Co-creating a cyber-physical world

6G development is now entering a more concrete phase of regulation [1] and standardization [2]. Even though much has happened in the industry, the Ericsson 6G vision first announced in 2020 remains relevant for transforming 2030’s business and society with 6G technology. We envision a future where individuals and digital entities have access to enhanced communication services delivered securely, efficiently, and sustainably. In response to this, many of the world’s leading telecom stakeholders are converging on the problems that need to be addressed and the goals that should be accomplished (discussed in [3]), for a continued successful global collaboration in communication networks. Simultaneously, the insights will be drawn from the ongoing expansion of 5G networks worldwide.

The future networks must tackle numerous societal challenges including resilience and sustainability while also supporting new and evolved use cases and capabilities, as well as the traffic increase, predicted to be 3X in 2029 [4]. The network services must continue to respond to actual needs of people and enterprises and be delivered in a way that fits the users’ needs and creates value. Ericsson’s proposal for achieving this – the 6G platform – aims to deliver advanced telecommunication networks that offer enhanced optimization and efficiency for communication service providers, along with new monetization opportunities. It will also grant access to a wider set of network services to enterprises and developers, while delivering new experiences to end consumers. 6G should support applications with guaranteed services involving more than communication, with value in focus.

By leveraging technological breakthroughs and applying new design concepts, flexibility and efficiency can be achieved while meeting the rigorous demands of new services. On the 6G platform, continued enhancements in the communication technology will be complemented with adjacent services beyond communication, providing a wider range of solutions to a larger customer base compared to 5G. Generative AI, among other AI technologies, will play a crucial role in automation of network management and operations [5]. The networks should build on and expand further on the 5G capabilities and be better integrated into the rest of the information and communications technology (ICT) ecosystem of cloud platforms and applications. Specifically, drawing from the insights gained from 5G, 6G networks will need to address challenges related to complexity, operating costs, device scale, and expansion into new business segments.

This white paper presents a scenario for future communication needs and beyond, gives an overview of the main components of the 6G network platform, and a projection of its potential applications. Finally, a roadmap is being outlined to establish viable 6G systems by the 2030s, ready to transform business and society.

The use cases in Figure 1 entail increasing demands on performance, efficiency, sustainability, and privacy of the underlying communication networks, but also have needs for services beyond communication. As these are instrumental in enabling desired consumer experiences of applications, 6G can expand into a platform offering a wider range of services. Applications in the cyber-physical era will benefit from a richer and easier access to network capabilities through service APIs (Application Programming Interfaces), and the ability to use them in national networks or globally through aggregators.

A platform is a system that enables other businesses to share information, or build applications or content on top of it. Communication services are a key foundation for all forms of digital businesses and will continue to evolve towards tailored and programmable connectivity services leveraging customized quality of service guarantees and network slices. With the advent of 6G, networks will continue their journey towards becoming a platform that provides services going beyond connectivity, building upon the foundation laid by 5G:

Information services: The network contains and can produce information with the potential to significantly improve applications. This includes sharing network information for performance prediction and preventing mobile fraud; using positioning information to enrich existing device location sensors, including global navigation satellite system (GNSS); and providing sensing data from the radio access network (RAN) to applications in need of updated spatial data of a specific environment.

AI and compute services: the expected AI native architecture of 6G will include many functions in the network related to AI and distributed handling of large amounts of data. Some of these functions might not only be useful to operate the network but could also be exposed as services like Artificial Intelligence as a service (AIaaS) to the benefit of external parties, for example, supporting digital twins. With the same rationale, computing resources in the network could be exposed to mobile applications, for instance to offer dynamic computational device offloading to improve both device behavior (e.g., such as battery, heat) and user experience.

The 6G platform [6] as depicted in Figure 2, is a network platform built on 6G technologies and services. Based on these assets, the platform has capabilities for rich service interaction and assurance towards application service providers (ASPs), enterprises, and consumers. Communication service providers (CSPs) are in a good position to offer crucial digital services in their current local markets. However, reaching a global market has historically proved challenging. By leveraging aggregator platforms, such as network aggregators like Ericsson GNP or existing Cloud providers acting as service aggregators, CSPs can offer global services to a broad customer base and leverage developer communities. Scaling 6G-empowered applications through distribution channels that have global reach, that is, application platforms such as app stores, would create synergy effects both for telecommunications and the App economy – covering applications and application platforms, in-app purchases, third party ads, and so on – spurring further growth in both sectors.

APIs are the means for the 6G platform to expose its capabilities to application developers. In the era of 6G, application developers will have access to a much wider range of APIs. These APIs will build on capabilities from devices and servers, multiple network domains, and services beyond just connectivity. The 5G exposure solution continues to evolve to support the 6G requirements, that is, to provide new and customized APIs more easily and rapidly. We see that service composition will be increasingly used to facilitate customized APIs. Automation support by the 6G platform will therefore be crucial to foster rapid service innovation and creation as well as to simplify life cycle management of APIs.

For services offered with performance and reliability commitments, typically expressed in a service level agreement (SLA), there is also a need for mechanisms to assure that they are fulfilled. Depending on the customer type, the network platform would provide customized SLAs, or template-based SLAs through a service-portal or from within an API. The introduction of intents provides a structured way of capturing both service and network requirements and their relative importance, including requirements from SLAs needed for service setup and assurance.

To give a concrete example, different 6G platform components could work together to support an immersive mixed reality (MR) application based on a lightweight head mounted display (HMD). To facilitate smooth application performance, the 6G platform will assist devices and ASPs with several services (see Figure 2): Tailored connectivity with user/ application specific performance guarantees; positioning and environmental sensing information from the network to enhance existing device sensors for improved localization and mapping; and compute services to seamlessly offload and execute compute-intensive tasks (for example, map optimization or object detection [7]) within the device-to-cloud compute continuum, improving both HMD behavior and application quality of experience (QoE).

As illustrated in this example, such an immersive experience requires a complex interworking of many services. The user, however, will judge the quality of the overall experience mostly by the steady and smooth interaction of virtual objects with the real world as well as the convenience of carrying the HMD in a mobile scenario. Good QoE will thus reflect favorably on the ASP’s offering. ASPs and consumers have an incentive to use the best communication and other services to reach that QoE.

AI will have a central role in realizing the 6G system, both as enabler to increase efficiency of existing network features, and introduce new features that were previously not possible to implement using non-AI technologies. To this end, we introduce AI native, defined as “the concept of having intrinsic trustworthy AI capabilities, where AI is a natural part of the functionality, in terms of design, deployment, operation, and maintenance” [9].

AI is pervasive throughout the entire AI native architecture (see Figure 5). This can be achieved in new products by considering AI from the start. Legacy products can evolve into AI native where this makes business sense. We see four main aspects of an AI native architecture (see also Figure 6):

1. Intelligence everywhere
2. A distributed data infrastructure
3. Autonomous operations
4. AIaaS

The aspect of intelligence everywhere is about executing AI workloads wherever it makes sense based on a cost-benefit analysis. That means, in every network domain, on every layer of a stack, on every physical site from central to edge sites, and possibly even on mobile devices. This also implies that AI execution environments need to be available everywhere, and AI training environments might be co-located if needed.

We envision a fully automated trustworthy model lifecycle management, constantly improving and following data changes, to achieve system-wide end-to-end gains. It ensures that models operate accurately and robustly even with unseen data. It decides what model version to use and where and when to perform (re)training. Models may require data spanning several layers and network domains, which may imply that layer and domain borders blur. Models with similar input features might be combined. Models might also be trained jointly with other models solving similar tasks using different inputs to benefit from multi-modality, thus reducing the number of models that need to be maintained. Additionally, models may be split into different locations, and trained in a collaborative fashion to preserve data privacy and utilize resources more efficiently by reducing the volume of data that is exchanged [10].

Intelligence everywhere is interconnected with the aspect of a distributed data infrastructure (see also the chapter on 6G Architecture). Executing and (when needed) training AI models in any location is only possible if data and necessary compute resources are available everywhere. Generative models may be used alongside network observations as data sources. Observed data may have a best-before date or legal constraints. The sheer volume of data could impose limitations, restricting when and where data can be utilized. Data streams might need to be processed or combined. Data observability needs to be flexible to adapt to the requirements of the data consumer and the available resources of the data producer and the transport infrastructure. This implies that the data infrastructure and the model orchestrators need to interact; sometimes data can be transported to the intelligence, and sometimes it is more efficient to bring the intelligence closer to the data.

There is a need to further automate the management of the network. Instead of introducing new manual operations (where humans decide what to do and how) or automated operations (where humans design workflow executions), the aim should be for fully autonomous operations [11]. Instead of instructing the system on specific actions to take, humans would express requirements to the system and supervise that those requirements are fulfilled.

Intents are a vital technology for achieving autonomous operations. We define an intent as a “formal specification of all expectations including requirements, goals, and constraints provided to a technical system” [12]. This includes requirements from diverse customers (for example, derived from SLAs) and the service provider for expressing its business policies. Various AI technologies, including Generative AI and Language Models, may be used to implement intent handling.

The AI native architecture aspects of intelligence everywhere, distributed data infrastructure, and autonomous operation require novel functions in the network related to AI and data handling. Some of these capabilities may be exposed as services to customers of the service provider; also known as AIaaS [13]. Examples include a model training or execution environment, or data exposure. Exposing such services turns the network into a platform for innovation.

For all the four main aspects of AI native, standardization has already started. This includes standardization of distributed intelligence in 3GPP SA2 (intelligence everywhere aspect), and standardization of intent interfaces in TM Forum, O-RAN, and 3GPP SA5 (autonomous operations aspect). We expect standardization of the AI aspects to continue into the 6G timeframe.

Cloud infrastructure

As telco cloud infrastructures evolve towards pure cloud native-based solutions, network functions are deployed as cloud workloads (for example, in containers) to provide the necessary network agility and innovations required for 6G. For this purpose, cloud infrastructures are becoming highly heterogeneous, proposing a combination of advanced CPUs and hardware (HW) accelerators to further enhance the overall performance and energy efficiency of existing and future compute-intensive workloads. This combination of newly emerging hardware can efficiently perform specific tasks to fulfill stringent 6G requirements, allowing energy efficiency and sustainability solutions without compromising on performance. Additionally, it is fit to enable AI technologies to empower an innovative 6G ecosystem. This means that 6G standards need to be designed to accommodate the integration of the latest hardware and software innovations in future network implementations, including commodity HW technologies (such as CPUs and new instructions, special-purpose accelerators, smart networking devices, server and interconnect technologies, and so on) and industry-standard frameworks and APIs (for example, DPDK, CUDA, DOCA, OneAPI, P4, Kubernetes, etcetera) enabling the adoption of the latest and most relevant HW technologies.

Computational offloading

As cloud infrastructure used for 6G will form the base of the vastly distributed mobile networks, there is an opportunity for CSPs to take advantage of these assets to offer additional services alongside communication services. An example is a computational device offloading service [16] exposed by developer-friendly APIs. Computational offloading aims at a dynamic and flexible deployment of highly granular application tasks, triggered by situational changes. This is a use case agnostic, offering sandboxed, operator- deployed compute to any application running on a mobile device such as smartphones, MR headsets, or drones. Such a service is attractive for subscribers as it facilitates prolonged battery life, reduced device heat, and improved application experiences. Offloading can also be used by application providers to differentiate their offering, by customizing application features and performance profiles, or to address privacy concerns with guaranteed data locality compared to many current public cloud alternatives. Finally, as discussed in the Chapter on the 6G network platform, richer service offerings like computational offloading allow CSPs to leverage new revenue streams by bundling it with connectivity services or other network services like sensing data or AIaaS, while reusing existing network procedures like authentication and management functions.

Integrated sensing

Integrated sensing and communication (ISAC) is an example of a service where the same network infrastructure can be used for multiple purposes, such as sharing base stations, spectrum, data planes and pipelines, and compute between sensing and communications services [17]. This is, of course, helpful from a cost and deployment perspective. Still, some additional functions need to run in the network to build a new service such as sensing. Base stations must employ specific radio signals for sensing. Control mechanisms are needed to establish sensing in the network, and processing is needed to interpret the raw data. In addition, a rich API towards applications is needed so that they can request different sensing services needed in diverse immersive and awareness types of use cases. Privacy questions are also paramount. Sensing needs to be performed in a manner that ensures the privacy of individuals as well as the society as a whole, and sensing applications should prioritize privacy assurance.

Device diversity

The diversity of device types is expected to increase further in 6G timeframe; to serve both new and evolving use cases, see Figure 9. This device diversity, which will be active concurrently in a 6G system, makes it pertinent to understand the characteristics and requirements brought about by both novel and evolving device types. This is, in addition, to shared common requirements including improved energy efficiency, enhanced user privacy, and the adoption of relevant novel spectrum such as in the centimetric range.

Pertaining to insight use cases, we envision two 6G IoT device types. First one is the low power wide area (LPWA) devices, which desirably operate in 3-5MHz, have further reduced device complexity, similar or lower coverage enhancements, reduced signaling and UE power-saving features than NB-IoT or Cat-M1. The second one, likely in a later 6G phase, is the zero energy IoT device that obtains energy from harvesting or rely on backscattering communication and thus very low powers resulting in inherently short-range.

To realize immersive experiences, the devices such as MR devices will bring about requirements [18] on guaranteed data rates with low latency (for example, less than 20 ms) driven by motion-to-photon latency and compute location considerations. The different service flows and their attributes such as compressed, or uncompressed, along with application aspects like the exact scene or its parts [19] that are processed together will drive the communication link reliability requirements. The device form factor limitations would bring unique requirements, for example, on antenna locations and the devices with compute capability constraints would benefit from compute offloading located on network edge. The latter in turn would bring forth requirements on joint compute-communication resource optimization, communication protocol optimizations, and so on. The devices with enhanced AI capability could facilitate semantic communications, reducing bandwidth use.

Inclusivity will be ensured with not only devices that support non-terrestrial network access and fixed wireless access, but also ones that are cost-effective.

The resilient use cases require similar network and device capabilities, to ensure the appropriate level of end-to-end resilience availability and reliability. Different approaches are possible, where some of them require reliability improvements for devices used for critical use cases. When the modems are unable to fulfill reliability requirements or when economies of scale are insufficient for the necessary modem adaptations, it is possible to leverage devices with redundant modems. These redundant modems can be connected to different networks and help avoid certain common points of failure. Balancing requirements between networks and devices requires close co-operation with respective vendors to evaluate suitable approaches.

Transport

The transport network plays a vital role in the mobile network by providing the necessary connectivity between the core network and the RAN, and between different entities within the RAN, such as RU, DU, and CU. The mix of technologies used to realize the transport network should evolve and enable CSPs to build, or upgrade existing, transport networks that are scalable, easy to maintain, smart and lean, and support the various deployments and new services provided by 6G.

The architecture of the transport network is determined by the mix of mobile network deployments, which come in many different sizes and shapes. For example, the more demanding centralized RAN (CRAN), where baseband and radio units are in different locations and are connected through fronthaul interfaces, or the simpler but likely more common distributed RAN (DRAN), where baseband and radio units are located at the same cell site and cell sites are connected through backhaul interfaces.

Furthermore, the transport networks need to concurrently support multiple services that can have strongly differentiated requirements such as eMBB, MR, or critical services. This means that the transport network also must deliver high precision network synchronization. Hence, we need transport technologies that can also handle these extreme cases when and where needed. To meet these 6G requirements, the transmission and switching technologies (packet, optical, and wireless) used in the transport network need to evolve by taking an end-to-end network perspective into account and identify solutions for network orchestration and automated management that results in a smarter network that can do more without being over-dimensioned.

Finally, it is important to understand that for most deployments, an ideal transport network is one that meets the diverse service requirements in a reliable, cost-efficient, automated, and sustainable manner, always maintaining appropriate and relevant dimensions.

The security in 6G evolves from 5G security, with additional security needed to support new use cases, technology, and paradigm shifts. The role of mobile networks as part of critical infrastructure comes also with new regulatory requirements on security because the availability and proper functioning of the networks are of interest to the society at large. Examples of such requirements are demands on higher security assurance and adherence to zero-trust principles [20].

Meeting 6G requirements requires combination of standardization, implementation, deployment, and operational aspects and processes. It is important to take a holistic view on the security of 6G networks to keep up with the evolving threats. These threats not only include traditional issues such as data confidentiality and fraud but also large-scale attacks on network availability, spyware attacks, and attacks on critical national infrastructure. Find more details on security in [21].

The focus of 6G discussions has shifted from research to pre-standardization alignment, where major regional organizations are formulating their views on the content and objectives of 6G. This includes both timeline discussions as well as the content of the first phase of 6G. There is a need to align these activities and find global consensus without making the scope too wide.

From a timeline perspective, several steps have been taken. As of 2023, 3GPP has committed to developing 6G specifications. A general expectation in wireless industry is that the first commercial deployments of 6G will happen around 2030. More of this is elaborated on in the Ericsson blog post [22], where the relation to the ITU-R work and timeline is also discussed. It is also important to note that 3GPP is not the only standard development organization (SDO) shaping the future, other SDOs and regulations will matter as well (for example O-RAN Alliance, TM Forum, and CAMARA).

6G should emerge as a global standard, offering support not only for the development of new use cases but also for addressing the widespread adoption of use cases initiated in 5G. Supporting multiple use cases is not unique to 6G, but in previous generations features targeting new verticals and deployment scenarios were added in later releases. The late introduction of these, sometimes complex, features resulted in delayed commercial availability, also considering the large development costs required to enable initially relatively small markets. Therefore, 6G should support new and evolved use cases with total cost of ownership (TCO) efficiency and sustainability in the center.

Considering this, it is crucial that the initial 6G radio access provides fundamental support for key use cases and deployment scenarios, while refraining from optimizing for the most extreme scenarios. This includes paving the way for massive immersive communication driving traffic and new requirements, low-power wide-area IoT, fixed wireless access, time- critical communication, and non-terrestrial access.

Some steps will be crucial in paving the way for 6G, as shown in Figure 10. The first step is to move to 5G standalone supporting a rich set of MBB and Enterprise services. As a follow-up there are possibilities to add 5G-Advanced features to further expand 5G. Therefore, CSPs have the tools to monetize 5G services and capabilities including slicing, SLA assurance, observability, and network service exposure (APIs). Following these steps it is possible to quickly deploy 6G in existing bands using low overhead 5G/6G multi-RAT spectrum sharing (MRSS) enabling wide area support for new 6G features and use cases. It is therefore critical to ensure that adequate support is in place from the beginning.

To conclude, the following are key points for 6G:

• The day-1 baseline of 6G should offer broad support for evolved and new use cases enabling cyber-physical interactions. A diverse range of devices and network deployments should be supported.
• Programmable connectivity and beyond-communications services should be exposed as APIs to applications, creating a platform for innovation with a global market.
• An evolved 5GC should support smooth introduction of a standalone 6G RAN with a lower-layer split. The architecture should be AI native and support a high degree of automation.
• All spectrum currently used by 3GPP systems can be used by 6G, complemented by new centimeter-wave bands. Highly efficient 6G-5G sharing (MRSS) is essential in primarily FR1 and FR2.
• The 6G RAT should deliver higher performance and higher efficiency.
• Overall, focus should be put on minimizing complexity and generating value.