTM) and Individual (PTP) delivery methods depending on NG-RAN capabilities and mobility needs.
• Session activation/deactivation, update, and area management procedures are refined to support the new states and on-demand triggers.

Benefits and Target Use Cases

Enhancement	Main Benefit	Primary Use Cases
Multicast in RRC_INACTIVE	Lower signalling and power; more scalable groups	Public safety, large audiences, IoT updates
Shared unicast + broadcast processing	Better concurrent service quality	Mixed media + unicast applications
RAN-sharing resource efficiency	Reduced redundant transmissions	Multi-operator deployments
On-demand / AF-triggered sessions	Flexible, demand-driven activation	Event-based or application-driven services
Scalability for large device numbers	Support for dense public-safety groups	Mission-critical communications

Overall Impact

MBS Phase 2 in Release 18 makes 5G multicast and broadcast significantly more practical and efficient. By enabling multicast reception in the power-efficient Inactive state, improving concurrent unicast/broadcast operation, and optimising shared-RAN deployments, the features reduce network load while expanding the set of viable applications. Public-safety, media distribution, software/firmware delivery, automotive, and large-scale event services all benefit from the improved scalability and resource efficiency.

These enhancements complement other Release 18 capabilities (for example, XR media handling, network slicing, and energy-saving techniques) and strengthen the overall 5G System’s ability to deliver efficient one-to-many services. Further refinements continue in subsequent releases, but Phase 2 establishes the key operational improvements needed for broader commercial and mission-critical adoption of 5G MBS.

---

17) Edge Computing Phase 2

Edge Computing places application servers and related functions close to the UE’s point of attachment (typically beyond a local PSA UPF in the Data Network). This reduces end-to-end latency and transport-network load. Release 17 defined the basic 5G System enablers (local traffic steering, AF-influenced routing, session and service continuity modes, and the application-layer architecture with Edge Enabler Client/Server and Configuration Server). Phase 2 in Release 18 addresses practical multi-operator, roaming, group, and continuity scenarios that were left open or only partially covered earlier.

System Enhancements for Edge Computing (Primarily TS 23.548 / EDGE_Ph2)

Key system-level improvements include:

• Roaming support for Edge Hosting Environments — A UE can access an Edge Hosting Environment (EHE) located in the visited network. This enables local breakout and low-latency services while the user is roaming.
• Finer-grained UE policies — Policies can be applied to more specific sets of UEs rather than broad groups, allowing more precise traffic steering and edge selection.
• (Re)location of UPF and EAS for collections of UEs — The network can relocate the User Plane Function and/or Edge Application Server for a group of UEs together, improving efficiency for correlated users (for example, participants in the same application session or event).
• Multi-organization Edge Hosting Environments — Solutions are improved to align with GSMA OPG concepts for EHEs operated by different organizations, supporting federation and partner models.
• AF-maintained IP-to-DNAI mapping — An Application Function can obtain and maintain a mapping table between IP address ranges and Data Network Access Identifiers (DNAIs). This helps the AF influence routing more accurately.
• Security enhancements — Authorization and authentication procedures used in edge computing scenarios are strengthened.

These features make edge deployments more flexible across operator boundaries and for groups of users.

Architecture for Enabling Edge Applications Phase 2

The application-layer architecture (Edge Enabler Client – EEC, Edge Enabler Server – EES, Edge Configuration Server – ECS, and Edge Application Server – EAS) receives several practical enhancements:

1. Edge node sharing — A partner edge computing service provider can share its edge resources with a “home” service provider (where the application server resides) under a federation agreement. This supports multi-party edge ecosystems.
2. Discovery of a common EAS — Participants belonging to a specific group can discover and use the same common Edge Application Server, facilitating shared sessions or collaborative applications.
3. Enhanced dynamic EAS instantiation — The Edge Enabler Server can trigger the Edge Computing Service Provider management system to instantiate an EAS. This is useful when discovery fails or during Application Context Relocation (ACR).
4. Notification Management Service — SEAL notification management services are reused (EEC acting as VAL client; EES/ECS acting as VAL server) for more standardized event and notification handling.
5. Application Context Relocation between EAS and Cloud — Architecture and procedures support service continuity when the application context moves between an edge server and a central cloud.
6. Bundled EASs — An Application Client can discover multiple EASs that are treated as a bundle and maintain service continuity across them.

These capabilities improve discovery, instantiation, sharing, and continuity for edge applications.

Management Enhancements

Related management work (edge computing management enhancements) addresses:

• Network Resource Model (NRM) updates for edge-related entities.
• Performance measurements and KPIs for edge servers and functions.
• Collection of 5GC network-function alarms to support Edge Application Server fault supervision.

These provide operators with better visibility and control over edge deployments.

Benefits and Target Scenarios

Enhancement Area	Main Capability	Primary Benefit / Use Cases
Roaming EHE access	Local edge services while roaming	Consistent low-latency experience for travellers
Group / collection relocation	Coordinated UPF/EAS moves for sets of UEs	Multi-user applications, events, collaborative XR
Edge node sharing & federation	Partner resource sharing	Multi-operator / multi-provider edge ecosystems
Common & bundled EAS discovery	Shared or coordinated application servers	Group services, multi-EAS applications
Dynamic instantiation & ACR to cloud	On-demand EAS creation and cloud continuity	Elastic capacity, hybrid edge-cloud continuity
Finer policies & IP–DNAI mapping	Precise steering and AF influence	Better traffic control and application awareness

Overall Impact

Edge Computing Phase 2 makes edge deployments more practical in multi-operator, roaming, and multi-party environments while improving service continuity and operational flexibility. By supporting visited-network edge access, group-aware relocation, federation/sharing models, dynamic instantiation, and stronger continuity mechanisms (including to the cloud), Release 18 reduces barriers to commercial edge services.

These enhancements complement other Release 18 features such as network slicing, XR media handling, and AI/ML-assisted management. They provide a more mature platform for low-latency applications in enterprise campuses, industrial sites, consumer immersive services, and hybrid edge-cloud architectures. Subsequent releases continue refining automation, exposure, and integration with broader 5G-Advanced capabilities.

---

18) Media and Real-Time Communication

Release 18 builds on earlier media frameworks (5G Media Streaming, MTSI/IMS multimedia telephony, and initial XR studies) to better support delay-sensitive, interactive, and immersive applications such as AR conversational services, multi-party immersive calls, split rendering, and real-time media sharing.

Real-Time Media Communication (RTC) Architecture

A dedicated architecture for Real-Time Media Communication over the 5G System is defined (TS 26.506). RTC is characterized as the delivery of delay-sensitive media from one peer to another with 5G network support.

Key aspects include:

• Mapping of RTC functions onto the 5G System architecture, leveraging 5G Media Streaming concepts where appropriate.
• Reference points and interfaces that support both operator-managed and third-party media communication service providers (including AR media providers).
• Collaboration models between the MNO and external application/media providers.
• Support for functionalities relevant to AR, such as split rendering and spatial computing, on top of the 5G System.
• Procedures for media session handling, provisioning, dynamic policy, reporting, and media/content transport (including peer-to-peer and client-to-server paths).

This architecture provides a standardized foundation for modern real-time media services that go beyond traditional conversational voice/video.

Immersive and AR Media Capabilities

Media Capabilities for Augmented Reality define the media formats, codecs, and related capabilities needed for AR/MR experiences (including speech, video, real-time text, still images, and associated metadata).

IMS-based AR Real-Time Communication (TS 26.264) specifies how IMS can be used for AR conversational services. It covers:

• Terminal and end-to-end reference architectures.
• Immersive AR media components (speech, video, real-time text, still images).
• AR metadata handling.
• Media configurations and transport (primarily RTP).
• Quality-of-Experience considerations and call-flow examples (session setup, split-rendering negotiation, scene description).

These specifications enable IMS-based AR calls with proper media negotiation and transport of immersive content and metadata.

WebRTC-Based Immersive Real-Time Communication

Release 18 advances support for WebRTC-based real-time media over 5G. Related specifications (including protocols and APIs in TS 26.113) define stage-3 procedures, APIs, and protocols for the RTC reference points, with a focus on WebRTC media transport.

This includes:

• Media session handling and provisioning procedures.
• Network and UE media session control.
• Media content and signalling transport (including peer-to-peer options where permitted).
• Integration points that allow WebRTC applications to benefit from 5G QoS, edge computing, and policy control.

Studies on further enhancements for immersive WebRTC communication continued to identify interworking, discovery, and protocol-level improvements.

Real-Time Media Transport Protocol Configurations

To support immersive media efficiently, Release 18 specifies RTP (over UDP) configurations optimized for XR and real-time immersive services (TS 26.522). Key elements include:

• RTP Header Extensions for carrying immersive media and associated metadata.
• RTCP Feedback Reporting mechanisms tailored to real-time immersive content.
• Configurations usable by IMS-based services, WebRTC-based services, and XR split-rendering scenarios (for example, between edge and device).

These configurations improve performance and Quality of Experience by enabling cross-layer optimizations with the 5G System (including interactions with the XR awareness and PDU-set features defined in the RAN).

Interaction with System and RAN Features

Media and real-time communication work is closely linked to other Release 18 enhancements:

• XR-aware RAN scheduling and PDU-set handling (already covered under XR enhancements) allow differentiated treatment of media frames or slices within a QoS flow.
• Edge Computing Phase 2 capabilities support low-latency split rendering and media processing close to the user.
• Network slicing and QoS frameworks provide the necessary isolation and prioritization for interactive media.
• Updates to IMS multimedia telephony media handling (TS 26.114 and related) incorporate the new immersive capabilities where relevant.

Summary of Key Areas

Area	Main Specifications / Focus	Primary Benefit
RTC Architecture	TS 26.506	Standardized real-time media framework over 5GS
IMS-based AR Communication	TS 26.264	Immersive AR conversational services via IMS
WebRTC Immersive RTC	TS 26.113 and related	WebRTC-based real-time immersive media
Media Capabilities for AR	TS 26.119	Defined codecs, formats, and capabilities for AR
RTP/RTCP Configurations	TS 26.522	Optimized transport for immersive media & metadata
System/RAN Integration	Cross-layer with XR, Edge, QoS, Slicing	Better latency, prioritization, and efficiency

Target Use Cases and Impact

These features enable:

• AR/MR conversational services and multi-party immersive communication.
• Split-rendering XR applications (device + edge/cloud).
• High-quality interactive media with low latency and improved congestion handling.
• Both operator-managed and third-party real-time media services.

By defining architectures, media capabilities, transport configurations, and integration points, Release 18 significantly strengthens 5G support for modern interactive and immersive media. The work provides a more complete toolkit for AR, real-time collaboration, and low-latency media applications while remaining aligned with existing IMS, WebRTC, and 5G Media Streaming frameworks. Subsequent releases continue refining these capabilities for broader commercial deployment.

---

19) Real-World Applications of 3GPP Release 18 (5G-Advanced)

Release 18 (the first 5G-Advanced release) strengthens the foundational performance of 5G while expanding its reach to more devices, environments, and verticals. The result is a more efficient, flexible, and inclusive platform that supports both evolutionary improvements to existing services and new classes of applications.

1. Industrial IoT, Smart Manufacturing, and Enterprise Campuses

Key enabling features: eRedCap, network slicing Phase 3 (partial/localized slices), edge computing Phase 2, positioning enhancements (including RedCap and low-power high-accuracy), energy savings, and AI/ML-assisted optimization.

• Low-cost, long-battery sensors, actuators, and trackers become practical with eRedCap (≈10 Mbps peak rates, reduced complexity, extended eDRX). These replace or complement LTE Cat-1/1bis devices while running on modern 5G SA infrastructure.
• Highly localized network slices support dedicated campus or factory networks without requiring full Tracking-Area-wide deployment.
• Edge computing improvements (roaming support, group relocation, dynamic instantiation, federation) enable low-latency control loops, predictive maintenance, and local AI inference.
• High-accuracy, power-efficient positioning supports asset tracking, automated guided vehicles (AGVs), and worker safety in indoor/outdoor industrial environments.
• Network energy savings and AI/ML load balancing reduce operating costs for always-on industrial connectivity.

Typical applications: Smart factories, process automation, warehouse logistics, utility metering/smart grids, and private enterprise networks.

2. Immersive Experiences, XR, and Real-Time Media

Key enabling features: XR enhancements (PDU Set awareness, configured-grant improvements, DRX alignment), media and real-time communication architecture, IMS-based and WebRTC-based AR communication, edge computing, and improved mobility (LTM).

• Application-aware scheduling and PDU-set handling improve capacity and reliability for high-rate, low-latency XR traffic (video frames, pose data, split rendering).
• Power-saving adaptations matched to XR frame rates (60/90/120 fps) extend battery life on glasses and headsets.
• Real-time media architectures and RTP configurations support immersive AR conversational services, multi-party collaboration, and hybrid edge-device rendering.
• Faster, lower-interruption mobility (L1/L2 Triggered Mobility) maintains experience for moving users.

Typical applications: AR remote assistance and training, immersive collaboration and virtual meetings, cloud gaming and entertainment, industrial visualization, tourism/navigation overlays, and consumer XR glasses (including tethered or RedCap-based form factors).

3. Automotive, Intelligent Transport, and Public Safety

Key enabling features: Sidelink evolution (unlicensed spectrum, carrier aggregation, LTE–NR coexistence), sidelink positioning/ranging, NTN enhancements, MBS Phase 2 (multicast in Inactive state), and advanced positioning.

• Sidelink in unlicensed bands and carrier aggregation in ITS spectrum increase data rates and reliability for sensor/video sharing between vehicles.
• Coexistence mechanisms allow smooth migration and concurrent LTE/NR V2X operation.
• Sidelink ranging and hybrid positioning support relative localization even in partial or out-of-coverage scenarios.
• Multicast reception in RRC_INACTIVE scales efficiently to large groups of first responders or vehicles.
• NTN coverage and mobility enhancements extend connectivity to remote roads, maritime, and aeronautical environments.

Typical applications: Advanced driver assistance and higher levels of automation (sensor sharing, platooning), vehicle-to-everything safety messages, public-safety group communications, disaster recovery, and fleet management in remote areas.

4. Massive and Critical IoT, Including Remote and Aerial Connectivity

Key enabling features: eRedCap, IoT NTN enhancements, NTN coverage/mobility improvements for handhelds, UAV/aerial platform support, and energy-efficiency techniques.

• eRedCap brings affordable 5G connectivity to high-volume, modest-throughput devices (meters, trackers, basic sensors).
• NTN uplink coverage enhancements (PUCCH repetition, DMRS bundling) and new bands improve performance for commercial smartphones and IoT devices via satellite.
• Network-verified location and discontinuous-coverage handling support regulatory and operational needs.
• UAV-specific measurement, identification, and mobility features enable safer and more reliable drone operations.

Typical applications: Global asset tracking, agricultural and environmental monitoring, maritime and rural broadband/IoT, direct-to-device satellite messaging, drone delivery and inspection, and urban air mobility support.

5. Consumer Broadband, Energy Efficiency, and Network Operations

Key enabling features: Advanced MIMO (higher-rank uplink, multi-TRP, mobility-robust CSI), evolved duplexing (SBFD for uplink capacity), L1/L2 mobility, network energy savings, and AI/ML for energy/load/mobility optimization.

• Higher uplink capacity and better cell-edge performance improve experience for fixed wireless access, content uploading, and interactive applications.
• Sub-band non-overlapping full duplex and multi-TRP techniques increase spectral efficiency in existing TDD spectrum.
• Network energy-saving techniques (antenna/power adaptation, cell DTX/DRX, SSB-less SCells) deliver measurable reductions in base-station consumption, especially at low-to-medium load.
• AI/ML-driven decisions further optimize energy, load balancing, and mobility.

Typical applications: Enhanced fixed wireless access (FWA), better smartphone experience in dense or high-mobility areas, lower-cost sustainable network operation, and improved quality for latency-sensitive consumer services.

Summary Mapping of Features to Application Domains

Domain	Standout Rel-18 Enablers	Primary Real-World Benefit
Industrial / Enterprise	eRedCap, localized slicing, edge, high-accuracy positioning	Cost-effective, low-latency private networks
Immersive / XR / Media	XR awareness, RTC architecture, edge, LTM	Reliable, power-efficient immersive experiences
Automotive / V2X / Safety	Sidelink evolution, SL positioning, MBS Inactive, NTN	Higher reliability and coverage for safety-critical services
Massive / Remote IoT	eRedCap, IoT NTN, aerial support	Affordable global connectivity
Network Efficiency & Broadband	MIMO evolution, SBFD, energy savings, AI/ML	Higher capacity, lower energy cost, better UX

Broader Impact

Release 18 does not invent entirely new verticals; instead it removes practical barriers—cost, power, coverage, latency, scalability, and operational rigidity—that previously limited 5G adoption in many of the above domains. Operators gain tools for more efficient spectrum use and greener networks; enterprises obtain more flexible and localized slicing plus edge capabilities; device makers can produce lower-cost 5G IoT modules; and end users experience more consistent performance for immersive, mobile, and remote services.

Commercial traction is already visible in early trials and deployments of LTM, RedCap/eRedCap modules, NTN direct-to-device services, and XR optimizations. As chipsets, devices, and network software mature through 2026–2028, these Release 18 capabilities are expected to move from advanced trials into mainstream production across industrial, consumer, automotive, and public-safety markets.

In short, the real-world value of 3GPP Release 18 lies in making 5G more capable, more efficient, and more widely applicable—bridging the gap between advanced technical features and practical, scalable services that benefit industries and society.

---

20) Consumer Broadband Improvements in 3GPP Release 18

Consumer broadband remains the highest-volume use case for 5G. Users expect consistent high throughput for video streaming, cloud applications, video calls, content uploading, gaming, and general web use—whether stationary or mobile. Release 18 strengthens the radio foundation that delivers these experiences without requiring entirely new spectrum or radical network redesigns.

Advanced MIMO Evolution

Massive MIMO continues to be a primary capacity and coverage lever. Release 18 delivers several targeted upgrades:

• Higher-rank uplink — Support for up to rank-8 / 8-layer PUSCH transmission, along with multi-panel simultaneous uplink transmission. These capabilities are especially valuable for higher-end devices and FWA customer-premises equipment (CPE), enabling substantially higher uplink data rates.
• Expanded multi-user MIMO capacity — Up to 24 orthogonal DMRS ports for downlink and uplink increase the number of users that can be paired simultaneously in MU-MIMO.
• Improved CSI for medium- and high-velocity scenarios — Time-domain prediction and compression within enhanced Type-II codebooks reduce channel aging effects, improving beamforming and throughput for users in vehicles or on public transport.
• Multi-TRP enhancements — Extension of the unified TCI framework to multiple simultaneous TCI states, dual timing advance support, and coherent joint transmission for up to four TRPs (under ideal backhaul assumptions). These improve both spectral efficiency and reliability, particularly at the cell edge or in non-line-of-sight conditions.

Collectively, these MIMO advances raise system capacity, improve uplink performance, and deliver more consistent experience across a wider range of channel conditions and device types.

Evolved Duplexing and Uplink Capacity

Traditional TDD operation allocates entire time slots exclusively to downlink or uplink, creating an inherent uplink bottleneck. Release 18’s study and related work on Sub-band non-overlapping Full Duplex (SBFD) allow simultaneous downlink and uplink transmission on different frequency sub-bands within the same TDD carrier.

Evaluations showed meaningful uplink coverage gains (approximately 5.4 dB in FR1 Urban Macro and higher in some FR2 scenarios) relative to conventional TDD patterns when combined with techniques such as PUSCH repetition. More frequent uplink transmission opportunities directly benefit latency-sensitive and uplink-heavy consumer applications such as video calling, live streaming, cloud backups, and interactive services.

Mobility Robustness – L1/L2 Triggered Mobility (LTM)

Handover interruption has long been a source of freezes, rebuffering, and degraded experience for mobile users. LTM enables the network to trigger a cell switch using a lightweight MAC CE based on Layer 1 measurements, with candidate cells pre-configured and early downlink/uplink synchronization performed in advance.

Typical interruption times fall into the 20–30 ms range (or lower in optimized cases), a substantial reduction compared with conventional Layer 3 handovers. Combined with support for carrier aggregation and certain dual-connectivity scenarios, LTM provides smoother mobility for users moving through dense urban environments, on highways, or on high-speed transport. This is particularly noticeable for video streaming, real-time collaboration, and gaming.

Supporting Coverage and Deployment Features

Additional Release 18 elements further strengthen consumer broadband:

• Coverage enhancements (including uplink-focused techniques and random-access improvements) improve cell-edge rates and initial access reliability.
• Network-controlled repeaters and mobile IAB expand the practical reach of high-capacity coverage in a cost-effective manner, benefiting both outdoor mobile users and FWA deployments in challenging locations.
• Multi-TRP and advanced beam management improve robustness in environments with blockage or rapid channel variation.

Impact on Fixed Wireless Access (FWA)

FWA CPE devices often have more flexible antenna configurations than smartphones. They can more readily exploit higher-rank uplink, multi-panel transmission, and multi-TRP operation. The combination of increased uplink capacity, better cell-edge performance, and more robust mobility (for nomadic or outdoor CPE) enables more consistent multi-hundred-Mbps or gigabit-class service, expanding the addressable market for fixed wireless broadband as a fiber alternative or complement.

Summary of Consumer Broadband Gains

Aspect	Key Release 18 Features	Primary User Benefit
System capacity	Higher MU-MIMO ports, advanced multi-TRP, better CSI	More users served at high rates simultaneously
Uplink performance	Rank-8 UL, multi-panel, SBFD	Faster uploads, better video calls & interactive apps
Cell-edge experience	Multi-TRP, coverage enhancements, SBFD	More consistent rates farther from the base station
Mobility	L1/L2 Triggered Mobility (LTM)	Fewer freezes and interruptions while moving
FWA suitability	Higher-rank UL, multi-panel, multi-TRP	Higher and more reliable home/small-business speeds

Overall Effect

Release 18 does not redefine the consumer broadband service itself; it makes the underlying radio system more efficient and robust. Operators can extract more capacity and better coverage from existing spectrum and sites. End users experience higher average and cell-edge throughputs, improved uplink performance, and smoother service during mobility. FWA deployments gain particularly from the uplink and multi-antenna enhancements.

These improvements complement energy-efficiency and operational features (antenna adaptation, AI/ML-assisted load balancing, etc.) that help operators deliver the enhanced broadband experience at lower cost and with reduced energy consumption. As network software and compatible devices roll out, the consumer broadband gains of Release 18 become a practical, incremental but meaningful step forward in everyday 5G performance.

---

21) Network Energy Efficiency

Energy efficiency moved from a largely implementation-specific concern to a core system capability in Release 18. Rising energy costs, sustainability targets, and the recognition that the radio access network accounts for the majority of mobile-network electricity consumption drove both radio-layer innovations and management-plane enhancements.

Why Network Energy Efficiency Matters

Base stations operate with significant fixed power draw even at low load because of always-on signalling (SSB, PDCCH monitoring opportunities, etc.) and the need to maintain coverage. Daily and geographic traffic variations mean that large portions of the network spend considerable time in low- or medium-load states where substantial savings are possible without harming user experience. Release 18 provides operators with standardized tools to exploit those opportunities while preserving interoperability and service quality.

Core RAN Energy-Saving Techniques

Release 18 specifies multi-domain adaptations that create additional sleep opportunities and reduce transmission power or active hardware.

Spatial and Power Domain Adaptations

The gNB can dynamically reduce the number of active antenna ports or elements, or lower transmission power, according to traffic demand and channel conditions. A new CSI reporting framework allows the UE to report multiple CSI sub-reports corresponding to different spatial or power adaptation hypotheses in a single reporting instance. This gives the network the information needed to select an optimal reduced configuration efficiently.

These techniques typically deliver the largest gains—commonly in the 15–30 % range under low-to-medium load, with higher figures reported when antenna and power adaptation are used jointly.

Time Domain: Cell DTX/DRX

Cell discontinuous transmission and discontinuous reception allow the gNB to enter sleep states during periods of inactivity for UEs in RRC_CONNECTED mode. Configurable active/non-active patterns can be activated independently or jointly and should be aligned with UE connected-mode DRX cycles to avoid mismatched wake periods. This enables deeper sleep (micro-sleep or deeper states) and further reduces energy consumption.

Frequency Domain and Supporting Features

• Extension of SSB-less secondary-cell operation to additional carrier-aggregation scenarios reduces always-on signalling overhead.
• Conditional handover enhancements help maintain mobility robustness when cells enter energy-saving modes.
• Mechanisms for managing legacy UE behaviour and inter-node coordination support consistent operation across mixed networks.

The techniques are designed so that legacy UEs can generally continue to access the network, and the evaluation methodology explicitly balances energy savings against user-perceived throughput, latency, capacity, and other KPIs.

AI/ML-Assisted Energy Saving

Network energy saving is one of the three priority use cases for the first normative AI/ML support in the NG-RAN. Models can generate energy-cost predictions and load forecasts that are exchanged between neighbouring nodes via standardized Xn data-collection procedures. These predictions enable more proactive and accurate decisions on cell or carrier activation/deactivation, reducing the risk of quality degradation or ping-pong effects that can occur with purely reactive approaches.

Management frameworks further support AI/ML-assisted optimization using analytics from functions such as the Management Data Analytics Function or NWDAF.

Management and KPI Frameworks

Parallel work in SA5 extended energy efficiency into measurable system properties:

• EE KPIs defined for NG-RAN, 5GC, and network slices (typically performance such as data volume or coverage area divided by energy consumption).
• Refined measurements of power and energy consumption for physical and virtualized network functions.
• Support for operators to express energy-efficiency requirements for slices and to receive corresponding reports.
• Energy-saving use cases and solutions that apply across RAN, core, and slicing.

These capabilities turn energy efficiency into an observable and controllable attribute rather than a purely local radio implementation choice.

Performance Gains and Trade-offs

Technique / Approach	Typical Gain (Low–Medium Load)	Key Considerations
Antenna + power adaptation	15–30 % (higher when combined)	Load-aware control needed to limit throughput impact
Cell DTX/DRX	Additional deep-sleep opportunities	Alignment with UE DRX; latency for bursty traffic
SSB-less SCell + supporting features	Reduced always-on overhead	Timing and power-difference constraints
Combined multi-domain + AI/ML	Best daily-average trade-off	Requires good prediction quality and coordination

Gains are highest at low-to-medium loads and diminish as cells approach full capacity. Careful, load-aware configuration is essential to keep impacts on user experience minimal. Multi-vendor interoperability and legacy-UE support were explicit design goals.

Broader System Impact

Release 18 elevates network energy efficiency from vendor-specific optimization to a standardized, multi-domain capability. RAN techniques reduce the dominant energy consumer, while management and AI/ML frameworks provide the visibility and intelligence needed for end-to-end optimization. Operators can lower electricity costs and carbon emissions while continuing to meet performance targets. Device-side power-saving features (extended eDRX, ongoing low-power wake-up work) further contribute to overall system efficiency.

In practice, these tools allow networks to scale energy consumption more closely with actual traffic demand—an essential step toward sustainable 5G-Advanced and future 6G operations. Early evaluations and trials have confirmed the magnitude of the savings under realistic load profiles, confirming the commercial relevance of the Release 18 energy-efficiency features.

---

22) Network Operations and Automation

Release 18 marks a clear step from largely reactive, rule-based Self-Organizing Network (SON) features toward more predictive, data-driven, and closed-loop automation. Operators gain standardized tools to optimize energy consumption, load distribution, and mobility in a multi-vendor environment while retaining control over performance and service quality.

AI/ML for the NG-RAN – Priority Use Cases

The first normative AI/ML work for the NG-RAN focuses on three high-value operational use cases:

1. Network Energy Saving

AI/ML models generate energy-cost and load predictions that help the network decide when to activate or deactivate cells, carriers, or antenna configurations. Predictions can be exchanged between neighbouring gNBs, enabling coordinated rather than purely local decisions. This reduces the risk of coverage holes or quality degradation that can accompany aggressive energy-saving policies.

2. Load Balancing

Predictive load balancing uses traffic and mobility forecasts to steer users or traffic across cells and carriers before congestion occurs. This improves resource utilization and reduces the need for reactive load-balancing actions that can cause temporary service disruption.

3. Mobility Optimization

Trajectory prediction and related models improve handover success rates, reduce unnecessary handovers, and support more robust mobility under energy-saving or multi-TRP configurations. Combined with L1/L2-triggered mobility (LTM), these capabilities lower signalling load and improve user experience during movement.

Data collection for these use cases is standardized, including the exchange of measurements and predictions over the Xn interface. Model lifecycle management, performance monitoring, and integrity aspects are addressed at the system level so that operators can deploy and govern AI/ML functions with appropriate trust and control.

Data Collection, Exposure, and Closed-Loop Automation

Release 18 strengthens the information flows needed for automation:

• Enhanced measurement reporting and data-collection frameworks supply the inputs required by AI/ML models and traditional SON functions.
• Prediction exchange between NG-RAN nodes enables distributed intelligence rather than purely centralized decisions.
• Management frameworks (including energy-efficiency KPIs, performance measurements, and analytics functions such as MDAF/NWDAF interactions) provide operators with visibility and the ability to set policies or constraints.
• Support for intent-driven and autonomous-network concepts continues to mature, allowing higher-level goals (for example, energy targets or load thresholds) to drive lower-level configuration and optimization.

These elements move networks closer to closed-loop operation in which measurements → analysis/prediction → decision → action occurs with less human intervention.

Operational Flexibility through Network Slicing Phase 3

Slicing enhancements directly improve day-to-day operations:

• Partial Allowed NSSAI and partially rejected S-NSSAIs allow a slice to be supported only in a subset of cells within a Registration Area. This removes the previous rigid requirement for uniform availability across an entire Tracking Area and greatly simplifies localized (campus, enterprise, event) deployments.
• Network slice replacement enables operators to redirect traffic from one S-NSSAI (or slice instance) to an alternative during congestion, maintenance, or planned changes while preserving service continuity.
• Enhanced Network Slice Admission Control (NSAC) and quota management (including roaming scenarios) give finer operational control over resource consumption per slice.
• Optimized signalling for dynamic or temporary slices reduces overhead when slices are activated and deactivated frequently (events, emergency response, short-term enterprise needs).

RAN-side support includes zero-resource configuration for slices outside their intended service area and the exchange of this information over Xn, allowing precise resource isolation and mobility decisions.

Complementary Management and Orchestration Advances

Additional operational capabilities include:

• Refined Network Resource Models, performance measurements, and KPIs (including energy-related metrics) for better observability.
• Support for asynchronous lifecycle management and scheduled slice availability.
• Improved fault supervision and alarm collection relevant to edge and other advanced deployments.
• Charging and exposure enhancements that align operational control with business and multi-tenant models.

These management features ensure that the new radio and AI/ML capabilities can be configured, monitored, and governed consistently across multi-vendor networks.

Benefits for Network Operations

Capability	Operational Benefit	Typical Impact
AI/ML energy saving	Proactive, coordinated cell/carrier management	Lower energy cost with controlled QoS impact
AI/ML load balancing	Predictive traffic steering	Higher utilization, fewer congestion events
AI/ML mobility optimization + LTM	Fewer failed or unnecessary handovers	Reduced signalling, better user experience
Partial NSSAI & slice replacement	Flexible, localized, and resilient slice deployments	Simpler enterprise/campus operations
Enhanced data collection & KPIs	Better visibility and closed-loop control	Reduced manual tuning
Management model updates	Consistent multi-vendor configuration and monitoring	Lower opex for complex networks

Overall Impact

Release 18 shifts network operations from largely static or reactive configuration toward predictive, data-driven automation while preserving operator control. AI/ML use cases for energy, load, and mobility address three of the most persistent operational challenges. Flexible slicing and improved management frameworks reduce the complexity of supporting diverse services and multi-tenant environments. Together these capabilities lower operating costs, improve resource efficiency, and enable more consistent service quality—even as networks become denser and more heterogeneous.

Early commercial interest focuses on energy-saving AI/ML functions and LTM because they deliver measurable opex and experience gains with relatively contained deployment complexity. As models, data pipelines, and multi-vendor interoperability mature, broader closed-loop automation is expected to become a defining characteristic of 5G-Advanced networks.

---

23) Enhancements for User Equipment (UE) in 3GPP Release 18

Release 18 continues the dual track of proliferating 5G to more device classes (especially lower-cost and lower-power IoT) and strengthening capabilities for advanced consumer, industrial, and immersive devices.

1. Reduced-Capability and IoT-Oriented UEs (eRedCap and RedCap Evolution)

Enhanced Reduced Capability (eRedCap) further scales down complexity relative to Rel-17 RedCap:

• Peak data rate limited to ≤ 10 Mbps (downlink and uplink).
• Optional reduced baseband bandwidth for data channels (approximately 5 MHz while RF remains up to 20 MHz).
• Extended discontinuous reception (eDRX) cycles in RRC_INACTIVE, reaching up to several hours (on the order of > 10 s minimum, with much longer supported values).
• Focus on FR1; coexistence signalling with other NR device types; new RAT type indications for network control and roaming.

These changes target cost- and power-sensitive use cases previously served by LTE Cat-1 / Cat-1bis, while remaining native to 5G SA networks.

Additional RedCap-related UE support includes long eDRX for RedCap UEs in RRC_INACTIVE and positioning enhancements (frequency hopping beyond the RedCap maximum RF bandwidth for PRS reception and SRS transmission) so that lower-complexity devices can still achieve useful positioning accuracy.

2. Higher-Capability and Broadband UEs (MIMO and RF Enhancements)

Uplink MIMO evolution significantly raises peak and average uplink throughput for capable devices:

• Support for up to rank-8 / 8-layer PUSCH transmission.
• Simultaneous multi-panel uplink transmission (especially relevant for mmWave and multi-TRP scenarios).
• Expanded orthogonal DMRS ports (up to 24) for improved multi-user MIMO pairing.
• Enhanced CSI feedback for medium- and high-velocity scenarios and coherent joint transmission.

RF and antenna-chain enhancements target form factors less constrained than handheld smartphones:

• Expansion of simultaneous uplink transmission chains (for example, up to 4 chains in single-carrier operation for FWA, CPE, vehicular, and industrial UEs), enabling higher-rank MIMO or better diversity and higher total transmit power configurations.
• Support considerations for 4Rx handheld UEs in low NR bands (< 1 GHz) and additional Tx configurations for inter-band UL CA and EN-DC.
• Updated TRP/TRS requirements and test methodologies for FR1.

These features primarily benefit fixed wireless access CPE, high-end industrial devices, and vehicle-mounted UEs, while still delivering capacity gains for multi-user operation with conventional smartphones.

3. XR, Media, and Immersive UEs

XR-specific UE enhancements improve both capacity and power efficiency:

• UE assistance information that makes the network aware of XR traffic characteristics, frame rates, and processing constraints.
• Support for non-integer DRX periodicities aligned with common XR frame rates (for example, 60, 90, 120 fps) to reduce unnecessary wake-ups.
• Configured-grant enhancements (multiple occasions, dynamic indication of unused occasions via UCI) and PDU-set / data-burst awareness so the UE and network can handle video frames more efficiently.
• Signalling for shared processing resources when simultaneously receiving unicast and MBS broadcast, allowing the network to adjust unicast configuration accordingly.
• Media capability definitions and real-time transport configurations (RTP header extensions, RTCP feedback) for AR conversational services and split rendering.
• Potential reduced-Rx antenna classes for AR/VR form-factor devices with corresponding performance requirements.

These features enable longer battery life and higher supported XR user density on power- and form-factor-constrained devices such as glasses and headsets.

4. Mobility, Coverage, and Robustness Enhancements

L1/L2-Triggered Mobility (LTM) requires UE support for Layer-1 measurements, early downlink/uplink synchronization to candidate cells, and MAC-CE-based cell-switch commands. The result is substantially lower handover interruption times (typically in the 20–30 ms range or better), improving experience for mobile users and latency-sensitive applications.

Uplink coverage enhancements applicable to ordinary commercial smartphones include:

• PUCCH repetition for Msg4 HARQ-ACK (especially valuable in NTN but also beneficial terrestrially).
• PUSCH DMRS bundling with phase continuity across transmissions (including under timing drift in NTN).
• Enhanced multi-slot aggregation and TB repetition for PUSCH, dynamic PUCCH repetition factors, and related PRACH/MSG3 improvements.

Additional coverage and coexistence features (frequency-domain spectrum shaping, power-class reporting, in-device coexistence enhancements) further improve practical UE performance.

5. Positioning and Location Capabilities

UE positioning support is broadened and refined:

• Carrier-phase positioning (CPP) for centimetre-level accuracy potential.
• Low-Power High-Accuracy Positioning (LPHAP) enabling UL and DL+UL positioning while the UE remains in RRC_INACTIVE.
• Bandwidth aggregation of positioning measurements across up to three intra-band contiguous carriers.
• Frequency hopping for RedCap UEs to achieve effective measurement bandwidth larger than their RF bandwidth.
• Sidelink positioning with SL-PRS for relative and absolute ranging, including out-of-coverage scenarios.

These capabilities extend high-accuracy and low-power positioning to a wider range of device classes.

6. NTN, Sidelink, MBS, and Multi-USIM Support

• NTN-capable UEs (especially commercial smartphones): uplink coverage techniques noted above, support for new L/S and Ka bands where applicable, network-verified location procedures, and mobility enhancements (including RACH-less handover and conditional handover with time/location triggers).
• Sidelink UEs: operation in unlicensed spectrum (SL-U), carrier aggregation (for example in ITS bands), and coexistence with LTE sidelink.
• MBS-capable UEs: multicast reception while in RRC_INACTIVE, and signalling for concurrent unicast + broadcast processing resource sharing.
• Multi-USIM (MUSIM) devices: paging collision avoidance, network-switching notification, and temporary capability restriction/removal mechanisms to improve dual-SIM behaviour in idle and connected states.

7. Measurement, Policy, and Other UE Features

• Application-layer QoE measurement collection from the UE for streaming, MTSI, and VR services (including multicast/broadcast sessions in relevant states).
• UE policy enhancements in the 5GC.
• RRM, SON/MDT-related reporting improvements, and support for mobile IAB scenarios (cell reselection prioritization, RACH-less handover involving mobile IAB nodes).

Summary of UE Enhancement Categories

Category	Key UE Capabilities	Primary Device Types / Benefits
Complexity reduction	eRedCap (≤10 Mbps, optional ~5 MHz BB, long eDRX)	Low-cost IoT sensors, trackers, wearables
High-performance UL	Rank-8 UL, multi-panel, more Tx chains, expanded DMRS	FWA CPE, industrial, vehicular, high-end devices
XR / Immersive	Frame-rate DRX, PDU-set awareness, assistance info, shared processing	AR/VR glasses, headsets, immersive terminals
Mobility & Coverage	LTM support, PUCCH/PUSCH repetitions & DMRS bundling	Smartphones, mobile users, NTN handhelds
Positioning	Carrier phase, LPHAP (Inactive), BW aggregation, RedCap FH, SL positioning	Trackers, industrial, public safety, automotive
Connectivity expansion	NTN smartphone features, sidelink unlicensed/CA, MBS Inactive, MUSIM	Satellite-capable phones, V2X, dual-SIM, group services

Overall Impact

Release 18 expands the UE landscape in both directions: lower-cost, longer-battery eRedCap devices bring native 5G connectivity to high-volume IoT, while higher-rank uplink, multi-panel, RF-chain, XR, and positioning enhancements raise the performance ceiling for advanced consumer, FWA, industrial, and immersive devices. Mobility, coverage, NTN, and multi-USIM improvements make everyday smartphone operation more robust. Collectively these UE enhancements enable broader 5G adoption across price points and form factors

---