6G Networks: Casting their shadows before (Part 2)
In the first installment of this series, we looked at the path that technology will take to reach 6G mobile networks, as well as some potential use cases for 6G networks. Now, we turn our attention to the technology, efficiency and sustainability changes we can expect once 6G becomes a reality.
New network technologies – Starting today
In its primer 6G Explained, Nokia lists six network technology areas characteristic of 6G. More importantly, we need not wait for 6G to be a reality before deploying these technologies — especially since 5G-Advanced (3GPP Rel-18) is much closer. The six technology areas include:
- Artificial intelligence and machine learning (AI/ML). What we know today as self-organized networks (SON) are a forerunner of the enabling technology in 6G. SON technology is perfectly suitable to be used for 5G and 5G-Advanced. Atos has immediately available telecom 5G/ML offerings that reach from application space all the way down to networks and to OSS and BSS.
Nokia considers the "dynamic AI/ML-defined air interface" a key component of future 6G networks, giving radios the ability to learn from each other and from their environments. More details are available here
- New spectrum technologies. New frequencies go beyond the centimeter waves already used in today's 5G and will include sub-THz and THz bands for short-range metropolitan or indoor coverage to allow peak data rates in excess of 100 Gbps. Today, such bandwidth is only achievable with quad-lane QSFP28 optical Ethernet.
- Networks that can sense their environment, people and objects. According to the Nokia report, rich situational information allows the creation of mirrors or digital twins of the physical world, and is most useful in combination with AI/ML-based analytics.
- Extreme connectivity, extending 5G URLLC (ultra-reliable low latency communication) to sub-millisecond latencies and even higher reliability ratings. The Nokia 6G Primer asserts that these new features "will improve the experience of real-time video communications, holographic experiences or even digital twin models updated in real-time through the deployment of video sensors."
- New architectures. As of today, most network functions for 5G core and RAN (radio access network) purposes are already designed as cloud-native meshes of microservices running on Kubernetes cloud/edge platforms – a trend known as telco cloud. Both private cloud and hyperscaler services are viable for telco cloud.
6G architectures will take the cloud-native service mesh idea further, supporting heterogeneous environments — as in private, public and hybrid telco cloud.
- Security and trust. This technology area will primarily address privacy issues from the deployment of mixed-reality worlds combining physical and digital twin representations, and the danger of jamming attacks.
Jamming a private 5G network – especially when built from a dense mesh of micro-cells – in order to bring down operations in a factory or hospital campus is difficult. However, innovative darknet tooling will no doubt become available over the years, and anti-jamming provisions need to keep pace.
Figure 2: Six technologies for 6G, according to Nokia
6G and energy consumption
Energy consumption is an important factor in cellular networks. Not only is there a strong societal expectation to reduce technology’s carbon footprint, but there are also commercial incentives. The energy bill makes up a significant portion of operating expenditure.
Open RAN (Radio Access Network), for example, was rejected by many telecom providers in its early days because the energy consumption to operate suitable servers turned out to be significantly higher than the energy consumption of dedicated appliances.
The energy efficiency challenge is being intensively discussed for 6G as well, with the following aspects in consideration:
- Radio unit broadcast energy efficiency. 5G spectral efficiency is already very good, as can be seen in Figure 3 – a compilation of various Internet sources on the comparison of 5G and anticipated 6G in various performance metrics. In a blog post entitled "6G networks will be energy efficient from the get-go thanks to AI/ML" Nokia claims that applying AI/ML optimization to the native air interface can, in single-antenna scenarios, boost radio efficiency by nearly 20%. In 5G, up to 15% of broadcast energy is used for the transmission of reference signal. Similar results are expected with massive MIMO 6G broadcasts.
- Edge system energy optimization. Running RAN DU (distributed unit) software on data center servers that have never been optimized for edge usage is extremely inefficient. Atos has a true edge server product line which can also be used for many different telecom purposes. They are equipped with FPGA, GPU and other accelerator cards for the efficient operation of 5G/6G cloud-native RAN and other edge workloads.
- Central data center system energy optimization. In 5G, there is already a trend to relegate control plane core functions to cloud data centers (which have a more favorable energy balance), and to use on-premises computing facilities selectively. Whether such an "escape to cloud" is feasible for a telecom network is mostly a question of the operator's business model.
- Deploying cloud-native network function software. In all cases (including 5G already) this is highly advisable. It not only anticipates the new architectures of 6G, but can be scaled within wide ranges, so energy waste is avoided and operating costs remain in-line with workloads.
Figure 3: Comparison 6G vs. 5G – spectral efficiency, energy efficiency et al. (var. Internet sources)
In conclusion, it’s safe to say that we will see many 6G developments during the 5G era. 5G-Advanced, in particular, is almost there already. Careful open and multi-vendor network planning, ideally assisted by an independent telecom system integrator, can take decision makers a long way towards protecting their investments and avoiding sunk costs.