Hacker News Reader

None of this is really what this service is intended for.

If you read the whitepaper, they list examples of what this would be useful for; namely, covering your own space with your own 5G for your own devices.

Deploy it in these areas instead of WiFi:

1. A stadium's remote ad/video/informational displays 2. A logistics distribution hub's stock-tracking robots/systems/handhelds 3. A corporate campus's smart displays or door access systems 4. Oil and gas drilling/processing sites's systems monitoring in remote, non-covered areas

So this isn't about creating a new provider in your local city, but rather about connecting your devices in your space in cases where WiFi is insufficient or overloaded.

Is there any reason it _couldn't_ become that?

Probably not, except cost. It would be too expensive to compete with the national providers and then you would still need to interface with them for any service outside of the city. It's possible, just not the intended use case and not a logical use.

I remember people making very similar claims about AWS EC2 in 2007.

Ah, so the mythical beast of future Amazon 5G use cases can be anything I want then, right? How about unicorns?

I see what you’re saying, but just because a similar statement was made about another service, does not mean the outcome will be the same in this instance with vastly different dynamics at play.

Wifi adapters are cheap and everywhere. For 5G tot surpass wifi it will have to match that. Laptops with 5G are not that common yet. ESP like controllers with 5G I haven't seen yet. One of the advantages is 5G operating outside of the ISM bands. That helps a lot with stability and preventing interference. It also means this is not for individuals in their home. If licensing is made so accessible that people at home can use it you will again get overlapping frequency use and increasing interference. If it not made this accessible I don't see the availability of cheap clients solved because the market will be limited.

It's fairly early but within 10 years, every device will have 5G

I got the impression this is Amazon providing Airave-type "femotcells" that emit a low-powered cellular signal and use your wired network for calls and data service. Seems to also imply Amazon would be providing SIM cards.

I'm not sure of the advantage of this over Wi-Fi, though, except to get devices that have no other option other than cellular connectivity to be forced to go over your own network.

> Cellular technology such as 4G/LTE and 5G augments existing networks with higher bandwidth, lower latency, and reliable long-range coverage to an increasing number of devices. With AWS Private 5G, you can build private cellular networks to take advantage of the technology benefits of 5G while maintaining the security and granular application and device controls of a private network.

That's from the Amazon website. Why not just deploy Wi-Fi?

Large-scale, long range Amazon-managed 5G deployment should be easier than large-scale WiFi deployment, at least if the promises are accurate. Especially outdoors where deploying a large number of base stations is challenging.

Lower frequencies mean improved building material penetration in environments like warehouses and docks, meaning fewer base stations are necessary.

Generally 5G also supports higher density than WiFi for use cases like hotels and conventions, where too many devices on too few channels will quickly destroy WiFi functionality.

WiFi is pretty crappy for many-device midrange solutions (let's say a dockyard, a 1km long facility with 1000 employees).

Commercial deployments on 5G are difficult because it's fairly new . Wifi is very tried and tested. It'll take some time to get practical experience but once done, it'll be easier than Wifi

> Why not just deploy Wi-Fi?

Everyone has a potential wifi access point in their pocket. The only people who have the ability to run their own cellular networks are either a) respectful of frequency allocations or b) know a thing or two about SDR. Interfering with 5G systems is difficult to do on accident and has a much higher barrier to entry to do intentionally.

And this is important for one of the use cases listed in the PDF: sports and entertainment venues. When you have a large venue, you will always have people who are using their phones as wifi hotspots (many of them having simply forgotten to turn them off), and this eats into valuable airtime. You can have your APs send deauth packets to rogue networks, but you could still end up not having enough bandwidth for your own purposes. I'm reminded of an article I read about the unveiling of the first iPhone[0]:

> The software in the iPhone’s Wi-Fi radio was so unstable that Grignon and his team had to extend the phones’ antennas by connecting them to wires running offstage so the wireless signal wouldn’t have to travel as far. And audience members had to be prevented from getting on the frequency being used. “Even if the base station’s ID was hidden” — that is, not showing up when laptops scanned for Wi-Fi signals — “you had 5,000 nerds in the audience,” Grignon says. “They would have figured out how to hack into the signal.” The solution, he says, was to tweak the AirPort software so that it seemed to be operating in Japan instead of the United States. Japanese Wi-Fi uses some frequencies that are not permitted in the U.S.

While wifi has come a long way since 2007, if you need a reliable, high bandwidth system in an environment with lots of interference, this is a good choice.

[0]: https://www.nytimes.com/2013/10/06/magazine/and-then-steve-s...

Edit: this comment has some more good reasons for why you would use this over wifi: https://news.ycombinator.com/item?id=29398181

I think the whole point of the service is "convenience". If deploying private 5g is as convenient as deploying Wi-Fi, why not just deploy private 5g?

We all like Wi-Fi because we can attach a Wi-Fi router to an internet link and it "just works". Similarly, if you can make a mobile network "just work", you are getting the same functionality plus the benefits you outlined.

This may not seem very impressive to someone who has no need to support large Wi-Fi deployments, like a typical home or office wireless network. These are not the cases this service is targeting anyway.

The target customers are enterprises that are managing large-scale fully-automated sites like warehouses and factories that need to support 10s-100s of thousands of IOT devices which may include some time-critical systems that need ultra-low-latency reliable connections to a backend. Today these enterprises have to rent a 4g network from a telecom provider to enable these sites which takes months of planning and is expensive to operate and extend.

This is where the "convenience" becomes a game-changer. Using a private-5g service, an enterprise can deploy new large-scale networks in days, manage the networks like just another cloud resource, and extend the network at will. I presume the pricing model for the service will also be much more affordable where instead of paying for each end-user (IOT device), you'd likely be only charged for your data usage on pay-as-you-go basis.

5G will displace WiFi. 5G has better authentication, handshake etc

So will 5G actually be better than 5GHz WiFi? I'm not too concerned about speed but rather a stable connection.

So here's the thing, currently I have 2.4GHz and 5GHz WiFi in my home. Any device that is stationary connects to the 5GHz one and any device that moves around connects to the 2.4GHz one. Why? Because initially I had everything on the 5GHz network, but noticed if I move around with my phone/tablets/laptops (even as simple as standing still and turning around), the connection would become unstable/choppy and introduce some lag and in rare cases a drop in connection. The 2.4GHz one works no matter what I do in my home. I can also get 2.GHz all the way out to the curb and it still works, where the 5GHz one drops significantly once I'm outside my house. In some cases it will still show the 5GHz one being connected but any app that tries to use it just hangs and never loads whatever it tries to fetch. So the 5GHz version of WiFi has been utterly useless to me, thus I'm only using it for devices that I know have decent line of sight, are within certain distance and aren't moving around. I currently have no Bluetooth devices chatting in the same airspace and I've tried it with single network too (only 2.4GHz enabled or only 5GHz enabled) - still the same. It could be that there is interference from neighbors: I can see quite a few WiFi networks of neighbors, most have both frequencies active - but I also had the same issue before I live in this specific house. So if 5G can improve the situation, that will be pretty great.

5GHz range is less than 2.4GHz. So that's the reason

Huh?

This doesn't seem to use AT&T at all, and amazon is actually bringing their own hardware too.

This isn't MVNO-aaS. Its Antenae-aaS.

Ah, you're right. I am mistaken for AWS Wavelength.

Hate to be that guy, but...

AWS Wavelength is Ec2 instances running on-prem on existing telco locations.

Yes I dont think OP's description is an accurate usage of Private 5G. Where it is aiming at industrial ( warehouse ) or cooperate usage within certain location ( cooperate HQ ).

Not sure how Tesla or Uber would get their own private 5G.

>This isn't MVNO-aaS. Its Antenae-aaS.

I think a lot of people are missing this point. This will not allow someone to become their own carrier. It allows someone to install their own "cell towers" and have devices connect to them without having to use a 3rd party carrier.

Btw, MVNO-aaS does exist (for 5G, too): https://en.wikipedia.org/wiki/Mobile_virtual_network_enabler

Yeah, while you might theoretically be able to build a national 5G network using this, it would be way more expensive vs just building it yourself. Amazon wouldn't be marketing this if they didn't expect to have a healthy profit margin, after all.

This seems more applicable in cases where you need 5G but there is no/insufficient 5G already, when it would be prohibitively expensive to go to existing carriers (mobile data rates in the US are ridiculous), or when the threat model necessitates a private network.

AWS wants to sell their storage. Margins are there + it's a new exciting market

Like installing a PBX or email server for internal communications?

Please excuse my vast ignorance but wouldn't Tesla, Uber or whomever need to deploy a massive network of 5g towers for that? Or is more that Tesla/Uber/etc could much more easily become an MVNO-like-provider because of network slicing?

The second comment. The telecoms via 5G have been positioning themselves to rent their infrastructure on-demand. The AWS partnership is this but Amazon then re-extending it to developers on their platform as part of their "Cloud" offerings.

I doubt Amazon has many towers of their own here and are almost entirely through one of the telecoms.

But this doesn't use telco towers? Amazon is literally selling antenae.

Isn’t this service nothing related to using existing infrastructure?

It looks like Amazon is sending you base stations to set up a private network in one location.

>I doubt Amazon has many towers of their own

Their offering page specifically says that AWS "delivers your network hardware (small cell radio base-station and servers) Attach power and internet connectivity to smart cells and servers"

They are not using telco partnerships here, or at least if they are, it's not on the level of carving out a chunk of the telco network for private use.

I struggle to see how this remains a "cloud" offering rather than a hardware rental.

I think this is more like long haul wifi for your corporate campus, factory, port, university etc. Maybe even for things like parking meters (san francisco's parking meters are famously being updated/upgraded because they're ending EDGE network used by the modems in the parking meters)

at this point these are 2 sides of telecom have nearly evolved to completely separate industries

one services towers on the ground. base stations. maintenance. physical network deployment

the other is the services of running the network. roaming agreements. customer management. internet breakout

tesla and other vehicle makers may see value owning the service side but very unlikely they want to maintain a network

For those at home reading this... MVNO = "Mobile Virtual Network Operator". Aka a network that doesn't have its own wireless infrastructure.

https://en.wikipedia.org/wiki/Mobile_virtual_network_operato...

>Uber offering free fast Internet to their drivers

Amazon is talking about installing actual local hardware infrastructure here though. It seems like that only makes sense where there is no existing 5G, otherwise it's probably just cheaper to use the telco's infrastructure since, as you said, they could work directly with the telco to get their own slice.

Sort of like the difference in price between a dedicated hosted server (the AWS 5G) and a VPS (a slice of the telco's 5G)

I'm sure I'm misunderstanding some aspect of this whole thing though.

It's actually cheaper if you already have access to cheap broadband such as fibre internet where the GB can be as little as 10 cent/gb. But if you're reselling the telco bandwidth then it doesn't make sense

> This could be extremely useful for security. Maybe even the death of VPNs.

You can already get this - if you think you need it, you should google 'Private APN'. It's been available for years, assuming you're a corporate user looking for a few hundred SIM cards.

> This could be extremely useful for security. Maybe even the death of VPNs.

You can already get plans from the existing cellular providers to drop you onto a private secure network that behaves like a VPN though... that's common for people who need secure OOB access to their network gear: but install routers with 3g/4g/LTE expansion cards, get the SIMs on one of these plans, and voila -- OOB remote network access that isn't exposing your devices to the internet

You can take this a step further and make the telecom as a service entirely decentralized: Earn "tokens" by providing wireless area coverage & data backhaul, spend "tokens" to transmit data.

This is exactly what the Helium (HNT) project is doing. They started with LoRa coverage (super low datarate but long range for IOT) and are moving into other protocols such as WiFi and CBRS 5G (same as this offering) via the FreedomFi project.

> 5G isn't really like 4G upgraded. It's more its own thing.

This is my understanding as well, but I have no idea about any of the details. I know there's something cool about "beams". Do you (or anyone) have a good "entry-level" article/doc that outlines some of the major features that makes it so different than 4G?

Search Massive MIMO, Beamforming ( Which isn't really new ), Small Cells, NR.

That is about it really. You can ignore mmWave which is pure hype. Most of the other enhancements are on the carrier / operation side and not consumer.

You can also ignore all the 5G Self Driving Vehicle crap.

>5G Self Driving Vehicle crap.

I live in a dense suburb of a large city and 5G coverage seems not that great. I'm not sure why self-driving requires or gains much benefit from 5g anyway, but I wouldn't want to rely on it. You can certainly do car-to-car communication & coordination without it, and you wouldn't want a minor network outage turning the system into chaos.

Because certain vendor from certain country have a huge patent portfolio on Autonomous vehicle (AV) and wanted it to rely on 5G so that everything is centralised and could easily be controlled in the name of traffic shaping (cough). And somehow (cough) EU was brought into the whole thing until a Saint appeared, after being banished to hell he finally came back and convinced enough people AV with 5G would need to total state surveillance. I think, if I remember correctly EU finally abandoned that idea. But as with everything it may come back someday. ( Hope not )

Any reason for the obfuscation?

Those who knows it would instantly understand. Those who dont could fact check it themselves.

>Those who dont could fact check it themselves.

That's me, and no they can't. At least not without searching the full breadth of possible players looking for a set of facts that matches. If I could figure out who you meant, I wouldn't have commented.

Seems unlikely many firms would want to scale out the hardware of their own network - rented from Amazon no less- instead.

Actually telecom don't have their own infra. They lease the infra from tower companies, here telcos will get hit as well as the tower lease companies

About the payment, we're working on a model to compensate the users. Email me if you're interested

> This could be extremely useful for security. Maybe even the death of VPNs.

If that's true, I wonder if this affects Tailscale's business and how.

It's not new. You can get private address space that's routed into your network, but if you trust telco network security, you probably don't have anything worth protecting (or run a VPN over the top and only use the private APN for persistent IP addressing).

For example, AT&T's private APN service has been around since the 3G days: https://www.business.att.com/products/private-mobile-connect...

This requires a physical cell tower and 5G receiver so it’s not as flexible as running a VPN client on your laptop.

Short-term, things like "Tesla Free Network" could exist for their self-driving cars. Or, Uber offering free fast Internet to their drivers or a truly private device.

So like WhisperNet, except 5G, and anyone can make their own?