Categorizing the 130 projects recently discovered in 6G discussions as expected to require AI/machine learning, we expect the AI landscape to look like this:
Across the top, the diagram shows the categories of AI project (please see previous blog post) – described as“not really AI” through to having “6G specific needs”.
Network Use Cases Network management use cases include discussions about practical, money saving uses in areas such as power management – to more complex capabilities such as signal processing. There are then projects in areas which are more “6G concepts” and discussions tend to be very general around the use of AI to power some popular 6G “talking points” such as increasingly distributed computing.
Going down the diagram, there will be a need for AI as part of cross-network co-ordination (eg satellite) and also a variety of security capabilities which will use AI to combat increasing use of AI by bad actors.
RAN Use Cases There are quite a number of new algorithmic techniques for the radio environment – some related to MIMO and some for theoretical technologies not developed yet such as Reconfigurable Intelligent Surfaces.
Customer-focused Use Cases This box captures AI use cases which relate to services for verticals. There are a variety of ideas captured here under “new capabilities”, for example, the improvement of slicing and use of improved locational accuracy. And then a box for offering AI “in a box” to customers, possibly at the edge.
Use-Case Specific The last box should have a range of options for adding AI into particular telco products for customers, however, not much was seen in these early stages beyond current hot topics such as drone management.
