South Korea's biggest telco says 5G has failed to deliver | Hacker News Reader

South Korea's biggest telco says 5G has failed to deliver

theregister.com

63 pointsjnord41 comments

I think a big problem with the 5G rollout was the confusing mish-mash of low, medium and high band "versions" of 5G, each of which had different marketing names depending on the carrier (not to mention "4G+" that was marketed as 5G).

All the marketing speak just said "5G is superfast and will let you download a movie in 20 seconds!" or whatever, when that reality was that only applied to mmWave 5G, and oh by-the-way mmWave signals are blocked by a window. It was just ridiculously over hyped.

I do not think multiple bands are the problem, they vary and are meant to vary across mobile network deployments by design.

The biggest promise 5G has failed to deliver on so far is the capacity. We were promised 5G cells of being able to handle "huge" numbers of subscribers in the high density areas, and that has not been the case.

In the middle of a typical business day, my 5G MiFi device has pre-4G downlink speeds, and the office is located next to a cell tower. Speed improves in the evening once people leave the CBD and hovers slightly under 1 Gbit/sec as it should.

So my understanding is that because telcos have had to pump vasts of money into the core network upgrade, which was effectively a complete core network revamp rather than upgrade, after seeing an eye-watering bill, the telcos have decided to underinvest into the cell tower capacity to break even.

So, yes, the low latency and high downlink speeds are there, but only when the cell has the capacity to provide the bandwidth.

> the authors recall being sold as an enabler of autonomous driving, unmanned aerial vehicles (UAM), extended reality (XR) and digital twins

I only ever saw 5G sold as a solution to cell towers being overwhelmed by dense crowds (e.g. at sporting events, music festivals, etc.). Has it succeeded at that?

At my local stadium, which holds something like 40-50k people I have no issues with mobile throughput. However the whole stadium as of 4 years ago is now full of small cells with maybe a dozen or so in each "section" so I'd say that's a far bigger factor than 5G just by itself.

Many, small, cells _is_ 5G :)

My phone works great at sports events with tens of thousands of people in an outdoor environment. 4ish years ago it did not. This isn’t with Wi-Fi trickery, simply more capable networks!

For sports stadiums, it's normally solved by installing ~10 cell towers inside the stadium itself to cope with the load.

Cell networks are getting better at installing small mobile cells now, and some even have 'cell-in-a-truck' designs that can be driven to a busy beach or other place that is temporarily busy to give people service where otherwise the network would be overloaded.

My 5G still breaks down at every large event I go to, or even sometimes at a busy train station.

I'm assuming it's more to do with how it's deployed than the technology itself, but purely as an end-consumer, it feels like a dud.

From the article:

> For example, Level 4 full autonomous driving will require 6G.

My understanding of Level 4 is that it's basically full autonomy, but with geo-fencing. Surely that can be achieved with GPS and there is no need for any 2 way comms? (this is assuming that all compute is "on board").

I feel like driving a car from "the cloud" is just asking for trouble, surely no one's planning on doing that right? Maybe a-sync updates like where the geofence should be, or new software would be fine but not something realtime (which I assume is where they're leading with this 6G business).

Could you imagine the crash reports? "Sorry you plowed into the pack of school children, RDS was failing over at the time."

After the Bambu malfunction, I'm imagining a thousand self driving cars backing through the garage door in the middle of the night and driving to a location creating massive traffic jams for everyone to wake up to

No, the main reason geofencing exists on modern autonomous cars is to assure connectivity. A ton of work is offloaded. Not immediate driving decisions, but classification problems and similar are offloaded. Object detection is local, route decisions and object classification is offloaded. Further, there's fairly limited cache of "landmarks", the route significantly relies on things like street signs and other roadside identifiable objects to identify it's position; without a database of those it doesn't know it's current road position very well. Lane markings turn out to be too ambiguous for these AI cars to use. It has a bunch of those cached, but street signs go stale very quickly, and two-way communication is used to keep these dbs up to date.

> No, the main reason geofencing exists on modern autonomous cars is to assure connectivity. A ton of work is offloaded.

Which means that they're not autonomous. Autonomous means being able to operate on its own.

"will require" dose not mean "has a technical prerequisite of"; it simply means no Level 4 car will ever be available that dose not include always online tracking.

I think there are regulatory requirements on connection as well.

There's no strong benefit to consumers, and the industrial use cases never materialised because telcos have not rolled out services like URLLC, MMTC, compliant edge-compute (MEC) platforms with full 5GC integration (e.g., MEC app registration, auto routing of UE traffic to MEC services, etc.)

The telcos I deal with don't even have a business model or a way to price enterprise services delivering these new 5G features, they're just selling a pipe with newer tech and asking for more money.

There are a bunch of companies, research organisations, and universities building and making demos but hardly any of these make it to production. Here's an example of a working system: Local 5G and TETRA interconnected at Narita International Airport: https://www.ntt-east.co.jp/en/release/detail/20230519_01.htm...

I work in this space... 5G coverage is a lot worse in most places than 4G or even 3G. As for the overhyped applications like remote surgery, remote tele-operated robots, 5G-driven cars, etc., don't hold your breath.

mmWave 5G is a dud - it has some promise but again telcos are not investing money to roll out mmWave. Telcos aggressively marketed 5G to consumers at a premium price point but they offer very limited benefits in limited situations. Even routine functions like cell to cell handoffs are often smoother in 4G than with 5G, not because of the spec but because of other factors.

5G was overhyped and it failed to deliver any value except in a few situations.

I have had a 5G capable phone for a few years, but have chosen low cost carrier plans which are LTE/4G only.

Not gonna pay a premium for 5G.

If it wasn't a premium service/no additional cost I am sure adoption would be higher in Canada.

Canada cell service always sounds so atrocious. AFAIK there isn't a price differential between LTE and 5G on the major US carriers.

It is really bad here in Canada. There is very little innovation going on here and the goal is to fleece the customers as much as possible.

I have a 5g phone. 99% of the time it only makes a 4g connection. There is one place in my house where it can get 5g, and I don't notice the difference when it does.

It makes more 5g connections when I'm outside, but I don't tend to need high bandwidth connections there.

For those who keep screwing this up in the sibling comments: 5G is LTE.

LTE is a protocol managed by the 3GPP Working Group. LTE is a forever protocol that continues to add new technologies and enhancements to existing ones while maintaining that single technology. LTE has gone through many versions whose names are absolutely useless, such as "LTE Advanced" or "LTE Advanced Pro", which do not align with any major addition to the LTE spec; new versions of the LTE spec may not explicitly contain the name, such as "5G NR", which is a source of confusion for some.

Also, 5G is not, automatically, mmWave, nor are 5G handsets required to implement it, and many don't.

5G is LTE, except... it is designed to allow IOT devices in your home to send data back to the manufacturer without your consent.

Nope, again, not a 5G feature.

In 3GPP Release 13 (the release that did LTE Advanced Pro, the version that adds high QAM complexity along with high MIMO complexity to set the stage for the next version of LTE, now known as NR; LTE Advanced Pro is often referred to as 4.5 or 4.9G) introduces two new variants of the LTE family, NB-IoT and LTE-M

NB-IoT defines what is essentially an "LTE Lite" protocol for low computation devices, similar to, say, the relation of BT and BLE, where BLE uses the BT PHY in a low power/complexity state with a simplified protocol. This protocol is meant for dozens of kbps in half duplex operation.

LTE-M focuses on low bandwidth, but higher than NB-IoT, up to 10mbps down, 5mbps up. It, otherwise, uses existing LTE mostly as is, just reducing the LTE PHY to the lowest complexity allowable by the spec in an effort to reduce power usage. This is similar to how you can buy 802.11ac WiFi SoCs that are limited to 2.4ghz 1x1 20mhz (often seen in standalone WiFi Direct stuff).

NB-IoT is useful not for "spying on you", but for deploying commercial agricultural and industrial sensors using existing LTE networks instead of deploying expensive purpose built networks (such as DASH7, HART). LTE-M would be "useful", but unlikely to be deployed inside of consumer devices for the purposes of spying due to the cost.

Neither are an appropriate replacement for Zigbee or Z-Wave or Thread or any of those home automation protocols; and devices that use those can, and probably do, spy on you already.

Well yeah. LTE was already at the theoretical limit of information transmitted per hertz. What were people expecting?

For everything to become incredibly amazing. 5G was more pumped than AI, especially in Europe. For some reason people kept mentioning robo-surgery as a use case...

There were two parts that were hyped: Lower latency, and mmWave. I think the self-driving cars and telesurgery kind of hype came from the lower latency part. Can't have a callcenter staff remote-drive a car that is stuck in some weird situation with 600ms ping.

Back a few years ago I was working for a telco. There was a huge expectation that 5G would enable bringing workloads closer to the "edge", so deploying on-prem servers and avoiding the round trip to a centralized location. However, as far as I know this hasn't happened much due to the massive amount of rearchitecting companies would have to do. Very few applications actually benefit from such a low latency. I'm not sure Telcos even made it happen on their end, I left the industry before the full rollout.

I guess they're saying they, as a consumer, are disappointed? This is a confusing article, when they speak to UAV, Self driving vehicles etc, are they talking about industry consumers like Hyundai? (SK owns Uber in Korea). I don't think the average person on the street here in Korea cares that much about autonomous systems.

As a consumer of SK 5G, I'm certainly not disappointed. I recently visited other countries and I missed Korean internet so much, SK 5G is great!

Edit: reading the white-paper, I suppose they're talking about the average person on the street... Seems, odd.

"Even at the time when preparing for 5G, services such as autonomous driving, UAM, XR, hologram, and digital twin had appeared and expected, but most of them did not live up to expectations. We should have taken a more objective perspective. For example, whether 5G technology alone could change the future, or whether the overall environment constituting the service was prepared together. If so, the gap between the public's expectations for 5G and the reality would not have been large. 3D video, UHD streaming, AR/VR, autonomous driving, remote surgery, etc. are representative services that are not still successful presented by the 5G Vision Recommendation. Most of them are the result of a combination of factors such as device SK Telecom 6G White Paper form factor constraints, immaturity of device and service technology, low or absent market demand, and policy/regulation issues, rather than a single factor of the lack of 5G performance. Therefore, to successfully settle innovative services in the 6G era, all participants in the ecosystem should work together to prepare the environment along with 6G technology."

Preparing to overhype 6G by undermining 5G which does as much as LTE in practice at a higher power profile.

> at a higher power profile

Does that matter though, when the article reports that the paper authors concede that

> Despite these challenges, 5G has delivered some wins. SK Telecom's paper pointed to a 70 percent reduction in data cost per gigabyte compared to LTE. Customers on 5G therefore use 50 percent more data than those tied to the previous generation standard.

I was mostly referring to/reminding of the hype preceding the 5G transition, not critiquing the article. In the US I haven't witnessed much wins for the consumer but reasonable to assume it has reduced cost on the carrier front.

I'd say that 5G has finally delivered on the promises of 4G - roughly gigabit speeds. I pretty consistently get around 400-600 Mbit speeds, with occasionally hitting 1200 Mbps. This is without mmWave which is not launched on iPhones outside of the US.

I've noticed that 5G doesn't work as well as LTE when moving, like when on a train or a car on the highway. Has anyone else noticed this?

I've noticed longer transition times between towers, but while connected I can pull down a lot more data so the impact on streaming is fairly minimal for me so long as my "buffer" doesn't run out too soon.

What I’ve noticed is overall link reliability going down on my carrier in the las couple years as more 5G is rolled out. Areas where I had decent reception are often worse. I suspect it is more enshittification rather than 5G per se.

Yes, I think I have noticed this as well.

Ahahaha, only if they didn't trick clueless laypersons into buying much worse 5G plans by claiming that your shiny new phone needs 5G (no it doesn't, it has to fall back to 4G even or 3G networks when it's not available!).

Seems 5G is much more useful right now for factories and industries. Seen demo of autonomous ports and factories using 5G.

We also need the killer app for 5G to be useful for the end user.

That makes sense, because industrial applications and increased load-handling by the the telecoms were the entire purpose for 5G. There was never any realistic benefit to the end user over 4G aside from the promise of the system being able to handle more subscribers.

There will not be a consumer-level killer app for 5G. It wasn't really meant for us.

That's what they said before Netflix and Youtube on smartphone

mmwave was supposed to give more bandwidth in 5g but it was not widely implemented by SK'n telecoms which i am told also has non-technical reasons and now they are complaining that it failed.

Reply on news.ycombinator.com