Nothing. It's politician's bullshit bingo regurgitated so many times through the media that now people on HN start to echo it.

We need 5G for self driving cars

We need 5G for industry automation

We need 5G for IoT (because that makes so much sense, lol)

We need 5G for ...

I just saw a recent article touting the benefits of 5G for smart traffic signals.... but it's not clear why those same traffic signals couldn't use 4G, or even 2G, it's not like traffic counts need a lot of data bandwidth.

Or, you know, it's not like traffic signals move around on their own, so they could just be hardwired.

There are already traffic signals out there that use radio signals to communicate with each other. They are used to manage the traffic around road construction sites. E.g.: http://www.fabema.de/en/products/traffic-light-systems/wirel...

So 5G is not needed for that purpose at all.

It is my understanding that while traffic signals are almost always hardwired to mains supply (somewhat surprising observation is that traffic cameras often are not but use Pb battery packs that are recharged during the night from street lighting power) they very often use radio interfaces for communication, be it something PACTOR-like, point-to-point 802.11 on unusual frequency bands (ie. sub-GHz) or straight 802.11.

The sub-GHz 802.11 seems to be uniquely american thing. And maybe just pushing anything over RF is somewhat uniquely Czech thing (caused by the fact, that Czech spectrum allocation has somewhat unique additional 10GHz ISM band), but in Prague there is an giant nest of smallish microwave antennas on every tenth street light pole (the extreme is probably Malostranske namesti, where there is 10GHz link between two traffic signals that spans about 70m).

IRT to applying ultra-high bandwidth networks to devices which don't need a lot of bandwidth, the point of moving to a higher bandwidth network is to allow these devices to quickly report back what they want to say and then get off the air faster. The faster they can report their small heartbeat of data, the more devices you can support. Lets assume this traffic light sends a snapshot of the intersection at a frame a second, each frame being 100KB. So, 100KB/s would be is normal load, so there's no real need for it to have a dedicated 100Mbit link. However, if you've got a few thousand devices wanting to share that link space, you're going to need more bandwidth available to the whole. Its also good for the general responsiveness of the network for that 100KB transfer to happen in less time, as only a single device can talk on the network at once. Its the same thing with home WiFi connectivity. 54Mbit would theoretically be plenty of bandwidth for a movie stream and a few phones/tablets browsing social media, but if you're using 802.11g you'll probably have issues with your video stream as those phones and tablets hog the airwaves while they load a batch of images. It takes time servicing multiple devices in the same channel, time that takes away from max effective bandwidth available.

Also, while its true one could just wire up all of these fixed devices, these wiring costs would massively drive up the cost of implementation. Trenching city streets to lay new cables costs a lot of money, just ask Google.

You're talking about 802.11g and 100Mbit links, while I was talking about cellular data -- two completely different things.

A cell tower has more than a single channel and a city with a thousand traffic lights is going to have many cells, so using existing cellular generations is not going to run into problems with collisions from 1000 traffic signals trying to send data at once.

At the basic level these two things are pretty similar. Its the same concept, just a difference in scale. The tower can only slice time so much before clients have to wait longer and longer to transmit. The faster a client can send its burst of data and get off the air, the more clients you can have connected per tower. I've personally experienced plenty of circumstances where even though signal strength is fine, there just isn't enough bandwidth for reliable cellular service.

I understand that if a single cell tower is overloaded, then its clients will have issues talking to that single cell tower.

My point is that a city with a thousand traffic lights will be covered by hundreds if not thousands of cells, so the issue of traffic lights competing for bandwidth for a single source is non existent.

And this problem doesn't change for 2G, 3G, 4G, or even 5G. Perhaps 5G can scale better but if you had a single 5G tower covering an entire city, it too would run out of capacity.

Or even use LoRa for meshing which is 50kbps over 10km. As you said, there's only so much data a traffic signal needs.

The 4g network tower can only handle so many devices connected to it. 5g also solves that problem. Speed is great but the big change is that the towers can handle many more devices (which is why it’s touted as the reason why iot/selfdrivingcars/etc can now work. If you’ve ever been to a large conference or festival, you can see what happens when too many devices try to use the network. It just loses all reliability and speed.

So the shorter distance the 5g signals go also mean more towers with more connections. My point is, don’t focus on the speed as being the huge innovation, it’s the number of devices possible. Although once it’s available, you will see apps come out that you may not have expected that take advantage of the higher speeds to do cool things.

But yes, we need to move in the direction of 5g or something similar to allow the tech you listed to take off.

5g will be so pervasive that it will be abnormal for a device or appliance to not be online. That 5G will be the aether and devices will expect to be able to access it, whether it's tied to your personal connection/identity or not.