China is serious about 5G, as demonstrated at the recent Mobile World Congress in Shanghai.
If anyone thought China’s enthusiasm in 5G was more hype than reality, the Mobile World Congress (MWC) show in Shanghai would have been a wake-up call. The Shanghai show isn’t as big as its Barcelona sibling, but it’s much more representative of mobile directions in Asia, particularly in China and Korea. Exhibitors were showing user equipment (UEs), connectivity from the big players but also from a lot of emerging small-cell providers, lots of Internet of things (IoT) use-cases, also xR (VR/AR/MR) demonstrating immersive streaming to a group for interactive gaming — all of this 5G-based.
Exhibitors and participants also seemed more mature in their questions and responses — no longer only early enthusiasts, but people who are planning the practicalities of deployment at scale. As well they should. China Mobile, together with Huawei, has already switched on the start of a 5G network and plans to be in 50 cities by the end of this year. It expects to have the beginning of a standalone (SA) 5G infrastructure — distinct from the existing 4G infrastructure — next year, and full SA support by 2025. Across Asia, mobile network operators (MNOs) are expected to invest $370B in 5G between now and 2025.
There was an amazing demo of one video experience this could enable, from China Mobile and ZTE. This was a huge 8K screen in one of the exhibit halls with what appeared to be a still picture of a flower — no big deal. Except when you looked at it carefully, you could see petals moving ever so slightly in the breeze. This was a stunning demonstration of the kind of resolution you can get with 8K and the fact that this image could be transmitted over a 5G connection.
IoT in this area is very big. Deployments mentioned in China include 500,000 sensors in one region to measure water pressure in fire hydrants and environmental conditions, 100,000 fire-detection and alarm systems around the country and 25,000 gas and water-meter monitors using NB-IoT. Wait, isn’t NB-IoT based on LTE? It’s already advancing to 5G support, with enthusiastic adoption in China. I saw a lot of chip and device manufacturers in IoT at the show. Demos weren’t that exciting (how do you make a water meter exciting, after all), but with much more focus on deploying at very large scales and all through 5G. Which, incidentally, is part of why China Mobile and others see the need to build up the 5G standalone infrastructure, to support new industries and new business models.
On automotive applications, I didn’t see much of an emphasis at this show, which was curious. However, China has demonstrated fairly clear commitment to (5G) cellular V2X versus the DSRC standard. Note to western automakers.
Intelligent Connectivity was the headline of this show. Most of the talks were in Mandarin but one, by Mats Granryd, director general of the GSMA, was in English. Here’s a quick text summary. One topic in this area that I found especially interesting was intelligent network slicing. Network slicing is about virtualizing networks to accommodate multiple different classes of need (performance, predictable latency, etc.) on one underlying physical network. Configuring for these needs and adapting to maximize utilization against changing demand is not something that can be encoded one time into tables. Just as we have found for (multiple input multiple output) MIMO support whereby AI-techniques are becoming more common for link adaptation, network-slicing will also have to leverage AI techniques.
We had a lot of meetings at the show with new players in the small cell business, also with some big MNOs. Today these companies use x86+FPGAs, building either into very large all-in-one units, or split remote radio units (RRU) and huge farms of baseband units (BBU) connected to the RRUs by fiber. These units are obviously very large and not adaptable to small-cell configurations. Also given massive MIMO demands (for example Comba supports up to 64 Tx and 64 Rx antennae), existing chip solutions, such as those from Qualcomm, are viewed as primarily useful for femtocells (up to 32 users). As a result, interest in building custom RRU+BBU solutions is growing fast.
I expect that given the great diversity of use cases for 5G — from last-mile high-bandwidth video and gaming, to massive industry 4.0 automation (and private 5G networks), to ultra-reliable low-latency needs in medical, automotive and other areas — demand will explode for new 5G-based ASICs for infrastructure and for the edge. And it’s already started in China.
Published on EETimes