With the advent of 5G coming, we are entering a new phase of communications infrastructure. Today the network is mainly used for moving content, e.g. videos, web data, emails, and audio. However, neither are sensors widely connected to a ubiquitous network, nor can we remote control real and/or virtual objects. In the latter case we require the installation of the "Tactile Internet", i.e. a network which allows for 1-10ms end-to-end response latency. Once this is achieved, humans, machines, as well as robots will be able to interact in real-time with virtual as well as real objects, enabling applications spanning the economies of health & care, manufacturing, mobility, edutainment, events, and energy. Each new economic opportunity on its own is a market of similar value as cellular is today. However, each opportunity comes with differing boundary conditions on the Tactile Internet, as we have learnt from our project fastZwanzig20 and the 5G Lab Germany.
Many applications require that the system operates with a minimum downtime, measured e.g. as outage. For manufacturing this can be as low as 10^-8, which is in stark contrast to well-designed 4G networks that deliver an outage of 3% at the cost of a latency generating HARQ. We can show with first theoretical results that new concepts allow for low latency reliable communications.
To design the Tactile Internet, today's cellular networks with a best-in-class response latency of 25ms must be redesigned from a physical layer up to the application layer. Due to the slow speed of light of 200km/ms over fiber, central web servers must be augmented by a distributed Mobile Edge Cloud, with new server architectures, operating systems, and more.
In this talk some basic ideas will be addressed which have been developed at TU Dresden, showing solutions from silicon hardware architecture, signal processing and modulation, to edge server design.