As 5G Focuses on Enterprise Use Cases, 6G Might Focus on Virtualized and Self-Learning Networks

Mobile and fixed network operators constantly are challenged to reduce capital investment and operating costs as a way of compensating for low revenue growth, challenged profit margins and ever-increasing bandwidth consumption by customers whose propensity to pay is sharply limited. 

The very design of future 6G networks might work to help reduce capex and opex, while incorporating much more spectrum, at very high frequencies and basing core operations on use of machine learning (a form of artificial intelligence that allows machines to learn autonomously). 

New 6G networks might rely even more extensively on virtualization than do 5G networks, featuring now-exotic ways of supporting internet of things sensors that require no batteries, a capability that would dramatically reduce IoT network operating costs. 

It is possible 6G networks will be fundamentally different from 5G in ways beyond use of spectrum, faster speeds and even lower latency. 6G networks might essentially be “cell-less,” able to harness ambient energy for devices that require no batteries and feature a virtualized radio access network. 

The “cell-less” architecture will allow end user devices to connect automatically to any available radio, on any authorized network. Harvesting of ambient energy will be especially important for internet of things devices and sensors that might not require any batteries at all to operate, reducing operating cost. 

source: IEEE

The virtualized radio access network will provide better connectivity, at possibly lower cost, as user devices can use the “best” resource presently available, on any participating network, including non-terrestrial platforms (balloons, unmanned aerial vehicles or satellites). 

Backhaul might be built into every terrestrial radio, using millimeter wave spectrum both for user-facing and backhaul connections, automatically configured. That will reduce cost of network design, planning and backhaul. 

Researchers now also say such federated networks will be based on machine learning (artificial intelligence), which will be fundamental to the way 6G networks operate. Devices will not only use AI to select a particular radio connection, but will modify behavior based on experience. 

The network architecture might be quite different from today’s “cellular” plan, in that access is “fully user centric,” allowing terminals to make autonomous network decisions about how to connect to any authorized and compatible network, without supervision from centralized controllers.

Though machine learning arguably already is used in some ways to classify and predict, in the 6G era devices might also use artificial intelligence to choose “the best” network connection “right now,” using any available resource, in an autonomous way, not dictated by centralized controllers.  

To be sure, in some ways those changes are simply extrapolations from today’s network, which increasingly is heterogeneous, able to use spectrum sharing or Wi-Fi access, using radio signal strength to determine which transmitter to connect with. 

Architecturally, the idea is that any user device connects to the radio access network, not to any specific radio, using any specific base station, say researchers Marco Giordani, Member, IEEE, Michele Polese, Member, IEEE, Marco Mezzavilla, Senior Member, IEEE, Sundeep Rangan, Fellow, IEEE, Michele Zorzi, Fellow, IEEE. 

source: IEEE

Overall, many 6G features will be designed to reduce the cost and improve the efficiency of the radio access network, especially to create “pervasive” connectivity, not just to add more bandwidth and lower latency for end users and devices.