News: Microelectronics
19 November 2020
Imec’s Advanced RF program to develop scalable, energy-efficient 6G device technology
At the Japan installment of its ITF 2020 Technology Forum, nanoelectronics research centre imec of Leuven, Belgium has launched the Advanced RF pre-competitive research program, which aims to lay the groundwork for next-generation mobile communications, focusing on creating a roadmap beyond 5G device technology.
One of the paths that imec will explore includes the use of a hybrid III-V/CMOS approach to energy-efficiently and cost-effectively accommodate the very high bandwidths enabled by radio frequencies above 100GHz. Focusing on concrete proof points, imec and its partners aim to present prototypes that show this approach’s potential. Imec says that, to realize the program’s targets, it will tap into its broad expertise in the connectivity space – combining its system, circuit and network knowledge with its semiconductor know-how, both in the digital and analog domains.
With bandwidth demands doubling every two years and existing radio spectrum bands getting clogged up, the telecoms industry is looking continuously for novel technologies and approaches to anticipate future mobile communication requirements. Three generations of cellular networks have been introduced since the early 2000s – starting off with the roll-out of 3G, the commercial introduction of 4G in 2009, and today’s deployment of 5G networks. Still, demand for bandwidth remains insatiable.
Single-link data rates of 100Gb/s, µs latency and new, bandwidth-hungry apps
“There is still some discussion on 6G’s exact characteristics and performance specs, as the actual standardization effort has yet to start,” says Michael Peeters, vice president of imec’s R&D activities in the connectivity domain. “What is clear, though, is that next-generation wireless networks will largely outsmart their predecessors. Projected features include a 100Gbit/s single-link throughput, micro-second (µs) latency and a significantly higher energy efficiency – less than 1 nanojoule per bit.”
“This will be crucial to enable concepts such as federated learning between artificial intelligence (AI)-enabled autonomous systems such as self-driving cars,” he believes. “Other use-cases include the deployment of very high-speed, ultra-reliable mobile hot spots in dense urban centers, the support of immersive augmented reality (AR) applications, and holography.”
Tapping into higher radio frequencies in an energy and cost-efficient way
The quest for more bandwidth is inextricably linked with the use of higher radio frequencies: the higher the frequency, the more bandwidth that is available. Hence, while the 5G mobile networks that are on the verge of being deployed will initially operate in the 28GHz and 39GHz bands, future mobile networks will resort to frequencies above 100GHz to cater to the world’s growing bandwidth demands. A similar shift to higher frequencies is happening outside of the cellular space as well.
“That is where we come across one of the main challenges to making affordable solutions a reality – as today’s standard silicon device technology lacks the required transmit power and energy efficiency at higher radio frequencies,” says Nadine Collaert, program director for imec’s analog/RF activities.
“III-V materials – such as indium phosphide (InP) – might offer a way out, but do not yet lend themselves to be integrated onto a silicon platform. Hence, we will specifically look into hybrid III-V/CMOS approaches,” she adds. “We will investigate how III-V materials can heterogeneously be combined with CMOS technology, how those materials perform in terms of reliability, which degradation mechanisms are at work, etc. Building on those insights, we aim to create mobile device technology that efficiently and cost-effectively operates at 100GHz and beyond.”
The networks that will use these mobile links will likely also need new approaches to the protocol stack from the physical to network layer. The default choice nowadays for high-bitrate air interfaces, orthogonal frequency-domain multiplexing (OFDM), may no longer be the right choice. More directional beams at higher frequencies can make much more aggressive use of spatial multiplexing. Meshing and even stronger separation of data and control may become needed. Here, we will leverage the expertise we have built, e.g. on mmWave channel modelling, dense and short-distance meshing, and time-sensitive networking.
A prototype that shows benefits of hybrid III-V/CMOS approach
“It is widely recognized that future mobile networks will have to operate at frequencies above 100GHz to deliver data rates of 100Gb/s and more. So far, however, a clear and mature pathway to developing the underlying technology has been lacking,” says imec’s CEO Luc Van den hove. “With our Advanced RF program, we aim to extend our impact beyond our pioneering role in the semiconductor space – providing insight and technology across the connectivity ecosystem, and not only when it comes to enabling 6G but also to accommodate the next generation of Wi-Fi communications.”
Over the past few years, imec researchers have already conducted research into this topic. One example includes their work in the Electronics Components and Systems for European Leadership joint undertaking (ECSEL JU) ‘Taranto’ project, resulting in the development of an integrated, compact 140GHz radio module that enables single-link data rates up to 80Gb/s in an energy-efficient way.
