Inflight Internet

By Bjorn Fehrm

06 April 2016, ©. Leeham Co., Hamburg, Germany: On the second day of the Aircraft Interior eXpo, AIX 2016, we checked out the In-flight Internet situation. What interested us was why is there such lousy Internet performance on many airlines today and what will be done about it over the next few years. We got a good picture on why things are as they are and how this will change within quite a short time.

Rather than go through what a lot of vendors will do, we will try to describe the root problems involved and what is done by the industry to change that. We will also mention some of the key players that are attacking the subject.

But let’s start with the problem. Figure 1 shows where the world’s main airline routes are right now.

World trafic map

Figure 1. World air traffic taken from FlightRadar24 at 23.00 CET on 05.04.2016. Source: FR24. Click to see better.

As can be seen, the world air traffic is not very evenly distributed. One third of the world has intense traffic and two-thirds not so much or none at all.

The picture is a world-wide overview. Should we zoom in the major airports and large airline hubs we would see these are very congested with aircraft and that it quickly thinned out outside of these.

Serving In-flight Internet

Now imagine that 25% of the aircraft had on average 150 of their passengers that wanted to be on the Internet during their waking hours. The bandwidth that needs to be provided to the aircraft depends very much on what activity will be supported by the in-flight WiFi network, Figure 2.

Activity bandwidth

Figure 2. Bandwidth needed for different passenger activities on Internet. Source: Leeham Co.

It’s clear that watching Video is 200 times more demanding on bandwidth to the aircraft than writing Emails or Chatting, 40 times more demanding than normal Web surfing and 10 times more demanding than listening to Spotify or loading up photos on Dropbox.

So the actual bandwidth that the aircraft is needing is much dependent on what activity works during flight and one can understand if Video would not be supported by an airline system. It would destroy the experience for all other activities. The passengers should use the aircraft’s In-Flight Entertainment (IFE) system for video content until systems are available that can support that several of the passengers want to watch their Internet-based favorite video services.

Internet activity by the passenger is also dependent on the time of day. Activity is high during 07.00 to midnight and goes low during the body’s sleeping hours. The In-flight Internet demand follows the night shadow in Figure 1 with a delayed start on the western border by about six hours.

Serving the bandwidth demand

The first generation systems which are out there now almost universally build on using Satellites that were sent up for other purposes, like general communication or SatTV services. Free capacity on these satellites is now rented for In-Flight Internet by the providers. Typically these Satellites have a broad beam covering large areas (in the picture one for each major Geo) with a relatively low bandwidth, perhaps 2-4 Mbits/s per beam, Figure 3 (your DSL service is that fast per household).

3 beam BB coverage

Figure 3. Principal picture of present day Satellites covering large areas. Source: FR24 and Leeham Co.

The beams typically have areas that they do not cover very well and they have no means to increase coverage at congested areas like around large airports or their hubs for an airline with In-flight Internet.  In these hot-spots, it gets crowded and when a lot of aircraft demand bandwidth from the same Satellite beam the system jams; the wait symbol goes on in the aircraft.

Increased capacity

There are principally four ways to increase capacity to the aircraft:

  1. Increase the bandwidth of the Satellite beam. Today’s beam were built for text-based communication, the first bar in Figure 2. New Satellites optimized for In-flight Internet will have usable beam bandwidths 10-20 times higher than today’s (the companies advertise more but the practical end to end systems performance will be around 20-40 Mbits/s per beam).
  2. Get more beams so that fewer aircraft are present in each beam. In Figure 4, more beams have been added. Note how mini beams now reinforce hot areas and that micro-beams cover hot-spots like around major cities and airports.
  3. Have steerable beams so that the beam pattern can follow the change of demand which is caused by the wake hours sweeping over the globe each day.
  4. Get each beam to split into two beams by playing with beam characteristics like polarization so that the antenna and receiver can receive the beam as two or have steerable.
3 beam + spot + microspot BB coverage

Figure 4. Multibeam Satellites with mini beams (blue) and micro beams (yellow). Source: FR24 and Leeham Co.

The dividing up of the beams is similar to how cell-based mobile phone systems divides areas into mini, micro and even pico cells (shopping malls or city centers). It’s all to get a manageable number of clients in the cell/beam so that the systems can handle all clients with an adequate quality of service.

When will the better Internet be there?

All the vendors I talked to (Panasonic, Thales, GEE, GoGo, ViaSat, Inmarsat….) agreed that it will not take long to improve. Things will brighten up considerably come 2018 and by 2020 the type of Satellite systems shown in Figure 4 will be on-line.

There is much more to In-flight Internet than what is described here. This is just to explain why the experience as a passenger can be under-whelming today and that things will get better. The rest of the story I let to be told by specialist like Runway Girl Network.

15 Comments on “Inflight Internet

  1. Having an IT background and looking onto the usual race tracks of aircrafts i wondered quite a few times why people don’t use some kind WLAN (of course adapted for the high speeds and longer distances, but you can reach large distances with sophisticated antennas and boosters even with plain standard WLAN) and use some kind of meshed network to use some kind. Aircrafts in the vincinity of ground stations provide the downlink and route the traffic and aircrafts not in the vincinity route their traffic via other aircrafts to the aircafts in the vincinity of a ground station. So the aircrafts create their own network to provide in-flight internet.

    • Shhhhhh!
      Please be quite until the satellite companies have sold their fancy stuff.

    • Sounds like an ‘internet of the sky’. But I think they would have to use microwave frequencies to be viable and that may have a viable antenna problem. But It could work with same planes of an airline, as they would be loath to do favours for other carriers.
      Im off to the patent office as soon as I finish typing this

  2. With excessively deep buffering I understand these satellite connections for aircraft tend to have rather bad “bufferbloat” behavior as well.

  3. … which is worsen by the high latency of geostationary orbit round trip (~ 250ms).

  4. for overland flights there should be no reason to use satellites. upward facing antennas mounted on cell towers would be cheaper, lower latency and greater bandwidth by orders of magnitude.

    Additionally, flat panel antennas could easily be integrated into wing/gear fairings with zero aerodynamic penalty on the bottom of the aircraft.

    Long range overwater would also be better served by a combination of satellite comms in unpopulated regions and terrestrial antennas mounted to offshore windfarm towers in near-coastal offshore flights such as NY-Miami (killing 2 birds with 1 stone).

    • SmartSky Networks will be deploying in the US beginning at the end of this year a nationwide air to ground network that will use terrestrial towers to provide 4G service to the in flight aviation industry. Using that same technology they can take the same efficiencies elsewhere in the world.

  5. These are excellent ideas, but I’m afraid they run into major difficulties with frequency allocations. The blocks assigned for ground communications are separate from the ones assigned for ground mobile, for air mobile, and for satellite communications. These are set by International Telecommunications Union (ITU) policy, and are very, very, very difficult to change. IIRC ITU Plenary sessions are only every four years, so you’d be hard pushed to get any change until the early 2020s. As well as this slow-moving international bureaucracy, there are entrenched commercial interest who ‘own’ big chunks of spectrum and would be loathe to relinquish any.

    Much easier to hope for technical break-throughs than bureaucratic ones!
    🙁

    • one would think Verizon and AT&T would jump at the chance to extort huge sums of money from the flying public for in-flight data using their LTE spectrum. I would not be surprised if they hadn’t already started purchasing congresscritters to enact legislation requiring the use of the 700 MHz spectrum (which they own) for this purpose.

  6. In my humble view, a better solution is to block internet videos on airplanes.

    Rather have content brokers load say 100 movies on storage media inside the plane and using registers and software, monitor the usage and charge the airline for the movies watched. As it is a closed system, piracy is avoided and the copyright owners get their share. A movie on board can be copied quickly to the passenger display device and the user can start/stop fast forward their copy at their own convenience.
    The other advantage of storing the movie on board is that you will be able to show HD or 3D content with the blurring, stopping etc that occurs with internet based viewing.
    Using Wi-Fi, these movies can also be streamed to passenger’s private tablets.
    Now someone will say – hey that is what is happening now on some IFE systems. The problem today is that airlines are forced into purchasing movie rights in a very clunky and wasteful manner. So they either limit the number of movies or agree to pay whether movies are used or not- that is wasteful expenditure.

    • Its actually pretty easy. and has a whole lot more capacity and capability than you even state. A typical 1TB SSD can store roughly 250 720/1080p 2 hour films. Internet video content providers (amazon, Netflix, et al) regularly provide these caching boxes free of charge. A typical caching box with 8-12 SSDs (~16-120 TBs of data) plus CPUs (a simple 45w Xeon-D can easily handle it) would burn on the order of ~100W. The whole issue for cached content really isn’t a technical one. It is primarily a licensing issue. Amazon/Netflix would love to provide IFE and they’d likely do it extremely cheaply if the licensing allows.

      • “Internet video content providers (amazon, Netflix, et al) regularly provide these caching boxes free of charge.” Could you provide some examples? I assume you are talking about something beyond a Content Delivery Network agreement?

  7. The basic problem is you need a lot of bandwidth up north and above oceans, where nobody lives. That’s where the wide bodies fly. That was the problem in 1996, 2006 and still is in 2016.

    I was at the AIX too and folks that have been active in IFE for 20 years and even longer that wonder why new young guys at the airlines keep believing in the ever surfacing new start ups with great booths that proclaim the bandwidth revolution is around the corner and talk about business models while spending the millions of naïve investors onto something they don’t understand but feel good about.

    http://www.macworld.com/article/1017953/connexion.html
    http://padawan.info/en/2006/08/connexion-by-bo.html

  8. What about security? I can’t understand why everyone is talking just about speed? Speed is one feature, not as important as speed, but another very important must have feature is security. People do office tasks and check their emails using in-flight WiFi, which can be easily hacked. Those who understand the danger use a secure WiFi like Ivacy while using in-flight WiFi. Airlines should think about security as well.

Leave a Reply

Your email address will not be published.