3 ms·

5G doesn't scale: no mobile tech scales. It works well when it's new, and in remote areas where it is not congested, but whatever speed you can pull out of it d

by XorNot 4y ago

5G doesn't scale: no mobile tech scales. It works well when it's new, and in remote areas where it is not congested, but whatever speed you can pull out of it doesn't work in cities when people use it as a primary internet connection. The laws of physics have not changed and yet every new mobile iteration cycle someone wants to insist they have. Starlink is more akin to fixed wifi, which is essentially a point to point link depending on line of sight. That's the one possible game changer here: it's low enough, and the other end is somewhere you're unlikely to be cut off from. But I suspect even Starlink, at any reasonable satellite density, wouldn't be able to support a city's worth of internet connections - and any reasonable effort to do so would be plain better served by just installing terrestial cables.

hylaride 4y ago
Starlink has been very clear that it won't scale for most urban areas.
TaylorAlexander 4y ago
I remember 12 years ago, when my phone first got 4G, it was blazing fast. I was getting 50Mbps download speeds. Eventually it slowed down as more people got on the network. Fast forward to today and my phone has 5G. It’s blazing fast. I’m getting 50Mbps download speeds!
scarmig 4y ago
> The laws of physics have not changed and yet every new mobile iteration cycle someone wants to insist they have. What is the actual physical limit for the network bandwidth for the available frequency bands? Have we reached it in congested cities?
- Veliladon 4y ago
  There's no real physical limit right now because we have a LOT of mmWave frequency up in the 26, 28, 39, and 41GHz bands. The problem is that the higher the frequency the less propagation you get, the more often you need a base station, the more the network will cost. The problem on the low band is that you can eventually you get to a point where the cell is overloaded and you're at capacity but still you can throw in another cell and/or split the sectors up. Which, again, costs money. The problem is essentially economic, not technical.
cma 4y ago
> The laws of physics have not changed and yet every new mobile iteration cycle someone wants to insist they have. MIMO gets past Nyquist and directional antennas can be added in finer granularity as more users come online (they do this for home LTE, not offering until adding more directional transmitters).
- nayuki 4y ago
  The relevant underlying theory here is not by Nyquist, but by Shannon: data rate (bits per second) = bandwidth (hertz) × log2(1 + signal power / noise power). https://en.wikipedia.org/wiki/Channel_capacity#Example_application https://en.wikipedia.org/wiki/Channel_capacity#Example_appli...
- XorNot 4y ago
  Sure: but the point is that whatever tricks you can pull, you're just slightly bumping the capacity of what is fundamentally a shared channel. We already have ubiquitous wireless internet: it's wifi. And it gets faster every year, provided we deploy a base station every 10m with fiver backhaul (aka your regular home internet connection).
  - cma 4y ago
    Directional can scale enormously vs omni directional. And MIMO can scale to a ton of antennas, especially in something like home internet where device size isn't a limit.
    - XorNot 4y ago
      Do you think it's practical to put dozens of antennas on every residential property? That's the point: with enough effort you can make any sort of wireless link pretty arbitrarily large and more "wire-like". At an extremely disproportionate increase in cost of equipment and sensitivity of antennas to positioning. But now you're competing with the installation costs of fixed cabling anyway - we're not talking about small portable devices any more which a consumer puts "wherever" in their house. We're talking large fixed installations which need to be aimed and calibrated correctly, and consume substantial power (the Starlink rectenna for example is not a low power device). The moment you've got skilled labor having to come out to the property you've lost: it's just as easy for the skilled labor to install fixed line cabling, as well as less disruptive to the home owner and it is very likely a lot cheaper. Our cities are well setup to bring things into people's homes - we have water, gas, electricity lines that need to go everywhere. The cost is not in the device, it's in the labor.

by XorNot 4y ago

hylaride 4y ago

Starlink has been very clear that it won't scale for most urban areas.

TaylorAlexander 4y ago

I remember 12 years ago, when my phone first got 4G, it was blazing fast. I was getting 50Mbps download speeds. Eventually it slowed down as more people got on the network. Fast forward to today and my phone has 5G. It’s blazing fast. I’m getting 50Mbps download speeds!

scarmig 4y ago

> The laws of physics have not changed and yet every new mobile iteration cycle someone wants to insist they have. What is the actual physical limit for the network bandwidth for the available frequency bands? Have we reached it in congested cities?

Veliladon 4y ago
There's no real physical limit right now because we have a LOT of mmWave frequency up in the 26, 28, 39, and 41GHz bands. The problem is that the higher the frequency the less propagation you get, the more often you need a base station, the more the network will cost. The problem on the low band is that you can eventually you get to a point where the cell is overloaded and you're at capacity but still you can throw in another cell and/or split the sectors up. Which, again, costs money. The problem is essentially economic, not technical.

cma 4y ago

> The laws of physics have not changed and yet every new mobile iteration cycle someone wants to insist they have. MIMO gets past Nyquist and directional antennas can be added in finer granularity as more users come online (they do this for home LTE, not offering until adding more directional transmitters).

nayuki 4y ago
The relevant underlying theory here is not by Nyquist, but by Shannon: data rate (bits per second) = bandwidth (hertz) × log2(1 + signal power / noise power). https://en.wikipedia.org/wiki/Channel_capacity#Example_application https://en.wikipedia.org/wiki/Channel_capacity#Example_appli...
XorNot 4y ago
Sure: but the point is that whatever tricks you can pull, you're just slightly bumping the capacity of what is fundamentally a shared channel. We already have ubiquitous wireless internet: it's wifi. And it gets faster every year, provided we deploy a base station every 10m with fiver backhaul (aka your regular home internet connection).
- cma 4y ago
  Directional can scale enormously vs omni directional. And MIMO can scale to a ton of antennas, especially in something like home internet where device size isn't a limit.
  - XorNot 4y ago
    Do you think it's practical to put dozens of antennas on every residential property? That's the point: with enough effort you can make any sort of wireless link pretty arbitrarily large and more "wire-like". At an extremely disproportionate increase in cost of equipment and sensitivity of antennas to positioning. But now you're competing with the installation costs of fixed cabling anyway - we're not talking about small portable devices any more which a consumer puts "wherever" in their house. We're talking large fixed installations which need to be aimed and calibrated correctly, and consume substantial power (the Starlink rectenna for example is not a low power device). The moment you've got skilled labor having to come out to the property you've lost: it's just as easy for the skilled labor to install fixed line cabling, as well as less disruptive to the home owner and it is very likely a lot cheaper. Our cities are well setup to bring things into people's homes - we have water, gas, electricity lines that need to go everywhere. The cost is not in the device, it's in the labor.