6G system architecture: where innovation meets evolution for a more sustainable and connected world

5G is being successfully deployed worldwide, delivering outstanding radio performance, improved energy efficiency, and advanced network capabilities and services. By the end of this year, 5G traffic volume is expected to surpass 4G traffic. However, as AI-driven applications and immersive experiences become more prevalent, it is crucial to anticipate how traffic patterns will evolve in the future. At the same time, technological advancements are progressing faster than ever and are challenging to integrate in a legacy architecture. The sixth generation (6G) of mobile communication presents a once-in-a-decade opportunity, to make a substantial leap in the system design.

6G will not only build on the lessons learned from 5G research, standardization, implementation and commercial rollouts, but it will also leverage new technological advancements in cloud computing, proliferation of AI and a data-driven design to improve performance and network operational efficiency. Furthermore, 6G will also unleash unparalleled energy savings, trustworthiness and resiliency, reduced total cost of ownership (TCO), and new revenue opportunities to monetize network assets. 

This article presents a holistic blueprint on the 6G system architecture along with a set of key design principles and technical frameworks for a lean, resilient and secure 6G system architecture, which we propose as a foundation for the upcoming 6G standardization efforts in 3GPP, with the aim of shaping the future direction of the technology from 6G day one. 

The 6G system should be built on the principle of a powerful, while lean design for its system architecture. We have learned from 5G that multiple architecture options, broadly categorized as non-standalone and standalone options, resulted in market fragmentation and a slow uptake of 5G standalone deployments, therefore, only a single 6G standalone architecture should be supported. To avoid unnecessary system complexity, 6G standards should focus on true multi-vendor interfaces and avoid duplication of functionalities in different network functions (NFs) and domains to enable a truly interoperable and lean system. Similarly, interdependencies between NFs and domains such as RAN and CN should be minimized to allow them to evolve independently. Furthermore, using harmonized technical frameworks will result in unified tools and a simpler design across different layers and domains.

The 6G system should follow a modular protocol design approach, where network functions are separated in the form of strictly independent modules. This approach means that 6G should be designed with a foundational protocols layer that offers basic functionalities targeting low-cost devices that are common to all the different device types, whereas enhanced functionalities are provided by building upon the foundational layer depending on specific service requirements. This results in reduced complexity, thus improved uptake of new features in 6G devices. Such a modular design can be applied to the radio protocols (both user plane and control plane) as well as the non-access stratum (NAS) layer.

Although 5G provides full resiliency for 5G Core already, 6G should aim for an end-to-end architecture that is fully resilient to enable zero downtime and faster service delivery including resiliency for RAN. A primary means to perform seamless connection transfer from a failing function to an active function can be achieved by (geo)redundancy. As an example, in 5G, this concept is supported for control plane NFs in the Core using the concept of “NF Set” where an NF from a NF Set can take over responsibility for a device when the serving NF fails while maintaining the service continuity. Going forward in 6G, the same should also be supported in RAN. As another example, AI can be utilized for predicting network failures and to trigger necessary actions in a preemptive manner.

Enablers for the use of AI are already partially specified in 5G and 5G-Advanced with intelligence in 5G Core being concentrated only in a subset of NFs like the network data analytics function (NWDAF) and the management data analytics function (MDAF). In 6G, potentially every NF may be AI-powered to enable complex decisions, predict patterns, detect abnormal behaviors, and generate data to be used by other consumers. To enable an AI-native 6G system, it is important that the system is designed in an end-to-end manner supporting the lifecycle of AI components in RAN and CN domains simultaneously. This includes data collection, model transfer and training, model validation, model deployment, definition of APIs to be used by AI agents, and performance monitoring from 6G day one, as only in this way can we harness the transformational power of AI in 6G.

With increasing complexity for configuration and management of 6G networks, end-to-end intent-based automation across all network domains becomes imperative to reduce TCO and service delivery time. Different from 5G, which provides up to level four (high level) autonomy in some areas, in 6G we are aiming to enable full cognitive autonomy in the system. To achieve this, we need to evolve and better integrate technical enablers of automation, like intent-driven management, closed-control loops (CCLs), and digital twins (DTs) altogether with enhanced AI functionalities. Furthermore, data being an integral part of the automation framework, should be made available in a consistent manner to consumers such as AI agents, CCLs and DTs.

In 6G, we aim to leverage the benefits of a streamlined, unified and future-proof exposure framework based on the 3GPP common API framework (CAPIF) to support new value-creating use cases that drive monetization opportunities. CAPIF not only enables seamless integration between value-adding platforms exposing APIs to third parties and the underlying network domains but also can be adapted to new technologies and evolving requirements from developers, applications and their users. By streamlining the exposure framework across network domains, 6G can enable consistent consumer experience and enhance governed access to invaluable data to fuel innovation.

In the 6G ecosystem, collecting, consuming and exposure of data plays a vital role, making consistent data access and efficient data management of utmost importance. 6G relies on data as the fuel for AI training, inference, exposure, and monetization, as well as to achieve cognitive autonomy. As a result, the 6G system must handle and transfer a growing volume of increasingly sensitive data, necessitating concepts such as data minimization and adherence to regulations. A lean, secure and holistic data framework is essential to support these goals, ensuring the secure and efficient management of data from the system itself and the data flowing over the 6G network.

To ensure the security of 6G networks, a security-by-design paradigm must be adopted from the outset. Emerging threats in the 6G era must be anticipated and addressed from day one. For example, quantum-safe cryptography must be supported across the entire network. The pervasive use of AI must be protected throughout its lifecycle, e.g., by using only data with clear provenance from authenticated sources. To prevent data breaches during model exposure, privacy-enhancing technologies can be employed. From an end user perspective, user consent mechanisms must be provided inherently in the 6G day one design. Additionally, 6G security should be improved by mechanisms such as radio security, differentiated security for NAS modules, and support for forward secrecy.

Environmental sustainability is paramount for 6G design, encompassing energy efficiency, carbon awareness, and circular economy principles. This technical framework aims to reduce power consumption and emissions, increase device battery life, and promote an eco-friendly network. It is crucial to integrate this framework into all technical aspects, including AI, data and hardware. AI functionalities should be designed in a responsible way, including governance and resources, for example, to minimize energy demands. There is no AI-native without sustainable AI. Similarly, the data framework must be optimized for efficient data collection and processing by, e.g., minimizing redundant and excess data transfer and storage. Hardware efficiency can be achieved through a scalable architecture, and resource sharing, such as RAN sharing, contributes to circular economy principles.

The figure shows three separated domains: devices (6G UE), access networks (6G NB), and core network; where the 6G node B (6G NB) interfaces (e.g., 6G Uu, 6G N2, 6G N3) are point-to-point interfaces, the 6G UE connects with core NFs (e.g., 6G AMF, 6G SMF, 6G PCF) via 6G N1, and interfaces inside the CN are service-based interfaces like in 5G Core.

The 6G UEs in the devices domain and the respective protocols follow the modular design principle as described above, to optimally support low-cost devices (using only a foundational layer of the chipset stack), broadband IoT devices (providing fast, reliable connectivity with minimal latency), and extended broadband devices (supporting ultra-fast data rates, sensing, AI integration, and much more). 6G UEs enable universal connectivity across both terrestrial and non-terrestrial networks.

In the RAN domain, the 6G NB is composed of a 6G radio unit (RU) and a 6G RAN node based on an evolved fronthaul split (aka lower layer split – LLS), which is defined by the O-RAN ALLIANCE. Furthermore, in the 6G era, RAN sharing among multiple operators is expected to be more popular to meet the telco operators’ environmental sustainability ambitions.

Finally, in 6G Core, open service-based interfaces continue to be defined at NF granularity. Where possible, NFs will be reused from 5G Core (e.g., UDM, AUSF, NRF, NEF, NWDAF). Those NFs follow the evolution path, i.e., they may evolve from current Release-19 specifications. At the same time, the 6G Core will introduce new NFs when and where justified. To enable strict modularity and orthogonality, Nokia envisions 6G AMF and 6G SMF to be new NFs providing essential access control, mobility management and session management functionalities, but with no interdependencies among them, and introducing distributed NAS security termination points. It is worth noting that the 6G AMF and 6G SMF can be a software upgrade from their 5G counterparts, enabling also collocated functions in real deployments. Moreover, new 6G NFs may be introduced in 6G day one or in later releases to support new services such as sensing.

At the core of the proposed 6G system architecture lies a commitment to cost optimization that still fulfills the diverse needs of deployment flexibility, introducing new services and adapting technology advances such as in AI. By leveraging the power of AI and cloud technologies, this architecture enables full and intent-based automation, streamlining operations, and facilitating a cost-effective migration to 6G, thereby not only reducing operational expenses but also paving the way for a more sustainable and accessible future for all.

Beyond cost efficiency, the proposed system architecture prioritizes performance and resiliency. It is designed to efficiently support a diverse range of devices and enhance the network capacity through an optimized system design combined with the introduction of a new radio technology. The focus on full end-to-end resiliency aims to achieve zero downtime, ensuring uninterrupted connectivity and service availability from an end user standpoint.

The proposed 6G system architecture is ready to enable various monetization opportunities. Programmable networks and a unified exposure framework empower developers and businesses to create new services and applications, unlocking a wealth of opportunities for growth and revenue generation.

Recognizing the importance of environmental sustainability, the 6G system architecture emphasizes substantial energy consumption reduction. Our environmental sustainability framework promotes the sustainable use of AI and data, while advocating for green energy usage, contributing to a more environmentally responsible future.

Security, privacy and trust are of paramount importance for the 6G ecosystem. The proposed 6G system architecture prioritizes secure and privacy-preserving data usage, incorporating quantum-safe cryptography to safeguard against future threats. Highly automated identity management and scalable authentication mechanisms ensure flexible and robust security, while AI trustworthiness initiatives foster confidence and transparency.

Finally, the proposed 6G system architecture is committed to digital inclusion. Ubiquitous connectivity through terrestrial and non-terrestrial access, coupled with reduced costs for basic connectivity, aims to bridge the digital divide and ensure that everyone has access to the benefits of the digital world.

By embracing these key design principles and technical frameworks, the Nokia envisioned 6G system architecture is a balanced blueprint for a transformative, future-proof network design. With the above, 6G is expected to be the most impactful, intelligent, sustainable, and cyber-resilient network ever, ushering the society into a hyper-connected world.