The first version of 5G has been rapidly deployed in large parts of the world. This is still only the beginning of a continuous 5G evolution. Starting from the deployment of 5G Standalone (SA) as a basis, communication service providers (CSPs) will enjoy further improvements in the next few years by providing new use cases, higher performance, and leaner networks. To support this, 3GPP has from Release 18 started the specification of 5G Advanced.
In 5G Advanced, the 5G new radio (NR) and 5G core (5GC) evolution is being continued to ensure the success of deployed systems globally and to expand the usage of the 3GPP technology by supporting different use cases and verticals.
Sustainable network design is one of the cornerstones of the 5G Advanced system. Artificial intelligence (AI) and machine learning (ML) will play an important role in addition to other technology components providing support for extended reality (XR), reduced capability (RedCap) devices, and other new market segments. While Ericsson 5G networks already support AI/ML and XR use cases in an energy-efficient manner, 5G Advanced offers standardized solutions for enhancing performance and enabling a new types of applications for these use cases. The 5G Advanced standardization is an important step in the evolution of cellular wireless access toward 6G.
This whitepaper provides an overview of 5G Advanced to show the main advantages of its technology components regarding network performance and capabilities. Guidance is provided on which features to expect in 5G Advanced Release 18 and 19 and it is explained how 5G Advanced provides steppingstones towards 6G.
3GPP Release 18 marks the start of 5G Advanced. 5G Advanced builds on the 5G baseline defined by 3GPP in Releases 15, 16, and 17. Further enhancement of the 5G Advanced system is expected in the forthcoming Release 19 which will start in 2024.
Initial learnings have been derived from commercial 5G networks that have been deployed in large parts of the world. In addition to this, the need for the continuous evolution of 5G networks by supporting new market segments and use cases has prompted 3GPP to begin standardization related to 5G Advanced systems. 5G Advanced also provides steppingstones in areas that will be of importance in the future 6G systems.
This white paper provides an overview of 5G Advanced with a special focus on its most vital pillars. The main technology components and their use cases are discussed in addition to 3GPP’s future direction towards 6G.
After initial studies, standardization of the new 6G system may officially start from Release 21. Figure 1 provides Ericsson’s view of the 3GPP’s 5G Advanced and 6G timeline indicating completion of the first 5G Advanced release in the beginning of 2024 and completion of the first (basic) 6G drop in 2028, followed by 6G evolution.
3GPP technology evolution
Since its introduction in Release 15, 5G has targeted three main use case families, namely enhanced Mobile Broadband (eMBB), critical Internet of Things (IoT), and massive IoT. Together with the support for new verticals added in later releases, the 5G system enables many new use cases compared to previous generations of 3GPP systems. It provides superior network performance in terms of capacity and coverage and has increased the focus on sustainability. The need to enhance the network energy efficiency to reduce the 5G systems’ (5GS) carbon footprint is of vital importance.
Now, new use cases such as eXtended Reality (XR) and learnings from existing commercial 5G deployments demand further optimization of the 5G system. 5G Advanced is the next wave of 5G starting in Release 18. It includes additional capabilities for new market segments as well as architecture enhancements of 5GS. In Release 19 this evolution continues.
5G NR addresses eMBB use cases by supporting different duplex schemes, frequency ranges, MIMO, and multi-carrier operations. In 5G Advanced, eMBB performance will be further enhanced. NR and LTE dynamic spectrum sharing (DSS) enhancements that come in Release 18 facilitate a smoother and more efficient migration from LTE to NR. Thanks to MIMO enhancements the capacity and performance of mobile devices are improved. In Release 19 we anticipate that massive antennas for further improving performance will be in the spotlight.
Critical IoT (cIoT) refers to use cases with stringent requirements on latency and reliability. Some of the most prominent applications relate to factory automation and automotive use cases. 5G addressed cIoT from the very beginning by introducing support for ultra-reliably and low latency communication (URLLC). Releases 16/17 added support for Time Sensitive Networking (TSN) while Release 18 adds support for Deterministic Networking (DetNet) to enable industrial internet. Release 18 is also focused on enhancing the support for XR applications that demand a challenging combination of high data rates and bounded latency. XR is expected to grow in importance in Release 19 and beyond.
5G’s support for massive IoT (mIoT) use cases and low complexity devices was inherited from 4G in the form of LTE-MTC and NB-IoT. These tracks were created in Release 13 and have evolved ever since. The first NR-based optimizations towards lower complexity devices were introduced in Release 17 by standardizing support for reduced capability (RedCap) NR UEs. RedCap reduces the device complexity significantly compared to regular NR modems. Still, RedCap devices provide peak data rates comparable to that offered by LTE Category 1-4 devices.
5G is evolving its support also in positioning, mission critical, air-to-ground, and railways communication. In Release 18, 5G Advanced support for drones is introduced. Railways, mission critical, and utility use cases are supported by an NR system that can operate in dedicated bands with less than 5-MHz bandwidth.
Some NR functionality is beneficial across different domains and not connected to a specific use case only. For instance, NR’s lean design secures efficient use of energy, both on the network and the device side. While the UE energy savings were enhanced in 5G Releases 16 and 17, in 5G Advanced further network energy saving is in focus. One tool that is explored in Release 18 for increasing network energy efficiency is AI/ML. AI/ML based RAN enhancements are also being considered for enhanced mobility and load balancing applications. In Release 19 the use of AI/ML in further applications is anticipated, such as positioning and beam management.
The 5GS deployment flexibility and architecture enhancements in 5G Advanced include for example, enhancements for non-public networks, non-terrestrial networks (NTN), location services, edge computing, UE policy, and network slicing. In addition, the service- based architecture has been extended to the IP multimedia subsystem (IMS) telephony services. The IMS can now use service-based interfaces to the policy control function (PCF), the home subscriber server (HSS), and from the IMS application server to data channel related network functions (NFs).
5GS has inbuilt management features, for example, to virtually partition the system into different slices or to collect various types of measurements for self-optimization that will evolve in 5G Advanced.
5G Advanced pillars
5G Advanced will enhance network performance and add support for new applications and use cases. This paper is focusing on the following four important feature areas, where 5G Advanced is expected to bring significant enhancements:
- 5G performance
- Support for new market segments
- Sustainable Networks
- Intelligent network automation
Support for advanced antenna systems and MIMO is part of 5G’s DNA. In Release 18, MIMO capacity is boosted in both the uplink and downlink thanks to support for enhanced demodulation reference symbols. To improve the support for high data rates to mobile users the MIMO beamforming framework is improved to cater to switching between different beamforming methods depending on the user’s speed.
To support the recent trends of massive antennas becoming even more massive, an extension of the supported number of antenna ports to cater to more radiating elements is needed in Release 19. MIMO-related enhancements on UL coverage and capacity will be important for both XR, mobile broadband, and fixed wireless access (FWA) scenarios. Release 19 should also contain enhancements that enable realizable and cost-efficient coherent-joint transmissions across multiple transmission and reception points (mTRP) on a larger scale (aiming towards a truly distributed MIMO system). To improve current coarse link adaptation (LA) procedures, channel quality reporting enhancements are needed for more accurate LA. For reciprocity based massive MIMO, the noise and interference level experienced by the UE is not available to the gNB at a satisfying accuracy and needs to be predicted which leads to inaccurate LA and inefficient MIMO operation. Thus, enhancements related to CSI acquisition for reciprocity-based DL SU/MU-MIMO need to be specified in Release-19.
Mobility, like MIMO, is a key component of 5G. In 5G Advanced, it has been observed that there is room to improve the service continuity for mobile users. The new L1/L2 triggered mobility (LTM) handover procedure will shorten the handover interrupt time. LTM merges the beam managing framework with the mobility framework and introduces a low-latency mobility procedure for NR, supporting CA and applicable to both FR1 and FR2.
In Release 18 LTM is supported between cells served by the same gNB distributed unit (DU), or by different DUs belonging to the same gNB centralized unit (CU). In Release 19, the LTM framework should be extended to support handover between cells served by different CUs. Release 19 may also include UE reports of measurements performed on not configured neighboring cells that a UE may move to. Release 18 only supports UE measurement reporting of preconfigured cells.
5G Advanced explores, as mentioned earlier also, AI/ML as a tool to improve mobility. Areas of interest are for example to utilize AI/ML to perform beam prediction or to predict device mobility. The journey to enhance the performance with AI/ML has just started and more use cases are likely to be enhanced in Release 19 and beyond.
Support for new market segments
5G Advanced provides enhanced support for several new market segments. These include, amongst others, cloud gaming, immersive reality, indoor positioning, and industrial sensor networks.
The 5G bounded-latency communication capabilities will enable improvements for a wide range of new applications including cloud gaming (CG) and extended reality (XR) which refers to anything from virtual reality (VR) and augmented reality (AR) to mixed reality (MR). In AR, digital elements are added to a live view, usually via a camera on a smartphone or AR glasses. With VR one leaves the physical world and experiences complete virtual immersion. MR comprises the interaction of both real-world and digital objects. In cloud gaming, with the assistance of handheld and wearable devices, either human-to- machine or human-to-human interactions are performed.
There are many emerging applications of XR in media, remote control, and industrial automation, which will benefit from the time-critical capabilities of 5G networks. Mobile service providers can introduce XR to consumers, enterprises, and public institutions to define new practices in areas such as entertainment, training, education, social interactions, and communications1.
The XR and CG use cases require high data rates. The devices are usually expected to be mobile and with a small form factor, which puts a limitation on their available power resources. Moreover, low and bounded end-to-end latency is another challenge when it comes to providing coverage for these applications as outlined in2. In Release 18, the performance of XR services is facilitated by 5GC support for application rate adaption using the low latency low loss system (L4S) for scalable throughput feature. XR application information regarding packet periodicity, jitter, size, and latency requirement will be signaled from the 5GC to the RAN to make the RAN XR aware and allow XR-specific traffic handling for improving power efficiency, latency, and capacity. Release 18 also enables improved buffer management by making RAN aware of a set of packets constituting a media unit, allowing Active Que Management (AQM) to drop a complete media unit instead of individual packets if a packet drop is needed.
In Release 19, further work to improve the XR latency and capacity is anticipated. One opportunity is to use unused UE measurement gaps for data transmission. For UE power savings a promising feature is to add support for multiple DRX configurations for handling multiple XR flows and traffic jitter. Finally, it is important to consider extending the XR awareness framework to cater to XR services not considered in Release 18.
Applying AI/ML as a tool in specific scenarios opens new possibilities. A promising case is applying AI/ML for indoor positioning, e.g., in a factory, warehouse, or office environment. In these environments, GNSS coverage may not be available and 5G based indoor positioning is a valuable complement to outdoor GNSS services. In Release 18 it has been shown that the accuracy of well-established cellular positioning methods can be significantly improved with assistance from AI/ML functions, as mentioned in the section on AI/ML for physical layer enhancements.
Enhancements for network slicing
UE Route Selection Policy (URSP) is a UE policy provided by the network and essential for several network slicing use cases in different market segments. Release 18 provides several enhancements relying on UE and network support. The home network can provide URSP rules to the roaming UE which are specific to the visited network (in addition to URSP rules specific to the home network). URSP rules can be provided to the UE in EPS, enabling consistent use of the UE policy. The network is enabled to detect whether the URSP rules are enforced by the UE and the network can use analytics to adjust URSP rules (see also 5GS architectural enhancements).
Release 17 introduced NR RedCap for support of industrial wireless sensor networks, wearables, and wireless cameras. In Release 18, RedCap support for positioning and further device complexity reduction are specified. Reduced peak data rates to 10 Mbps will enable RedCap complexity in parity with LTE Cat-1 devices. In Release 19, RedCap support for satellite communication should be introduced to enable truly ubiquitous NR IoT coverage.
Industrial and Critical IoT has been an important 5G topic from the start. One example of an industrial IoT use case is media production and delivery which requires bounded low- latency IP communication. A framework of functionalities has been specified for 5G time- sensitive communication (TSC) supporting both Ethernet and IP, covering amongst others UE to UE communication through user plane function (UPF), time synchronization, and 5G time-sensitive networking (TSN) integration.
However, there is a demand to support deterministic networking (DetNet) for application areas requiring not only bounded low latency for IP but also low delay variation and extremely low loss. 5G Advanced has added support for DetNet IP flows in Release 18 based on the TSC framework defined in Release 17 (see also Figure 4, where 5GS is acting as a logical DetNet router in an IP DetNet network). DetNet IP redundancy solution is a candidate for Release 19.
From the start, 5G was designed to meet increasing traffic demands while limiting the power consumption of mobile networks. With 5G Advanced, the focus on network energy savings is further pronounced. The increasing energy consumption of mobile networks is neither sustainable from a cost nor an environmental perspective. Breaking the energy curve is an industry responsibility3.
Energy efficiency has always been an important part of 3GPP considerations by allowing smart sleep modes for mobile devices and exploiting lower bands to extend the coverage while increasing capacity and speed with carrier aggregation of higher bands. In 3GPP Release 18, a dedicated study on network energy savings has been carried out. Key performance indicators, energy consumption models, and evaluation methodologies are all defined. Focus areas, potential techniques, and features for enabling network energy savings were studied. Previously, similar work was performed for user equipment (UE) power savings in Release 16 and 17. For system-level network energy savings, traffic load balancing and sleep modes for gNB were studied for the urban micro and macro scenarios with massive MIMO. The study outcome resulted in support for network energy saving features within four key areas; reduced gNB broadcast transmissions, gNB discontinuous transmission and reception, dynamic gNB DL power, and antenna port adaptation. Antenna port adaption is for example useful together with radios targeting massive MIMO.
Release 19 should build on the work in Release 18 and introduce additional energy saving functionalities. Particularly for secondary cells (SCells), additional Synchronization Signal Block (SSB) transmission optimizations including an on-demand version can further reduce energy consumption. Also, given the number of shared components for RX and TX in a base station, there is energy saving potential from reducing the amount of time the base station receiver is activated. Here, dynamic adaptation of the periodicity of physical random access channel (PRACH) occasions based on the PRACH load of the cell, without going through a full system information update would be useful.
In addition to a specific item on network energy savings, further work is done within the AI/ ML area to support network energy saving, such as defining inter-node energy efficiency prediction signaling using data collected at the RAN interfaces via the AI/ML procedures.
Intelligent network automation
With increasing complexity in network design, for example, many different deployment and usage options, conventional approaches will not be able to provide swift solutions in many cases. It is well understood that manually reconfiguring cellular communications systems is inefficient and costly.
AI and ML have the capability to solve complex and unstructured network problems by using a large amount of data collected from wireless networks. Thus, there has been a lot of attention lately on utilizing AI/ML-based solutions to improve network performance and hence providing avenues for automating and inserting intelligence in network operations.
AI model design, optimization, and life-cycle management rely heavily on data. A wireless network can collect a large amount of data as part of its normal operations. This provides a good base for designing intelligent network solutions. 5G Advanced addresses how to optimize the standardized interfaces for data collection while leaving the automation functionality, for example, training and inference, up to the proprietary implementation to support full innovation flexibility in the automation of the network.
5GS architectural enhancements
5G Advanced provides enhancements of the architecture to support intelligent network automation including RAN management, analytics, and AI/ML model life-cycle management, for example, to improve the correctness of the models. 5G Advanced also supports intent-based management for simplifying network management.
The advancements in the 5GC architecture for analytics and data collection serve as a good foundation for AI/ML-based decisions in the 5GC NFs. Release 18 has added, e.g., Network Data Analytics Function-assisted (NWDAF- assisted) generation of UE policy for network slicing, where the Policy Control Function (PCF) is assisted by slice load analytics, allowing the PCF to adjust the UE Route Selection Policy (URSP) rules. Release 18 has also enhanced the analytics with possibilities to provide it on a finer granular location than a cell.
AI/ML for RAN enhancements
In Release 18, AI-powered network energy savings, load balancing, and mobility optimization are supported.
The selected use cases are supported by signaling enhancements to current NR interfaces such as the UE to gNB radio interface and the inter-gNB Xn interface. The targeted performance improvements are achieved by AI/ML functionality implemented in the RAN. By keeping the AI model implementation specific to vendor incentives, innovation, and competitiveness are ensured.
For Release 19, work will continue with the addition of new use cases and potentially finishing aspects not completed within Release 18. An example of a new potential use case is AI-assisted dynamic cell shaping.
Figure 5 illustrates 5GS support for AI/ML powered functionality including an intent-based management approach. The intent will be received by the RAN from the OAM and take action to support the configured intent.
AI/ML for physical layer enhancements
In Release 18, 3GPP performed a study to investigate how AI/ML can be used to improve functionality in the 5G physical layer (PHY). The work seeks to define a framework to support AI/ML on PHY including aspects such as Life Cycle Management (LCM) including performance monitoring and testing. In Release 18 beam management, channel state information feedback enhancement and positioning accuracy enhancements have been studied to get a good picture of how to standardize support for AI. Among these three use cases, applying AI/ML to positioning has the most promising gains, followed by beam management. Thus, in Release 19, specifying support for these two use cases should be prioritized. Potentially, 3GPP can in Release 19 continue to study the channel state information feedback enhancement and explore some new use cases.
Stepping stone toward 6G
5G systems are currently being deployed at a rapid pace, providing high-speed low-latency connectivity for a wide range of services. There is no doubt that the ongoing transformation will give rise to challenges beyond what 5G and 5G Advanced can meet. The increasing expectations set a clear target for those in the industry and research community—6G should contribute to an efficient, human-friendly, sustainable society through ever-present intelligent communication [ref: 6G – Connecting a cyber-physical world, Ericsson]. Nevertheless, several of the above-discussed 5G advanced technology components can be seen as precursors to some of the 6G building blocks. For example, XR will gradually evolve into immersive communication for human-machine interaction which may pose new requirements on 6G to provide an even better experience. In the area of machine-type communication, RedCap can be complemented by zero-energy devices, a class of devices harvesting energy from the surroundings and providing input to digital twins. AI/ML will also play an important role in the fully data-driven architecture of 6G and the intelligent network platform of the future.
The 5G evolution starting from Release 18 is branded 5G Advanced. As 5G Advanced builds on a baseline defined by Releases 15, 16, and 17, this new marker indicates the aggregated value of the 5G evolution from 2018 and onwards. In Release 18, there are both architecture enhancements and additional capabilities for new market segments. 5G systems are currently being deployed at a rapid pace, providing high-speed, low-latency connectivity for a wide range of services. New services will be introduced, for example, advanced XR services, which will further increase the expectations on network performance. Support for RedCap will widen the range of machine-type communication. Applications requiring real-time networking using IP will benefit from Deterministic Networking providing bounded low latency, low delay variation, and extremely low loss. To meet all these demands efficiently, service providers will increase the use of AI/ML and network automation while continuing the journey toward further reducing energy consumption. It is therefore important for 3GPP to focus on these areas as part of the 5G Advanced work while service providers prepare to leverage the benefits of 5G Advanced systems. These technology components are also important precursors to several 6G building blocks.
Imadur Rahman joined Ericsson in 2008. Rahman is a principal researcher in Research Area Radio at Ericsson Research in Stockholm, Sweden, and is currently managing the 5G Advanced standardization research project at Ericsson Research. Rahman holds a Ph.D. in wireless communications from Aalborg University, Denmark.
Olof Liberg joined Ericsson in 2008 and currently leads the company’s 3GPP RAN standardization team. Liberg has an M.Sc. in engineering physics from Uppsala University, Sweden.
Sara Modarres Razavi
Sara Modarres Razavi joined Ericsson in 2014 and is currently the Cloud RAN innovation program manager at Ericsson. Razavi holds a Ph.D. in infra-informatics from Linköping University, Sweden.
Christian Hoymann joined Ericsson Research in 2007 and led a research group at Ericsson Eurolab in Herzogenrath near Aachen, Germany. Hoymann holds a Ph.D. in electrical engineering from RWTH Aachen University. Currently, he is the Director for European Standardization and Industry Initiatives at the Ericsson CTO office.
Stefan Parkvall joined Ericsson Research in 1999 and is currently a senior expert based in Stockholm, Sweden. Parkvall holds a Ph.D. in electrical engineering from the Royal Institute of Technology, Sweden, and is an IEEE Fellow.
Göran Rune joined Ericsson in 1989 and is currently a senior expert in core network architecture. Rune holds a Lic. Eng. in solid state physics from the Institute of Technology at Linköping University.
Ralf Keller joined Ericsson in 1996 and is an expert in core network migration. His current focus is on packet core architecture and technology in his role as Chief Architect. Keller holds a Ph.D. in computer science from the University of Mannheim in Germany.
Patrik Persson joined Ericsson in 2007 and is currently a 6G program manager director at Ericsson Research. Persson holds a Ph.D. in electrical engineering from KTH Royal Institute of Technology, Sweden.
Asbjörn Grövlen joined Ericsson in 2014 and is currently the SWEA Technical Coordinator at Ericsson Research and head of the Ericsson delegation to 3GPP RAN. He holds an M.Sc. in electrical engineering from the Norwegian University of Science and Technology, Norway.
Daniel Chen Larsson
Daniel Chen Larsson joined Ericsson in 2007 and is currently a principal researcher at Standards & Technology. Chen Larsson holds a M.Sc. in electrical engineering from KTH Royal Institute of Technology, Sweden.