Networking trends 2023: Building the platform for next-level digitalization

Immersive user experiences such as virtual reality (VR) and augmented reality (AR) have gained a lot of attention in industry and society in recent years. At the same time, technology giants around the world have made significant investments to speed up technology development in this area, which has led to progress in both extended reality (XR) devices and applications.

The evolution of immersive communication that ties user experience to the physical world has been a recurring theme in my CTO Technology Trends articles for several years now. This year, I want to focus on two specific capabilities that are vital for a cyber-physical world. The first, massive twinning, connects and aggregates the physical world of senses, actions and experiences, and uses that information to generate a programmable digital representation. The second, situational awareness, refers to the use of sensing technologies to monitor the surrounding environment, locate objects, and navigate for smooth and safe interaction in a cyber-physical environment.

Massive twinning – real-time digital replication

The creation of a digitalized and programmable physical world without temporal or spatial constraints will require massive twinning – that is, the creation of accurate, fully synchronized digital twins of everything that exists in the physical world, including humans, physical objects, processes and networked systems. Each digital twin knows its history, can predict its future and has full control of its real-time status. In massive twinning, all digital twins are fully synchronized. Vast amounts of data from the physical world are sent in real time to update the digital twins for use by applications. Actuators and interfaces in the real world carry out functions from the applications for action and experiences.

Massive twinning offers support to minimize resource use and fault prediction. One example of such a use case is a real-time programmable digital twin of an entire smart city that enables precise modeling, monitoring and managing of almost any public or private service, such as utilities, public transportation, public health and environmental monitoring. Another important use case for massive twinning is process optimization in a smart factory to improve productivity, flexibility, quality and sustainability. Massive twinning will also be of great use to communication service providers (CSPs) themselves by making it possible for the CSP to generate a digital representation of the entire network that can be used to simulate network rollout, optimize network operations and initiate predictive maintenance.

Ubiquitous connectivity and integrated sensing are essential components of massive twinning solutions. Massive twinning can involve hundreds of sensors and actuators per square meter, which means that the network needs to provide advanced service coverage wherever and whenever it is needed to a broad range of devices and applications. This includes the aggregation of data rates up to gigabits per second, which leads to stringent requirements on very low latency and high throughput.

In the 6G time frame, the network will provide connectivity using new and significantly more energy-efficient signaling protocols, and many of the sensors and actuators used in digital twins will be zero-energy devices that do not require any batteries or manual charging. To ensure massive twinning deployments at scale, a fully automated process to handle sensor and actuator devices in bulk will be required. High network reliability and availability, as well as stringent end-to-end data and privacy protection, will also be needed to meet the required levels of trustworthiness.

To fully unlock the potential of zero-energy devices, a holistic approach is needed. For the devices themselves, the energy-harvesting technologies – as well as the packaging and low-power integrated circuit design – will need to be advanced. In collaboration with Massachusetts Institute of Technology (MIT), Ericsson is exploring new materials and ultra-low energy designs to harvest the radio signal itself.

Situational awareness – sensing surroundings and locating objects

Situational awareness is the ability to understand the environment and its elements in real time and derive meaningful insights in order to make informed decisions and take appropriate actions. This capability is instrumental to building a cyber-physical world and is based on massive twinning data. The digital world is created by processing the sensing and location information provided by simultaneous localization and mapping (SLAM) algorithms. The SLAM information is used for a variety of purposes including AR, 3D mapping, 3D reconstruction and real-time object tracking.

Spatial sensing capabilities will be an integral part of the future network and have been identified as an innovative feature of future 6G systems. The future network will support spatial sensing by reusing network communication signaling, utilizing the same principles as radar technologies to determine the presence, characteristics and trajectory of objects. This solution is referred to as joint communication and sensing (JCAS) and will enable services for evolving use cases such as smart cities, transportation systems and public safety applications.

The network will collect not only JCAS sensor data but also other data from non-JCAS sensors. The network platform has the capability to process the aggregated sensor data to be used for services such as SLAM, navigation and geofencing. The network platform will offer the processed sensor data through network APIs that are easily consumable for the developer ecosystem.

The main advantage of the network platform in terms of future sensing is that most of the network infrastructure is already in place, providing full-area coverage that facilitates a synchronized multi-static sensory mesh. Another useful feature of the network infrastructure is that the radio signals for sensing work in complete darkness and can even sense local weather conditions.

The foundation of modern network infrastructure is an open architecture with open interfaces and open APIs that enable continuous evolution in line with market expectations on business innovation and growth. Limitless connectivity, trustworthy systems and AI-powered cognitive networks will be instrumental technologies in the evolution ahead, from 5G to 5G Advanced, and on to 6G.

Limitless connectivity with greater openness

Open access interfaces, open service interfaces via network APIs, open network infrastructure layers to cloud, transport and data pipeline interfaces as well as open network applications, such as rApps, are all crucial architectural principles of the network infrastructure.

Service differentiation and deployment flexibility will be enabled with the help of end-to-end managed network slicing that dynamically optimizes the network resources within the stipulated QoS during the network slice life cycle.

Programmable networks will allow for the optimization of the overall network performance as well as the customization of device behavior targeting specific use cases. Programmable networks build on cloud technologies for faster feature development, faster bug fixing and DevOps-like operations. This will enable CSPs to adopt a software-as-a-service business model for network applications such as billing, management systems and core network functions.

Network-aware applications use knowledge of network capabilities to intelligently place application components where they will run with the greatest efficiency, performance and energy efficiency, taking into consideration factors such as where specific hardware-acceleration capabilities are available.

An architecture built on distributed network infrastructures also provides limitless connectivity – that is, connectivity that provides the required data rate and latency, and is available wherever needed. Multi-connectivity, multipoint connectivity and distributed MIMO (multiple-input, multiple-output) will be utilized to ensure robustness, performance and consistent quality, which will contribute to increased network reliability, availability and resilience (NRAR). Multipoint connectivity and distributed MIMO provide the ability to use many tightly coordinated network nodes to communicate to each specific device, which greatly enhances performance. To provide truly global coverage, the distributed network infrastructures will include non-terrestrial access components such as satellites and drone networks. As part of the longer-term radio-access evolution, spectrum utilization will extend into the sub-THz frequency bands. These very high frequencies create the possibility of highly accurate sensing for use cases such as JCAS.

Energy-efficient networks will be built by optimizing energy performance end to end. At node level, hardware accelerators and related software components such as operating systems and drivers are essential for low-energy solutions. They will provide scalable traffic processing including features like multiple sleep states and fast activation. Further, smart network and AI-based features for site energy orchestration will be embedded for network energy optimization.

In the 6G time frame, the network infrastructure will be capable of providing more than 100 times higher capacity in comparison with today’s network infrastructure, with peak rates over 100Gbps, wide-area consistency latency below 5ms and centimeter-level position accuracy. The always-available, high-performance network infrastructure will offer truly global coverage in indoor, outdoor and non-terrestrial environments. This is made possible through interoperable service interfaces that interconnect all ecosystem actors on a global scale.

Embedding trustworthiness

The openness of the network infrastructure requires security assurance based on zero-trust principles – never trust and always verify – to secure all communication while constantly monitoring the trustworthiness of the communication. A secure, resilient and privacy-preserving network infrastructure is a cornerstone in supporting all society-, mission- and business-critical use cases for humans and intelligent machines in a verifiable way. This requires seamless integration of AI-based automated security controls and assurance mechanisms to make it possible to respond in real time to both known and unknown threats. This will secure trustworthy network infrastructure that performs tasks reliably, maintains data provenance, confidentiality and privacy, as well as protecting against unauthorized access or manipulation.

In a trustworthy system, secure identities are needed at all layers of the system – infrastructure, connectivity, devices and the edge – as well as in network-slicing functions. This can be enabled by means of root-of-trust mechanisms for identities. To ensure both security and privacy, the network infrastructure will use state-of-the-art protocol stacks including quantum-proof algorithms and confidential computing capabilities to protect data when it is being processed and stored. Finally, the network infrastructure will support security assurance throughout the application communication path, including devices, all network domains and the application cloud. In the long term, this will include support for bystander and spatial privacy-preserving solutions for immersive user experiences.

Artificial intelligence in programmable networks

Network infrastructure must have the flexibility to handle a large variety of use cases, which will be achieved through greater automation of network management. Programmability and distributed intelligence are the best ways to achieve this. Fully autonomous system operation is based on zero-touch deployment and operation across all stages of the life cycle, as well as on the ability to continuously optimize performance in real time. Network services will be managed by well-defined functional models and by dynamic and fine-grained allocation of heterogenous network resources.

Distribution of AI-based performance optimizations in the network requires a data-driven architecture and data-driven operations. These optimizations are continuously supervised and tuned to deliver explainable optimization results. The network will aggregate the optimization results and place them into a larger context, where they can be combined, evaluated and adjusted to reach a defined intent. A hierarchical control-loop structure is used to manage and optimize the network and end-to-end services. In an autonomous network, humans can take a supervisory role relying on trustworthy AI that is lawful, ethically adherent and technically robust.

Generative AI has reached commercial maturity and has been proven to provide efficiency gains in areas such as research and development. Further, this technology will support network analysis and optimization tasks such as predictive maintenance, fraud detection and network planning.