Читать книгу Shaping Future 6G Networks - Группа авторов - Страница 35

A digital twin is a digital replica of an object, generally characterized by a very high level of fidelity that makes it possible to use the digital version as a reliable representation of the behavior and characteristics of the real object [3]. The concept of digital twins has risen to the forefront of the discussion on product life cycle management thanks to improvements in the design and capabilities of sensors (e.g. video cameras, laser scanners, and lidars) and sensor fusion algorithms, which now allow a rich and faithful representation of the real object, as well as thanks to advances in computation capabilities, which enable the real‐time manipulation and editing of the digital twin. Moreover, the concept of digital twin is often associated to VR and AR, as the digital representation can be visualized through any immersive visualization technique. Thanks to these properties, digital twins can improve the design, engineering, inspection, and maintenance of complex machines and devices. For example, a machine could be remotely inspected without the need for personnel on the ground, without any loss of realism, and with an improved (digital) access to components that would be hard to reach physically. Similarly, mechanics can monitor the performance and status of different components of a vehicle with a high‐fidelity representation without the need for the car to be in the repair center. Additionally, for product development, a digital twin would allow different teams to work on the same product, exploiting a 3D, shared visualization in various remote locations, and can enable advanced simulations of the product behavior.

In these scenarios, the role of the network is to provide a high‐throughput, low‐latency bit pipe to connect the sensors on the physical product with the computing platforms on which the digital twin is hosted. Several elements contribute to the need for ultrahigh throughput, which would not be supported by 5G technologies, as for the teleportation use case of Section 2.3.2. A digital twin will be generated by a large number of data sources, which need to be distributed around the physical device, and capture different properties, not only the visual aspect. Moreover, the twinning rate, i.e. the rate at which the physical and digital representations are synchronized, could be in the order of hundreds of Hertz, for applications that require a real‐time tracking of the evolution of the physical object. Therefore, the data rate required by digital twin use cases can be in the order of tens of gigabits per second, with the need for high‐capacity links between the different components of a digital twin system (sensors to database, and database to representation). Similarly, when real‐time interaction and control of the physical object through its digital counterpart is required, the latency should be in the sub‐millisecond range. However, if real‐time control is not of interest, or a lower level of fidelity can be accepted, digital twinning applications can tolerate higher latencies and lower throughput. Therefore, 6G networks should also focus on adaptability and openness to the applications, with open interfaces to enable cooperation between the wireless stack and the higher layers, for example, to optimize the number of sources and the twinning rate according to the capabilities of the network, or, vice versa, to allocate more resources according to the needs of the application.