Original 5G use cases

I recently bumped into a World Economic Forum (WEF) slide deck from 2019 with the title 5G-Next Generation Networks Programme Repository of use cases. It is a collection and analysis of projects and concepts that were publicly promoted. In hindsight this looks like the treasure vault of 5G hype. It is also a good starting point for a retrospective. ?

The mixed bag of IoT was the most often listed use case category followed closely by mixed reality and autonomous driving. Drones and robotics followed. ?

IoT and 5G is a so far difficult combination. Yes, the number of sensors and other connected devices around us is growing just as expected, but only high bandwidth applications such as HD video cameras really need?5G. Perhaps RedCap (when eventually supported by the networks) will give us 5G connected Point of Sales terminals and thermometers. ?

Walking around with VR goggles??

Mixed reality or the various combinations of AR and VR seem to become another disappointment for the 5G enthusiasts. The good news: there are some VR headsets with 5G support on the market. The bad news is that most use cases so far do not need 5G, and all the significant?device vendors connect their gadgets?using Wifi or a cable. What went wrong? ?

Today people engaged with headsets tend to be indoor and rather stationary. Against this background it is a bit odd that forecasts such as the recent Nokia Global Network Traffic 2030 Report propose that within some years a significant portion of mobile traffic would be AR/VR. That would require camera output or video input to be carried over the network. This was the original idea.?

In the early days of 5G mixed reality was classified as one of the key areas where low network latency?is necessary and where 5G would be king. From an implementation point of view immersive services like multiplayer cloud gaming are like tightrope walking. Processing and synchronizing the various data streams without ending up with a compromised user experience (it lags)?is difficult. Add haptic communication – the doctor doing remote surgery – and it is tightrope walking in with a?heavy backpack in windy weather. The risk of failing is obvious. When the delay between real action and what is shown to the user grows beyond 10 milliseconds, some people?notice. The longer the delay is, the more likely it is that users experience motion sickness or a reduced feeling of immersion. These are strong reasons to abandon the service.?

From this starting point telecom folks concluded that the network delay could only be less than 5ms. With or without 5G that is not easy to deliver.?In?5ms?light travels?1000km in fiber. Routers and gateways in the network add delays ranging from microseconds to milliseconds and jitter that must be compensated at the receiving end. If remote surgery requires the doctor and patient to be?just a few network hops from?each other?– why bother, especially in a heavily regulated industry. I am not holding my breath waiting for the offshored surgeon to be legalized.??

As immersive services with haptic communications are so demanding, the current products work with local content and inputs. According to an article of The Verge the Apple Vision Pro video pass-through?creates a 12ms latency within the device. Adding networking with some jitter does not sound appealing. This obviously is where we are today with state-of-the-art technology. ?

If you take off your goggles for a moment and look at what is going on at sports stadiums you see?fans now using their phones for checking?game statistics, slow motion replays?and chatting instead or in addition to what unfolds in front of them. This is not exactly augmented reality as engineers defined it, but it fulfills the same needs. ?

Drivers becoming passengers??

Four years after early 5G use cases were collected it is easy to say that autonomous driving will hardly become a major 5G revenue generators any day soon. The 5G engineer might say this is because it is a?Ultra-Reliable Low Latency Communications (URLLC)?use case and the necessary capabilities have so far been missing from the deployed networks. True, but Telsla provides what they call Full Self-Driving Capabilities as a subscription service (obviously with a lot of legal constraints). This feature does not seem to have any connectivity dependencies. Hard to see why the need for URLLC would be added later. Today, instead of seeing a?car driving on its own you are more likely to bump into a friendly delivery robot?that uses the pedestrian lane.

These devices mostly connect using 4G. I guess broadband?data is mostly needed for those corner cases where the poor robot gets into trouble. In the Finnish winter this happens sometimes, and even small accidents have been reported.?I am so happy it is still delivery robots and not cars.?

While autonomous driving is undisputedly making some progress, other traffic related 5G use cases seem to have been left lying at the roadside. There are many convincing YouTube videos on truck platooning, i.e. forming convoys?of trucks with a driver sitting in the first one and automation taking care of the rest. I guess this will be food for building 6G hype. The same applies to intersection assistants and other smart traffic concepts that would increase traffic throughput at current bottlenecks. What makes these theoretically brilliant systems?difficult to implement is the need to have all cars supporting them. Considering that in the EU the average car in use is 12 years old, getting to that point would take an awful lot of time as retrofitting automation to thousands of different car models is not practical. ?

Brainless robots??

The experiences with 5G smart traffic are quite sobering for the robotics use cases as well. The original 5G enabled demo robots had much of their brains working in a nearby cloud and were essentially relying on automated remote control. Commercial products are more autonomous and work fine even without URLLC. AI chips have come a long way and processing video and other inputs locally in the robot instead of sending it to a cloud service is now mainstream. Consequently, a?robot can now use a?regular 5G subscription or 4G, like the Starship delivery robots or your AI enabled mobile phone.? ?

What went wrong??

In the mobile networking history 5G was the next generation that was supposed to deliver something new. 2G gave us a digital and better scalable mobile telephony plus the possibility for text messages – a feature that went viral. 3G brought some data services to mobile devices and 4G was about using the Internet with your mobile phone. 5G specifications provide enormous flexibility. Speeds comparable to fiber access or latencies comparable?with high speed trading networks, unlicensed 5G or a continuation of the 4G service with just higher download speeds and more capacity. ?

Now we know that 5G delivered capacity – to the extent that fixed wireless broadband has become mainstream. With Standalone 5G it looks like network slicing would slowly become a real service. Many of the other options are still in the labs or on the drawing board, but the world outside of the 5G community is not standing still. ?

In the WEF slide deck many of the 5G use cases were expected to grow to an annual size of tens of billions of dollars. For augmented and virtual reality 815 billion USD were expected in?2025. More recently Fortune Business Insights forecasted the market to grow from 25b USD in 2023 to 166b USD by 2030. Cloud robotics was expected to grow to 170b USD in 2025. GMI Insights states that the cloud robotics market was 5B USD in 2023. And so on.?

As the inflated numbers did not come from one source but a wide variety of fairly credible actors, you can ask yourself what the big collective mistake was.?While the topic would be worth at least a doctoral thesis, my candidates for the two biggest contributors are:?

• A general overestimation of 5G and 5G feature availability: People involved in 3GPP or vendor roadmaps had in principle all the information needed for understanding that many of the hyped 5G promises would not come in products based on 3GPP release 15 (fully completed in 2019) but only in?later releases. But then again, I do not think too many bothered to study what exactly the topics like Vehicle-to-Everything Communications Phase 2 of the 3GPP release 15 consists of - or thought about the need for additional phases – or the possibility that the whole topic may lead nowhere. Instead of the future being now (2019 in case of the WEF slide set) four to five years were needed for chipsets to mature, networks to be built and the obvious use cases to be covered. ?
• Prudent product managers and development teams across industries continued to build their products to fit into as big markets as possible. 5G data speeds or low latencies have not been the norm so far. That is a?good reason why there is a strong preference to process inputs needed for autonomous operation or user aid locally - without going to the cloud. Another aspect is that local processing may well be cheaper than a mobile broadband subscription. ??

Obviously, we also had a pandemic and years of economic uncertainty. These have most likely reduced the pace of many futuristic businesses that expect people to buy costly new gadgets. ??

See you at the 6G summit??

Now that all industry events that used to be 5G this and that have either become Next generation or 6G this and that we see the next attempts for a building visions around the capabilities of the sub-terahertz networks that allow using the network as a sensor. This brings a lot of new futuristic use cases. While it is obvious that that some hype is needed for securing more frequencies for mobile networking and for obtaining (public) funding for the research that keeps the IPR portfolios of the big players up-to-date – don't be fooled. What many of us thought 5G would deliver immediately is now unfolding. In many countries 5G coverage is so high that you can count on it. Even without URLLC round trip delays have become shorter. You just need patience if you want to follow the evolution in progress.

And before you join the 6G carnival – please remember that all around the world a lot of spectrum was been reserved for 5G in the 26 – 71 GHz area. With a small number of exceptions, it remains unused. I think that topic needs to successfully addressed before there is a chance for anything more ambitious, like sub-teraherz.