Ericsson teams with MIT on hardware design for cognitive networks

Ericsson is exploring cognitive networks to reduce operational complexity and increase energy efficiency, with two major research projects in a new collaboration announced with MIT.

The partners aren’t disclosing the amount of funding or length of the project, but an Ericsson spokesperson said work is already underway. The projects have a long-term technology research focus for the design of advanced hardware that could help power future 5G and 6G networks.

Head of Ericsson Research Magnus Frodigh said in the announcement that 5G is paving the way to fully-realized Internet of Things, citing “massive amounts of tiny IoT devices and AI-driven cognitive networks” as “two drivers of the next leap forward.”

“Working with the brilliant teams at MIT, we hope to develop the hardware that will make that possible,” Frodigh stated.

A key focus is lithionics-based device research to enable neuromorphic computing. A December 2020 ZDNet article described neuromorphic computing, a model that’s inspired by the brain and uses less energy for artificial intelligence processing.

Cognitive networks that use AI for highly automated, secure and data-driven operations are of interest to Ericsson as the Swedish vendor looks to reduce complexity for operators that need to manage large mobile network environments with increasing levels of features. 

Specifically, Ericsson Research and the MIT Materials Research Laboratory are teaming up to investigate new designs in lithionic chips that would enable neuromorphic computing “offering exponentially more energy efficient AI processing” than currently available and that could lead to fully cognitive networks.

The scope of the research includes understanding both core semiconductor technologies, as well as AI approaches, so the potential result shouldn’t be seen as one specific circuit, an Ericsson spokesperson explained.

The project takes a holistic approach for mobile network research and lithionics isn’t limited to a specific subsystem and domain of the network.

A second focus is energy-harvesting technologies, primarily applied to cost-effectively powering low-power IoT devices supported by the mobile network.

Anantha P. Chandrakasan, Dean of MIT’s School of Engineering, said in the announcement that the pair are tackling key technical challenges as energy-efficient devices continue to advance.

“By combining our knowledge with Ericsson’s expertise in mobile technology, we aim to develop hardware that will power exciting new AI applications on the edge, and take significant strides in the next generation of mobile networks,” Chandrakasan stated.

Ericsson will work with the MIT Research Laboratory of Electronics to potentially show how devices, like massive amounts of sensors or other “zero-energy” devices, could tap directly into energy from received radio signals to connect to the network. The duo also will look at systems design for how this low power could be leveraged for simple tasks.

“Besides from device hardware technology our research will include how a mobile network (RAN and Core) may be designed to connect and control these devices,” the Ericsson spokesperson said via email.

It’s a long-term effort, and he said realistic realization of results is probably going to be in the 6G horizon timeframe. 5G systems are still in fairly early stages of deployment, and while some early 6G efforts are underway, 2030 has been cited as a rough timeline for initial 6G commercialization.  

“However, results can still apply for improving current and evolved 5G systems and are not necessarily bound to a particular architecture generation,” the spokesperson added.