There's a new air-to-ground provider though: https://smartskynetworks.com/connectivity/

(I worked on this technology)

Interesting. Is it based on 4G/5G or something ad-hoc?

<goes and checks press-releases to remember which bits I can talk about...>

It's basically LTE in the 2.4 GHz band, with a load of extra bits for beam forming, coping with speed and distance etc. Using the 2.4 GHz band allows them to build a network covering the US without spending billions on spectrum.

Yes, it's the same 2.4GHz ISM band that wifi uses, with the same power limits. And yes, it works well enough that you can do video calls hundreds of kilometres from the base station. That's how shit wifi is, and how well this stuff works.

> That's how shit wifi

Seems like you know a bit about this. Could you indulge my curiosity and elaborate why wifi is shit? I assume some combo of DSP bottleneck (i.e. want to keep hw cheap) and maintaining backwards compatibility with older standards?

Wifi is improving now. Newer standards like Wifi 6 are doing that basically by taking the good ideas from LTE (4G) and applying them to Wifi. The cellular radio community has been much better about evolving standards, coming up with new technology and rolling it out. Maybe they have it easier because everyone gets a new phone every two years, but they've still had fairly good backwards compatibility support.

Multiple Access - this is the general term for 'multiple devices on the same spectrum'. Cellular divides up the spectrum into frequency bands (Frequency division multiplexing) and time slots in those bands (time division multiplexing). The base station then assigns each device a slot in a band (or potentially several), it also provides the time reference for this scheduling. This makes efficient use of your available spectrum and nothing interferes with or talks over anything else.

Wifi gets none of this. There's no base station planning so two access points can interfere with each other. There's no FDMA (access point runs on one channel of one band), and no TDMA. There's no uplink channel and downlink channel separation. Instead it's "Carrier sense with collision detection". Basically "try not to transmit at the same time as someone else, and if you do come back and try again later". This results in nodes competing, talking over each other, interfering (see "hidden node problem") etc. and makes inefficient use of the spectrum. The problem gets worse as the traffic gets busier. Wifi 6 is bringing in some actual scheduling.

Modulation schemes (the fundamental way that you transmit bits on a radio) - cellular modulation schemes are now incredibly efficient and robust. They use a variety of techniques that allow scaling based on channel conditions so that you can get as close as possible to whatever Shannon currently says you're allows. A lot of wifi uses old-school spread spectrum, and newer standards use OFDM which is a bit better. Wifi 6 is finally moving to the Quadrature Amplitude Modulation stuff that LTE uses.

Sometimes I work in a factory (full of metal everything) that's a metal plated building with no signal repeaters inside, basically a faraday cage. U can get enormous signal boost by just opening doors trough the wall. Most places inside the building report low signal with no usable internet connectivity, but some special spots provide enough signal to get few KB/s trough both ways. I was always impressed by the amount of work needed to push few KBs of data to cell tower trough such conditions, given a phone is able to establish and hold a connection. With WiFi there would be no hopes of even holding a connection between two points inside the building 30+ meters apart.

That's an unfair comparison to cellular though as cellular has its own licenced spectrum with nobody else able to interfere, high transmit power limits and potentially a friendlier frequency for the sort of waveguiding you're talking about.

That's why I find the smartsky comparison so interesting - because it's a completely level playing field.

Thanks for this information! Do mesh wifi routers do FDMA now? I think my mesh router uses multiple bands

Not as far as I'm aware. I don't know much about Wifi mesh, but I'm guessing that it has two channels - one for being "master" and one for "slave".

I.e. the master channel is used to serve wifi to the nearby devices, and to the next wifi routers in a chain, whereas the slave channel is used to get wifi from any upstream routers.

We wouldn't describe that approach as FDMA - more that it has a separate channel for backhaul.

Wifi's use case was, and still largely is, to be a wireless LAN. This means few users, no mobility, static devices, short distances, devices plugged in (not so true now). It started pretty much as exactly that: Just transmit ethernet frames over the radio and everything else is exactly as wired ethernet.

Cellular standards cater for a very different use case and, as a result, as vastly more complex and robust. To manage a lot of moving users and to support mobility is hard, also taking into account that cellular has had to deal with quality of service from the beginning because even if data rate was low voice requires a constant throughput with constrained delays.

This kind of shocks me honestly. Maybe it is because wifi is shit as you say but I wouldn't have thought hundreds of kilometers was possible, much less with one target moving.

I've been looking into a way to get a wifi link between a couple locations maybe 500ft apart and wasn't sure if even that was possible.

> That's how shit wifi is, and how well this stuff works.

That's not really fair. Their solution requires knowing the exact location of the base station and the client at all times. It's significantly more complex and expensive than Wifi.

Obviously it's more complicated and expensive - the radio hardware is massively more capable.

Also, you're not entirely correct about how the system works.

Thanks. That sounds like the standard approach.

They are air to geo stationary, and that’s a long way. 50,000 km instead of Starlink’s 300 km, iirc.

The operational height is around 550km they're just launched to ~300km then boost and spread themselves out.

There‘s a new hybrid system in Europe called EAN, based on a specialized LTE network and an S-band satellite to fill in gaps in coverage (mostly over the oceans).

I recently got to use it, and as expected latency is significantly better than on a pure GEO system.

Gogo is also working on an air-to-ground 4G/5G network covering the continental US.

The thing is that once Starlink has the service working, and they have, it does not require a lot of extra investment to support ships and aircrafts.

On the other hand, an air-to-ground system requires a full cellular network on the ground (at the very least with dedicated RF and antennas), although with fewer sites because cells can be massive, and that still only covers land.

> it does not require a lot of extra investment to support ships and aircrafts.

Are you sure about that? I have a friend who worked in airspace comms his whole career (including at Gogo) and he described some of the challenges of the engineering problem (these systems have to be airgapped from the critical systems on the plane, the certification processes, the antennas have to be super highly directional to talk with satelites high overhead while both the plane and the satellite are moving at tremendous speed, etc). Maybe your "not a lot of extra investment" is relative, but it seems like there's still a bunch of work to make these things work on planes.

Yes, of course it is relative.

Developing new terminals is orders or magnitude cheaper than developing and deploying a cellular network.

True, but complex terminals increase the unit price and ongoing cost (higher weight and surface area both increase fuel consumption).

EAN supposedly works with only 300 base stations – in terms of terrestrial networks, that‘s nothing.

That's fine, I wasn't sure if you were making a relative point or if you were implying that equipping planes was trivial in some absolute sense.

I flew across the Pacific on Alaska Airlines last month and they specifically said they'd only have Internet until we left the continental US. Most of the trip had no Internet, just the in-flight movies that you can stream over their wifi.

Depending on the satellite provider, they may still be using satellite and still only work over the continental US. Even starlink basically bounces the signal off a satellite and down to a ground station; this doesn't work over the ocean. They just started doing satellite to satellite links to get "back to land".

I stand corrected, edited above.

If this has been working correctly for many years, how is it that we still lose planes? :(

Isn't such connection bulletproof for the trajectory of the plane, even if simply passively tracking?

Several changes came in post-MH370 https://en.wikipedia.org/wiki/Malaysia_Airlines_Flight_370#A...

Not all plans have these systems installed.