Five network trends – Towards the 6G era – Ericsson

Trend #1: Digital representation for the networked reality

With 5G, we already enable physical and digital worlds to converge into an augmented reality that serves communication needs for humans and machines. As humans and physical objects are only able to experience the physical world in a local context, the local presence of sensors, actuators and networks is a crucial enabler. The convergence of physical and digital worlds is enabled by digital representations of both humans and physical objects as well as their environments.

Data from embedded sensors and actuators enables the digital observability of the environment and physical assets. The future network will provide low-level processing of billions of different data streams from sensors and prepare them for applications. The preparation will handle aggregation, filtering and fusion of the data streams. Processes are monitored in real time, and actuators will enable autonomous operations. To further improve observability, the network will generate sensory data such as identity, positioning, time stamps and spatial mapping information. While 5G enables the basic functions, 6G will provide enhancements across these domains. The following functions and capabilities are the most crucial in the development of the future network platform.

Network-aware rendering and synchronization

Some use cases require upper-bound guaranteed end-to-end latency. As the network is context aware, from the capabilities of spatial mapping and dynamic object handling, rendering algorithms provides the dynamic latency and control required to optimize the quality of the user experience. The optimization includes synchronization between all sensory modalities and digital objects in the physical environment.

Collaborative contextual awareness and observability

Connected intelligent machines rely on contextual awareness and observability to relate to, and cooperate with, other machines nearby. This includes the registration and tracking of physical trajectories, intents and capabilities, for example. The network will serve intelligent machines through a context fabric of contextual information including real-time processing of such contextual information. The context fabric thus provides validated information to the collaborative machines about their personas and capabilities, as well as contextual data.

Interoperability among machines and devices

Diversity among connected intelligent machines and devices leads to heterogeneity in terms of semantic descriptions and information formats, and mediation and interoperability are thus required between them. The optimization of communication between machines and devices calls for syntactical and semantic interoperability, as well as programmability between different protocols and models. The network communication stack will provide support for this, thereby off-loading the issue of interoperability.

Real-time positioning

Physical and logical positions as well as time observations will be essential for relating the digital representations to the physical world, thereby generating digital context awareness. Such context-aware data needs to be presented in a uniform and standardized format to enable usage across platforms and to operate and manage systems of systems. This also enables marketplaces for continuously updated observations, models, lessons learned, insights and other digital representations in real time.

A key network feature is the coordination of safe, collaborative autonomous operations by real-time positioning of physical objects in a dynamic environment. The network also provides secure identification and authentication of objects and their related data for integrity and privacy. This enables the secure control and actuation of physical objects, taking into consideration regulations, policies and roles. These network abilities will be key enablers in the foundation of a trustworthy cyber-physical world.

Spatial mapping

Spatial maps are created by collecting and fusing sensor data from all networkconnected devices – such as cameras, lidars, radars, gyroscopes, accelerometers, level sensors and pressure sensors. These networkgenerated maps are used to correctly position the digital objects to the physical environment, thereby enabling multiuser interaction. This network feature will handle relevant spatial mapping processing to reduce form factors, power consumption and cost on IoS devices. Spatial maps will initially focus on visual representation and will, over time, be extended to spatial sound, touch, smell and taste.

Dynamic object handling

Based on spatial maps, the network will perform dynamic object handling and real-time tracking to merge the physical environment and digital content. Dynamic object handling will be particularly important for advanced occlusion, un-occlusion and the continuous update of moving objects in high-resolution spatial maps. For example, in a fire-rescue operation, un-occlusion of smells and temperature could be used to localize a trapped human. Other features include the ability to collect spatial data to generate personalized user experiences and enable context-aware communication.

Embedded data processing and enrichment

A fundamental capability of the future network is a lightweight pervasive dataprocessing fabric providing the right data pipeline and compute characteristics. As the connected intelligent machines, devices and sensors will generate massive data and information streams, the network will provide adequate processing for the preparation of data, metadata extraction and annotation. Further, in-network stream processing and enrichment such as event detection, filtering, inferencing and learning, as well as sensor fusion, will be provided and exposed. This in-network stream processing will be supported by specialized hardware acceleration embedded in the network. Data formats and new compression algorithms will be optimized for the need of machines rather than for human consumption and processing. The necessary transcoding and compression algorithms will be embedded in the network.

In addition to the network functions and capabilities described above, the realization of digital representation for the networked reality at scale will also require end-to-end solutions across the digital infrastructure of devices, edges, networks and clouds. As digital representations can include sensitive information, standardized, interoperable and secure use of collaborative spatial maps are prerequisites to build trust, making them a good example of an area that will benefit from ecosystem collaboration and innovation. We will promote openness and close collaboration within the global ecosystem to form the future business platform for innovation.

Technological advances in four key areas are essential for the future network to support the convergence of the physical and digital worlds, namely: limitless connectivity, trustworthy systems, cognitive networks and the network compute fabric (trends 2-5).