Blog Post

Why your in-flight Wi-Fi is slow and expensive: It’s all about the pipe

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

It has been five years since American Airlines first launched Wi-Fi on its domestic flights in the U.S. and, frankly, since that time, Wi-Fi on airplanes is still stuck on the runway. It’s expensive, slow and based on GoGo data, less than 10 percent of flyers even attempt to use the service, even if someone else is paying for it.

But there’s a new promise on the horizon, with a leaked memo showing that JetBlue plans to offer Wi-Fi on planes in 2013 from a new provider, a service that could offer up to 12 Mbps per passenger on a flight thanks to a new type of satellite. But this scenario is unlikely, and understanding why explains why Wi-Fi on planes costs so much and is relatively slow. It also offers lessons on the limits of wireless.

So as entitled as we might feel to the same Wi-Fi experience in the air as we have on the ground, it’s not going to happen overnight. It’s going to cost more, have less capacity and come with limits. Don’t blame the airlines (or worse, the flight attendants): blame physics. After all, getting a mobile connection in a metal tube flying at a speed of 500 miles per hour is actually pretty freaking phenomenal.

It’s the economics, stupid.

The most fundamental cost is the cost of delivering data over the network, measured as the cost per megabyte. Currently there are three different types of networks, with the Exede satellite that JetBlue will offer being a fourth.

GoGo: The top provider in the U.S. (American Airlines (s aamrq) and Delta (s dal) are GoGo customers) has built a ground-based network that provides capacity up to planes, as opposed to beaming signals from a satellite. Tim Farrar, a satellite analyst, estimates that it costs GoGo roughly 20 cents to deliver a megabyte of data on a plane. That’s about $200 for a gigabyte, which makes AT&T’s (s t) and Verizon’s (s vz) $10 per gigabyte charges for wireless data look reasonable by comparison.

Satellite in the Ku-Band: Row 44 and Panasonic Avionics are both buying satellite capacity in this band. Airlines such as Southwest (s luv) and some United (s ual) carry this service. To afford their connectivity plus operate their business Farrar estimates they have to charge roughly 20 cents a megabyte to break even. At that price downloading a standard definition TV show that’s an hour-long would cost roughly $100.

Satellite in the L-Band: This is really expensive, but on international flights Inmarsat is subsidizing the cost somewhat because it wants to keep customers while it waits to launch a cheaper alternative, said Farrar. Singapore Airlines and Emirates use this service, but if they were paying full price their connectivity would cost $5 per megabyte, he said. Because that’s essentially the connectivity they are buying. Needless to say, these don’t offer rapid speeds.

Satellite in the Ka-band: Here’s what JetBlue is going to implement, and what Inmarsat and even GoGo hopes to deploy in 2015 to its customers. Using the Exede service delivered by ViaSat, airlines will pay roughly 2 or 3 cents per megabyte for their connectivity said Farrar. The service depends on a new satellite launched in 2011 that can deliver up to 12 Mbps per person on a flight according to ViaSat.

What do airlines want their Wi-Fi to do?

Since network connectivity is the largest continuing cost of Wi-Fi on planes, it’s clear to see why the services are so expensive. It’s also why they are so slow. Even if the VisSat bird can deliver 12 Mbps per passenger on a plane, streaming a two-hour SD movie at a gigabyte will still cost between $20 and $30. How much of that will be subsidized, and who will do it?

The answer can depend on the agreements the airlines sign with the Wi-Fi companies as well as their goals in offering the service. GoGo isn’t designed for every passenger to hop online –it’s designed so premium business-class passengers will hop online. And they will pay for it, as evidenced by the recent price increases. This is a cold calculation that only a few people will pay the additional fee, but those people will pay a lot.

Other airlines may want to use Wi-Fi to entice passengers to fly with them as it seems JetBlue might. In that case, going with a cheaper service that you can offer on a limited basis to passengers makes sense.

This Wi-Fi is not home or event hotspot Wi-Fi.

Given the cost, the smaller pipe that satellite broadband offers, it should be easy to understand why you can’t expect to get the same Wi-Fi in a plane as you do at home, or even at Starbucks. At its median, it’s 20 times the cost of cellular data and the bandwidth is roughly that of a 3G network.

All Wi-Fi is not created equal — the backhaul to the Internet determines its capacity and how quickly you can download things. Your home Wi-Fi, if connected to a fast cable or fiber connection, is connected to a fire hose. The Wi-Fi from an LTE mobile hotspot is more akin to a garden hose and the Wi-Fi from current in-plane systems is like a drinking straw (some like L-band are like cocktail straws).

So don’t complain, after all, I managed to write this post on a plane with no Wi-Fi at all.

Alaska Airlines airplane image courtesy of Flickr user as737700.

22 Responses to “Why your in-flight Wi-Fi is slow and expensive: It’s all about the pipe”

  1. Martijn Moret

    Amusing comments from people who have been lying under a rock (there are more device than just laptops), to semi-philosophers (we might want to be connected everywhere if we choose to do so), to physics-fundamentalists (it is no miracle, but still very cool, as is a picture from a bot on Mars).

    Thanks for a very good and clear written article. Clearly the market (demand) is there, and we are just at the beginning of a changing onboard experience. Even without internet connectivity, possibilities to connect devices to each other onboard are endles, e.g. for crew-passenger interaction.

  2. Considering that a satellite can transmit its data to ground at rates exceeding a gigabit a second at a speed of 18,000 mph; transmitting wi-fi from an airplane at 500mph is child’s play.

    This isn’t a challenge of physics, it’s a challenge to provide the service at a low price.

  3. Satellite Spy

    Well, this little post has fired up a few folks!
    Stacey has got it right, as has HiSpeedWiFi.

    Never mind whether it’s Wi-Fi, phone, or whatever from the passengers’ point of view in the plane. That’s just bog standard TCP/IP stuff like you get at home. But imagine a home with 400 people all downloading videos at the same time. That’s a heck of a large composite data rate!

    The fundamental issue is the data rate that has to be supported between the plane and the satellite for a given quality of service (e.g. BER >1 in 10**7 for 99.9% of the time). That comes down to a few things, including:
    (1) The satellite transmit power (EIRP) and the plane terminal’s EIRP & G/T.
    (2) The issues of the active tracking antenna on the plane, be it a mechanical, steerable dish or a phased array. For a good design, allow 3dB in the link budget.
    (3) The effects of Doppler shift due to the plane’s movement, and how the modem’s carrier & clock recovery circuits (algorithms) cope with this.
    Couple this with the performance of modern modems using ACM (adaptive coding & modulation). This means scaling the data rate according to the time-varying channel degradations.
    (4) Propagation conditions – signal fading due to atmospheric attenuation, scintillation etc. This is frequency-dependant, but planes are usually above the weather, so it’s not such an issue. Perfect for the next generation, high-capacity Ka-Band frequencies.
    In a nutshell, it comes down to how big a tracking antenna can be mounted on an aircraft.

    OK, this is all boring techy stuff. But this is the reality! Very good blog post, Stacey. :-)

    Background: My company (CSR Ltd) developed the very first satellite modems used for passenger comms on commercial aircraft in the late 1980s – taken up by British Airways (SkyPhone service), Singapore Airlines, Lufthansa et al.
    Personal website and @satispy on Twitter.

  4. Richard Skaife

    WiFi access on commercial aircraft is all about the economics of the service. The cost driver is the backhaul from the aircraft, either satellite or direct to the ground (a la Airlink). There simply are not enough passengers wishing to use the service to make the business case close so you have to regard the overall system as an aircraft information service, providing information services to both passengers and the airline. The airline services are such things as real time credit card clearance, aircraft cabin operational management (maintenance reports to prepare cabin job cards in advance etc, passenger information – connecting flight data etc – using this system instead of ACARS), then you can get into aircraft data systems – engine performance etc for managing the aircraft maintenance, electronic flight bag…. the opportunities are almost endless. Once you view the service in these ways the business case for satellite backhaul begins to make sense. Standalone passenger services is a pretty tough case to make.

  5. What a great summary…….. Yes we are working on the right physics (RF photonics) to solve the capacity problem for satellite communication. (

    Our technology provides capability to receive multiple (=50) satellite transponders simultaneously (each delivering at 50 Mb/s of capacity and covering the 2000 MHz KuBand Rx bandwidth). And recently we proved multibeaming capability with the same technology, enabling real-time connectivity even whenchanging satellite constellation. So you wont miss a minute of your Live soccer (or basevball) match.

    Regarding the economics, our system is conformal and therefore saves € 100.000 a year on kerosine for the airline but more important with a capacity of 5GB/s it is a moneymaker !!!

    Paul van Dijk

  6. The current costs mentioned don’t factor in the effects of scaling, and inevitable drops in prices like almost any other product. It’s still a little chicken-egg though–more people won’t use it if it’s slow, and it’ll be slow until enough people sign on to lower the per/mb cost.

  7. It's got to be better than this!

    “So don’t complain, after all…….”
    GoGo service would fall flat on it’s face if 50% percent of people on a plane were actively using it.

    Where there’s a need, there’s a business opportunity. I hope to see new players in the market doing all of these competitor one better.

  8. I’m fairly certain Air Force One has no problem implementing high-speed internet bandwidth. And that’s alongside a veritable arsenal of military-grade communications gear commercial 747 aircraft don’t even carry on board.

    Fact: it’s doable. Author’s excuses aside.

  9. “Getting a mobile connection in a metal tube flying at a speed of 500 miles per hour is actually pretty freaking phenomenal.”

    Sorry Stacey, but dealing with the laws of physics, I don’t see why that statement should be true, because:

    (1) Radio waves travel at a speed of 186,00 MILES PER SECOND. Compared to that, 500 miles per hour is nothing. From the viewpoint of the radio waves, the airplane is basically standing still.

    (2) Okay, the plane is a metal tube. But the airplane itself is not used to receive the Wi-Fi signal from the ground. That’s what antennas are for, and an airplane has many of them and they’re used to transmit and receive communications and navigation signals that the plane needs to fly. All those antennas are designed to operate properly when mounted on a metal tube. Whichever antenna is used for Wi-Fi will do the same.

    So, using the laws of physics, I can’t see how either the speed of the plane or its being a metal tube is relevant to the performance of Wi-Fi. If Wi-Fi on planes is expensive and slow, it’s because of other reasons, probably related to business and economics., as you mentioned.

    • HiSpeedWiFi

      Actually this isn’t about how fast the airplane is moving with respect to the speed of radio waves. It’s about how rapidly the environment or the “channel” in which the radio waves are traveling to the plane is changing. The faster the plane moves, the more quickly the channel changes, and the harder it is to compensate for these changes to recover the data encoded in the radio waves. As usual, wikipedia has the details:

      • Keep in mind, the antenna is required to track the satellite, basically compensating for the satellite’s fixed position while moving. That is the real challenge, not the speed at which the radio waves reach the antenna.