Internet that flies

Yet despite the amount Australians fly, and despite increasing aircraft utilisation requirements among low-cost and legacy carriers, no Australian airline enables their passengers to get online in the air. Why? As ever in aviation, the answer is complicated, revolving around the technologies involved, government regulation and rulemaking, uniquely Australian issues and ever-evolving passenger demand.

The technologies

Inflight internet is delivered by two forms of technology: air-to-ground cellular transmission or air-to-space satellites.

Air-to-ground (often abbreviated ATG) has a sole provider and a strict geographical boundary: Gogo within the continental United States, most of Alaska, within 100-200 miles of the US border, and most recently in southern portions of Canada. Gogo’s system covers over 2,000 aircraft in the US.

The technology is, in essence, mobile phone technology with transmitters pointing up to the sky instead of down to the earth, with antennas installed underneath aircraft. Gogo’s technology requires between 175 and 200 transmitters to cover its current network, with each transmitter covering a radius of roughly 500km.

First-generation Gogo ATG

Speeds with the first generation of Gogo ATG (on all aircraft outfitted with Gogo prior to the November 2012 launch of second-generation ATG-4) max out at 3. 1 mbps per aircraft – the maximum of the 3G EVDO revision A cellular standard. Naturally, that maximum is a theoretical one, with real-world speeds lower than that on most ATG flights.

As anyone trying to share a connection with one or two other people knows, 3G can get very slow very quickly. Start dividing even the theoretical maximum 3. 1 mbps by 175 of your closest friends onboard a Boeing 737-800 and the connection soon becomes bogged down, as Gogo customers in the US have been discovering. If over 50 people are sharing that theoretical maximum connection, it drops to dialup speeds of 56 kbps.

The problem with the original ATG technology is one of bandwidth rather than speed: too many users slow the single 3G connection to a crawl. So, Gogo tripled the connections with ATG-4.

Second-generation Gogo ATG-4

The late 2012 ATG-4 technology is an aircraft-side upgrade, with two new directional antennas located on the underside of each aircraft, an additional modem and software update, and the ability to take advantage of the EVDO revision B standard.

Capable of a peak speed maximum of 9. 8 mbps, more than tripling the old system’s capacity, the technology increases the width of the pipe instead of its speed: three times as many users can have a reasonable experience.

No Australian airline enables their passengers to get online in the air. Why?

Third-generation Gogo GTO

An increase in speed per user will come from Gogo’s Ground to Orbit (GTO) product: essentially, adding the 50 mbps download speeds of a Ku-band satellite system to the existing ATG-4 product for a maximum download speed of nearly 60 mbps. Announced in late 2013 and in line for release in the second half of this year, GTO is expected to provide a step change rather than the incremental update of ATG-4 alone.

And so to satellite

Australians living outside urban areas will already be familiar with satellite internet, and the idea is similar for satellite-enabled aircraft. Geostationary satellites tens of thousands of miles above the earth communicate with antennas on top of aircraft, meaning that fewer satellites than ground-based stations are required to cover the same geographical area.

Airborne satellite internet currently spans three sets of frequencies within the microwave spectrum: in ascending order of maximum speed, L-band, Ku-band and Ka-band. (Ku and Ka are pronounced kay-yoo and kay-ay, rather than koo and ka, and refer to K-under and K-above, the bands surrounding the K-band. )

L-band satellite

L-band to aircraft is provided by SITA’s OnAir subsidiary, utilising Inmarsat’s SwiftBroadband-enabled Broadband Global Area Network (BGAN) system on the I-4 satellite constellation. Each channel supports up to 432 kbps, while a dual-channel link can be bonded together, resulting in a relatively slow maximum speed of 864 kbps. Hardly swift, and arguably not even broadband.

Realistically, a relatively small number of users at a time can have a decent experience on L-band – OnAir quotes 12 simultaneous users. It’s firmly in the “better than nothing” camp, but most people find themselves restricted to text-only email and the mobile web.

Inmarsat’s High Data Rate (HDR) standard debuted in December 2013, boosting maximum speed to 850 kbps per channel (and thus 1. 75 mbps per dual-channel link). The global coverage of the BGAN network, no matter how amusing the acronym may be, may mean that there is a place for it in areas of the world not served by newer, faster bands and standards. That may well be the case in Australia.

Pre-HDR L-band was the technology Qantas trialled in 2012 on six Airbus A380 aircraft. Pricing during the trial varied between $12. 90 and $39. 90 for data allowances of up to 35MB. These days, that’s barely enough for most people to check their email a few times.

Qantas concluded its trial without an order for further L-band systems, blaming a take-up rate lower than expected. The airline noted at the time that its A380 flights operated primarily during night-time hours and that usage was lower than expected. With slow speeds and eye-watering pricing, Qantas’s move wasn’t necessarily unexpected.

But what if faster internet were available above and on routes from Australia? Since Qantas’s trial began, Panasonic has significantly boosted its Ku-bandcoverage and redundancy, making it a significantly more attractive proposition than in 2012.

Ku-band satellite

Ku-band airborne connectivity is available from the largest number of providers, including Panasonic Avionics, Row44 and Gogo (not to be confused with Gogo’s US-only air-to-ground service). Each provider is able to contract with with a different set of Ku-band satellites, meaning that global connectivity is theoretically available, with the exception of some polar regions.

Panasonic’s coverage, for example, uses 17 satellites and 23 beams to reach 209 out of 210 countries, the exception being China. Naturally, that presents an issue for numerous airlines relevant to readers of Australian Aviation, airlines which cross Chinese airspace en route between Asia and Europe or North America.

Ku-band speed varies according to the amount of the globe a satellite covers: wide-beam connectivity typically reaches 25 mbps but covers a wider area, with 1-2° spot-beams covering a significantly narrower area but practically doubling speed to 50 mbps. At speeds like that, a good proportion of the airplane can be connected at relatively fast speeds. At a satellite level, current-generation Ku-band satellites support between two and six Gbps in total, spread across all the aircraft (and other ground users) connected to the satellite.

The speeds available to Ku-band satellite technology also enable airlines to offer live or semi-live TV content, over the same channel as inflight internet data. Panasonic currently offers Sports24, BBC World, BBC Arabic, Euronews, NHK World, CNBC, CNN World, Al Jazeera English and others to its airborne Ku-band customers. The lineup of channels varies and can be customised by region, and Panasonic confirms that it could easily offer cricket or footy over Australia if its airline customers demanded it.

Ku-band’s next step will be evolutionary and on the antenna side. Starting this year, a two-panel antenna with improved performance, lower weight and lower costs, yet fitting within the same radome and mounting provisions, will be offered. The next big step for antenna technology will come when the first flat antenna – with significantly less drag and lower costs – can be installed. Currently, both Ku-band and competing Ka-band antennas are the same size and use the same mountings, with a penalty of 1. 2-1. 4 drag points. Adding an additional antenna essentially doubles speeds, which is yet a further potential improvement.

The next generational step for the satellite side of Ku-band technology is to add new satellitesand extra spot-beams, notably using the EpicNG program from Intelsat. EpicNG enables tighter, high-density spot-beam transmission in addition to the existing 1-2° spot-beams, and each satellite has a capacity of 25-60 Gbps (an order of magnitude greater than the current generation). The company plans to launch satellite 29e over the North Atlantic in 2015 and satellite 33e over the Indian Ocean region in 2016.

Intelsat claims that the two EpicNG satellites will cover all the populated continents, yet its coverage maps do not include most of Australia – coverage ends at roughly 141° east longitude (the eastern border of South Australia) and the 29th parallel south (most of the NSW/Victoria border).

Existing Ku-band satellites already cover Australia, though questions over domestic satellite demand from users on the ground outside major urban areas as part of the National Broadband Network (NBN) are likely to affect the ability to provide airborne satellite connectivity. The existing Ku-band NBN satellite service, started in 2011 as an interim option, is already seeing contention issues affecting speed and usability, and is likely to reach its user capacity this year.

While numerous question marks still exist over the future of the NBN and its plans to transition from the interim Ku-band to faster Ka-band satellites, the significant lead-time required to add satellite capacity over Australia restricts supply, and the Commonwealth Government’s reassessment of plans for satellite ground-based broadband under the NBN raises significant questions over demand. Global satellite businesses, whose connectivity is under global demand, may reassess the capital assets they commit to orbit over Australia with rural broadband demand in doubt.

Ka-band satellite

Ka-band is only offered on commercial flights by ViaSat’s Exede in the Air service, which is limited to the United States. ViaSat’s Yonder service has some limited Ka-band coverage in Europe, the Middle East and North Africa, but airborne Yonder is restricted to business and VIP jets.

Being significantly less crowded than Ku-band, and with a larger frequency allocation, Ka-band already starts with advantages.

Inmarsat will be in the Ka-band game soon with its Global Xpress inflight service, based on three satellites, the first of which launched in December 2013 and two more of which will launch in the second and third quarters of 2014. A fourth satellite will provide redundancy and is intended to be delivered in mid-2016.

Ka-band is significantly faster than Ku-band offerings. Inmarsat is advertising download speeds of 50mbps for Global Xpress, and although ViaSat refuses to be drawn on speed maximums, it will say that its standard is 12 mbps per passenger and that it can sustain 50-70 devices per flight. That would work out to between 600 and 840 mbps per aircraft – over 15 times faster than Ku-band.

12 mbps per passenger is no marketing swizz either, as this writer can personally attest. When testing out the system on its launch day aboard a jetBlue test flight in mid-December, I reached speeds of over 25 mbps while dozens of others on the flight were livestreaming, watching HD video, using Skype, FaceTime and Google Hangouts, and listening to streaming music. With a planeload of aviation and technology journalists banging on the system and trying to make it fall over, it didn’t.

jetBlue announced that it had sustained 111 simultaneous users on two flights.

The following day, jetBlue announced that it had sustained 111 simultaneous users on two flights. The test will come, though, when more aircraft are connected to the single ViaSat-1 satellite: as of late January 2014, jetBlue had fewer than 10 aircraft enabled with the Ka-band product.

A second satellite, ViaSat-2, is scheduled for launch from Baikonur in Kazakhstan in mid-2016, covering North and Central America, the Caribbean, and most aeronautical routes between North America and western Europe. ViaSat-1 was delayed by nearly a year in 2011 owing to damage to the satellite while being moved for testing and the grounding of the Proton rocket program following a launch failure leading to the loss of an unrelated Russian satellite.

Regulator delays

As Australian Aviation goes to press, the first radome has cleared the US Federal Aviation Administration’s (FAA) significantly more stringent requirements for birdstrike testing. The Ka-band radome from jetBlue’s LiveTV subsidiary that supports inflight internet onboard Airbus A320 and Boeing 737-900 aircraft has passed real-world birdstrike tests to show survivability following a 1. 8kg birdstrike (somewhat amusingly, by a chicken) at design cruise speed at sea level or 85 per cent of design cruise speed at 8,000ft.

The supplementary type certificate (STC) for these aircraft – and numerous others – was granted in 2013 subject to additional birdstrike testing required under airworthiness standard 14 CFR 25. 571(e)(1). Essentially, this resulted in inflight connectivity providers installing systems at risk under a temporary FAA exemption. In 2013, the FAA had increased its scrutiny of birdstrike testing on Ka-band and Ku-band radomes, requiring some real-world testing in addition to simulator modelling.

Unaffected were Boeing 747 aircraft, with the unique shape of the upper deck precluding a birdstrike on radomes installed further back along the fuselage. Numerous single-decker aircraft, most notably the Boeing 777, Boeing 767 and Airbus A330, have seen connectivity installation delays due to the 2013 set of requirements.

So far the FAA has not required further testing for previously issued STCs in order retroactively to meet the new requirements. If it did, Row44’s Ku-band systems, installed most notably on hundreds of Boeing 737 aircraft operated by Southwest Airlines and Norwegian, would see the largest impact in terms of requiring change or rework.

With Australian regulators letting their international counterparts take the lead on radome testing, the impact has been widespread and opaque. Providers’ announcementsare significantly limited by public disclosure issues, and airlines obviously would not highlight delays to their rollouts.

Airlines new to the inflight internet game have mostly been waiting for Ku‑band and Ka-band availability, given the restricted speed of L-band and the restricted geography of Gogo’s ground-based systems.

As the risk of rework diminishes with further full STC clearances, it is likely that those airlines will be in the market for connectivity.

The always-on traveller

The FAA has also been shaking up the industry with new portable electronic device (PED) rules following recommendations from its PED Advisory and Rulemaking Committee. Gate-to-gate electronic use is now permitted in the US on an airline-by-airline basis, and increasingly airlines outside the US are following suit.

But there’s still an offline air-gap between pushback and the 10,000ft ‘bing’ that means you can turn on Wi-Fi. That’s a significant part of the overall flight time for shorter flights – and that’s especially true on Australia’s triangle routes.

The only option currently even technically eligible for gate-to-gate internet is satellite connectivity. Air-to-ground isn’t engineered to work under 10,000ft because the antenna is underneath the aircraft, and because significantly more towers are required to cover the same area at 5,000ft than 35,000ft.

Make no mistake, gate-to-gate connectivity is possible and will be achieved. After that, the next step is to link together Wi-Fi providers in the terminal and on the aircraft for a seamlessly connected journey.

Australian opportunities

The prospective fleet for inflight connectivity equipage in Australia is restricted by a number of different factors.

L-band could be installed on most aircraft, yet Qantas’s trials and the calculations of trying to split a SwiftBroadband HDR maximum of 1. 7 mbps between a jetful of passengers suggest that L-band lacks capacity for anything but the smallest of aircraft or the most price-insensitive users. Airlines are unlikely to equip their fleets with a technology that is unable to support the demands of existing users, let alone the projected requirements of users at the end of the aircraft’s service life with that airline many years hence.

At the smaller end of the scale, no turboprop has Wi-Fi on board, and small fleets of out-of-production jets like QantasLink Boeing 717s or Virgin Australia Regional Fokker 100s are unlikely to provide a return on investment.

Ku-band and Ka-band radomes have only just been granted STCs for the Airbus A320 and Boeing 737-900 – the former operated by Tiger and Jetstar plus two aircraft under the ex-Skywest Virgin Australia Regional banner, and the latter not in operation here.

Boeing 737-700 and -800 aircraft operated by Qantas and Virgin Australia are still subject to the risk of birdstrike STC rework for Ku-band installs, and have not yet been certified for Ka-band installs alongside the A320 and larger 737-900.

Medium-haul Boeing 767 and Airbus A330 aircraft are in birdstrike STC limbo, as are longer-haul Boeing 777 aircraft. In November 2013 Lufthansa was granted permission to retrofit

Australians are left with foreign carriers if they want connectivity beyond the occasional L-band disappointment.

Panasonic’s Ku-band systems to the Airbus A380, though Qantas’s statement that the A380’s overnight flights saw low Wi-Fi take-up rates are likely to steer the airline away from an early decision in favour.

At the top end of the aircraft capacity spectrum, only the Boeing 747 is currently a possibility for Ku-band or Ka-band installations, and those Qantas aircraft both operate similar routes to the A380s and also completed a $250m refurbishment as recently as December 2012. Qantas’s well-publicised financial issues are also unlikely to make a further capital investment into its loss-making international division a high priority.

So Australians are left with foreign carriers if they want connectivity beyond the occasional L-band disappointment. Westbound, that means a number of Boeing 777‑300ER aircraft in Etihad’s fleet, and around a half-dozen of Emirates’ 777-300ERs, though the Dubai-based airline is planning to install Panasonic’s Ku-band services on over a hundred 777s. Most of United’s 747-400s which serve Sydney and Melbourne are already fitted with Panasonic’s Ku-band service, but they will be replaced by 777-200ERs, which are currently in Ku-band STC limbo, from April 1.

Economic feasibility

While crunching the numbers for coverage on the Sydney-Melbourne-Brisbane triangle routes might sound reasonable, the capital expenditures required for the infrastructure, particularly given Australia’s population density, are problematic. Despite Australia and the continental US both being roughly 7. 7 million square kilometres, Australia’s population is roughly seven per cent of the United States’, similar to the population the state of New York alone.

And in terms of potential benefit, just over 100 daily flights operate on the strict triangle routes. While the nearly 50 flights between Tullamarine and Kingsford Smith alone make it one of the world’s ten busiest airport pairs, it may not be enough on its own to provide demand.

Use a five per cent take-up rate from 150 passengers on your average A320 or 737-800 on the 100 daily triangle flights, assume each of those takers will spend $10 for connectivity, and you get $7,500 a day or $2. 7m a year – to be split across all airlines on the triangle route. Take-up rates are higher on longer flights, but Australia has fewer of those, primarily Perth flights, many of which are red-eyes on the eastbound leg.

And airlines can’t count on being able to charge forever. Finnair openly admits that it is constrained in what it can charge for its L-band internet trial by the fact that competitor Norwegian offers faster Ku-band for free. That will be just the start of a trend worldwide.

Until the numbers add up to profitability, which of Australia’s airlines will be first to foot the bill?

The pace is inexorable

Inflight connectivity is happening elsewhere and it will happen in Australia. Regulators and hardware providers will get their houses in order. Technologies will enable gate-to-gate Wi-Fi, not just gate-to-gate electronics usage. Take-up rates will increase as passengers’ lives become more online, and the airline side of the benefits column will increase as more elements of the aircraft and airline operations are added into the Internet of Things. Installation costs will go down as linefits increase and aircraft lessors will be able to amortise capital outlays for radomes over an aircraft’s lifespan.

But for now? You’ll still be offline above the clouds.