For CEA-Leti, 6G is not just a new generation of wireless connectivity.
Like any generational advance in technology, the 5G-6G transition will greatly improve our ability to meet key performance indicators (KPIs). We’ll have the ability to link several-orders-of-magnitude–more devices; create zero-latency, zero-energy, ultra-reliable links; perform semantic-enhanced data mining; and seamlessly share knowledge between humans and machines in support of artificial intelligence and other advanced applications. At the same time, however, achieving broad acceptance will require attention to a whole new set of KPIs, such as energy efficiency, cybersecurity, and sustainability. As academic and industrial researchers have begun tackling 6G design and engineering tasks, several preliminary directions have emerged: exploration of new spectrum horizons, targeting of higher spectral efficiency through new MIMO schemes, and improvement of the overall environmental footprint.
The worldwide 5G rollout has resulted in some local states placing stringent limits on electromagnetic field (EMF) emissions, impacting network deployment and increasing the complexity of upcoming spectrum use. Because EMF limits are calculated over the entire spectrum, the envisioned increase in traffic throughput will likely require more efficient spectral-use schemes.
Recently, distributed massive multiple-input multiple-output (D-MIMO) networks, also referred to as cell-free MIMO, have gained attention. Unlike the co-located massive-MIMO approach used in 5G, the D- MIMO scheme involves a cooperative, dense network based on a massive deployment of cost- and energy-efficient RF units, which together build a virtual large array. This new paradigm promises efficiency gains even at low frequencies; this is hardly possible for co-located MIMO because of the necessary array size. The vision of a distributed network would lead to massively digital cells with terminal-like hardware specifications, denoted as an RF-less architecture. Necessary components include digital front ends, filtering and switching functions, and power amplifiers with advanced energy-efficient features such as digital pre-distortion (DPD) and envelope tracking.
Again, a continuum for new spectrum opportunities is being investigated. Spectrum resources below 6 GHz will still be essential to support wide-range radio coverage, and there may be ways to use frequencies currently not considered, between 6 GHz and 24 GHz, to deploy massive-MIMO base stations. Development will be fueled by continued hybridizing of CMOS and III-V technologies for output power in front-end modules (FEMs).
But the real breakthrough will come from considering frequencies beyond 100 GHz. The large bandwidths available here are of particular interest with, at first, D-band at about 140 GHz, and later the 300-GHz range. There are, of course, major challenges that will call for technical solutions capable of pushing silicon’s limits. Potential avenues include, again, hybridizing silicon and III-V technologies, with a goal of achieving transistor Fmax = 1 THz, and co-integrating very small antennas at wafer scale or in-package.
Until now, no convergence on these solutions has been found, but some trends are emerging. CEA-Leti teams are considering two specific aspects. On the transceiver side, to avoid large-bandwidth, power-consuming conversion, channel bonding/aggregation has attracted attention, and 100-Gbps throughput over 16 channels has been demonstrated. On the antenna side, efforts have focused on a transmit array utilizing a single transceiver/FEM rather than more classical phase-shifter architectures that require one transceiver/FEM per tile.
Figure 1 compares the average power of a beamforming RF front end assuming 34-dBm EIRP and 100-mW RF transmitter consumption. This would allow a transmit array with 26-dBi gain to reach a 7.2-Gbps/2-GHz channel at 200 meters with a backoff of 6 dB and expected power gain of a factor of 2 to 20 for xHaul or multi-user MIMO use cases.
Another major point of concern lies in the digital domain for processing of very high-throughput data streams. Silicon CMOS node technologies are evolving in response to calls for this type of processing capacity, but they come with higher prices and greater power dissipation. While this is acceptable for backhauling, it is not suitable for edge applications, and optimization work to meet 6G expectations is expected. As with 5G networks, the use of edge computing and AI blocks is seen as a key enabler for several strategic functions: development of ever-more complex wireless networks, reduction in data traffic through edge processing, and enhancement of signal-processing algorithms for channel estimation, the beamforming process, DPD, and optimization of RF settings.
Finally, there are the emerging KPIs associated with sustainability and environmental impact. A primary concern, energy efficiency, is addressed in 5G by avoiding useless resources when capacity exceeds the momentary and local actual need. 6G promises to advance this in several ways in anticipation of the envisioned exponential traffic growth. The use of AI to manage network power consumption and load is being studied, as are the impacts on equipment life cycles, as well as novel mechanisms such as reconfigurable intelligent surfaces and new MIMOs for limiting the impact of EMF radiation on both intended and non-intended recipients.
This article was originally published on EE Times Europe.
Jean-Baptiste Doré is 5G program manager, CEA-Leti.