presentation

We have read the paper "6G Takes Shape."

The goal of the paper is to provide a realistic view on what the 6G standard will likely look like (and not look like) in the early 2030s, and to discuss some of the main driving forces behind the development of 6G.

The authors have chosen to explore their goal by asking 20 questions, divided into four subgroups. We will present these questions and summarize what the authors of the paper have concluded.

The authors remind us that the paper is only speculative, but also believe that some of the outcomes are inevitable.

The paper has three authors, two of whom are doing research in wireless communications at the University of Texas. The third author is from the industry and is the Vice President of Engineering with Qualcomm Technologies.

The next part of the paper is called "6G Physical Layer" and tries to answer these questions:

Historically with every new generation, ever new "G", the physical layer has been the most changing part, often braking backward compatibility and needing new hardware.

The authors predict that 6G wont have these abrupt changes. Focusing more on optimizing the hardware and techniques already used in 5G.

which leads us straight into the first question:

The authors concludes that OFDM is still going to be the main workhorse of 6G. OFDM is extremely efficient thanks to the FFT and there is really no reason for replacing it.

This ties to the next question. And the answer is no. The codes and modulations already used is near optimal in a information theory sense and are very robust. Neural networks would not handle this job better.

And also neural networks are not power efficient at all.

However the authors still believe that Machine Learning will have an important role in other parts of 6G. Mainly where classic model-based approaches struggle. The mention tasks like:

This part was the most technical and mentions a lot of abbreviation for a lot of different coding and modulations. But the main points here is that 6G hardware is going to have to be backwards compatible with 5G, so a lot of the coding and modulation of 5G is still going to be used.

They do mention that 6G might adopt something called Probabilistic Amplitude Shaping (PAS).

The authors argue that 6G is going to adopt full duplex. The main drivers of the need for full duplex here is that the traffic, mainly on the Uplink, is growing (due things like Extended Reality (XR).

The paper predicts that cell free MIMO is he way forward, at least in dense urban areas.