A certain telecoms provider is advertising in the UK by showing off that you can operate air traffic control over it's home broadband service.

Is this that surprising?

I didn't think that air traffic control for commercial flights was that bandwidth hungry or relied on an ultra stable connection?

What bandwidth (in kb, mb, gb etc. per second), and how many nines of availability are required?

  • 5
    $\begingroup$ I think 'ultra stable' is required, you want air traffic control to be immediately available when needed. When there is an issue onboard, the airplane has to land immediately, and the runway needs to be cleared, traffic control cannot be offline for a few minutes. $\endgroup$
    – ROIMaison
    Mar 7, 2022 at 11:55
  • $\begingroup$ The entire Chinese railway system (one of the largest, fastest and most busy) run on GSM for two decades before switching to 4G and NB-IoT. Here I'd trust wireless much more than any landline. While each node and switch along the line could fail, the cable itself isn't 100% reliable either. Wireless has much fewer points of failure. $\endgroup$ Mar 7, 2022 at 13:54
  • 3
    $\begingroup$ There 31 million seconds in a year. Five-9s (99.999% the industry typical value) reliability would be about 5 minutes of downtime per year. A 5 minute outage at the exactly wrong-moment could be catastrophic. At any randomly chosen time, even a 1 minute outage could easily lead to separation loss. I would estimate that Five-9s is not enough reliability for ATC service. $\endgroup$
    – abelenky
    Mar 7, 2022 at 16:31
  • $\begingroup$ I think you’ll have to be more specific. There are a lot of different ATC roles that may be more or less suited to being deployed remotely. I could easily see clearance delivery being done that way. $\endgroup$
    – Jim
    Mar 7, 2022 at 19:27
  • 1
    $\begingroup$ @user3528438 Wireless has many more points of failure. But that isn't important. What is important are Single Points of Failures. Wireless systems can have redundancy added very easily. $\endgroup$
    – Aron
    Mar 8, 2022 at 8:00

3 Answers 3


You can, technically, but you wouldn't run ATC over a home broadband service. The problem is not the bandwidth of the broadband, but the availability (is it available when you want to use it) and continuity (the probability that keeps being available while you are using it) are too low, and the Mean Time To Repair is too high.

In terms of bandwidth for radar control, you don't need much (20 - 200 kbps for a radar, 500 kbps for radar tracker output with 250 aircraft and 1 second update rate) nor for Voice over IP (100 kbps per active channel). But if you operate a remote tower, such as in this case, you need to be able to stream a few 4K video feeds with low latency (<1 second end-to-end)(~20Mbps), which may be a bit of a challenge for some networks. In addition, in case of degradation of the bandwidth, you want to be able to prioritise voice over video for example.

Here is the making off:

Anyway, no air traffic control was done in this video. Notice the ground station call sign "air suburbia ops" and the phrase "runway 23 available for landing". If anything was actually transmitted from that living room (we don't know, we are looking at cut/edited scenes), it was purely informational, not air traffic control.


Actually very little bandwidth is required for most Aviation control related items. This is due to the substantial redundancy built into the "entire system." Each part plays a role in ensuring adequate control of the air traffic in the associated area. The FAA has over 30,000 stations around the US. They range from VOR's to weather camera's to AWOS (Automated Weather Observation Stations) sites, to regional ZAN's/ATC's. Regional ATC's require the most bandwidth but the FAA is just now getting out of the TDM world and moving to ethernet.

TDM, aka legacy circuits such as T1/DS1's, DS3's, and OC-3's, provided extremely predictable bandwidth that was completely serialized. The FAA took advantage of the serial nature of TDM and built an entire network of equipment, in excess of 150,000 unique service/location combinations, and it still works to this day. When digital bandwidth was expensive in the 1970's the FAA built most of its services to rely on 9600 bits/sec. Some things rely on only 2400 bps. With a 9600 bps requirement they could stack 5-6 services on a single 64k digital channel. This saved the FAA alot of money over the years.

What is more interesting is the detail the FAA has on each site. Of the 30,000 plus locations the FAA has on record they have a strict listing of the redundancy requirements, which generally translates into uptime requirements, for each one. In some locations a single circuit is ok. Other sites two circuits/connections are ok. In other sites two connections are required but they must be from different providers. In other sites the connections must be different providers, physically diverse from each other and terminate at different Central Offices. Going even further the FAA might require physically different transports or even three connections, all of which are physically diverse in all aspects. In the legacy TDM days the FAA would request a circuit path that would require the circuit to avoid certain central offices even when going across a small town. In some regional centers there will be multiple carriers with multiple physically diverse entrances for the remote connections.

That said the question is related to Traffic Control. The FAA needs very little bandwidth for traffic control. There are larger pipes, OC3's and OC12's, now moving to Gigabit ethernet, between the regional centers so they can share radar and traffic control data but generally speaking regional centers can operate independently, though that is far from ideal. Most of the data is not air traffic control though and even the centers migrating to gigabit ethernet are doing it for the cost savings, not the bandwidth needs. Telco's are raising their prices substantially for TDM (T1, OCx) services making it prohibitively expensive to stay on TDM Circuits. For voice traffic to the airplanes this is minimal. G.711 is ~90k when active but there are codecs that are more efficient. Most airline carriers have large networks that consist of remote SIP-to-RF radio's but most are single channel radio's so they would only need less than 100k when they are active.

That said the system has a massive amount of redundancy built into it to ensure that no one location knocks out the entire system. This is matched with the rules of flying that generally state the pilot is the one in control of the airplane and is responsible to be situationally aware at all times.

I know all of this because I work for a Telecom in Alaska that supports the FAA. Alaska has been a proving ground for the next generation equipment remote equipment for the FAA. Alaska's vast and remote distances pose significant challenges to the FAA in terms of reliability, both the equipment itself as well as the communications to that equipment, and the criticality of the equipment. In fact ADS-B was designed and tested in Alaska extensively before it was rolled out nationwide.

All said and done the bandwidth is not the issue for the FAA. Generally speaking control data is very little, be it in Aviation or controlling an oil field. The sheer number of sensors can add up, but it is nothing compared to the data a Netflix PoP pushes in a day.

  • 4
    $\begingroup$ Really nice answer! Welcome to Av.SE, Scotth. $\endgroup$
    – Ralph J
    Mar 8, 2022 at 13:43

I'll give it a shot... I don't think a bandwidth value can be provided because it's a distributed and decentralized system, so it's bandwidth on many channels, not bandwidth on a single fibre-optic cable running to someone's house.

So it's a nonsensical ad from the bandwidth perspective. ATC is mostly radars, with computers processing the radar information and transmitted data received from the airplanes (now moving to on-board GPS location transmitted to ATC by satellite for processing), for presentation and further processing by the supercomputer inside the head of the air traffic controller.

The heavy lifting in ATC is done by the unique characteristics of the brains of the people working the radar displays. The recruiting, testing and training systems of the ATC agencies is designed to sift out these brains from the general population. That's why previous education and experience doesn't count that much toward your chances of success when you try out for ATC. If you have the mental skills they're looking for, you'll make it in as long as you're free of other show-stopper defects in personality or character.

Beyond a moderate amount of bandwidth in the communication networks of the ATC computers (system development for those sorts of hardened institutional uses follows many years behind development for commercial and consumer uses, and if you went to a ATC facility the hardware is likely 10-20 years old), the big priority is hardware reliability.

You don't want the human computers monitoring the hardware computers, and making all the decisions, to lose the feed for any length of time, and to that end you will achieve reliability the same why the airplanes themselves do, with redundant network and processing hardware architecture to eliminate single-points-of-failure.

So someone in the industry would see that Ad and conclude it's claiming it has a hardened and redundant network with bulletproof reliability, not so much super-duper bandwidth or speed.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .