It’s a truism of the economics of connected aircraft that the majority of the cost-benefit calculations for inflight connectivity so far have been approved based on passenger use, with growing operational use in many cases depending on subsequent integration, innovation and widespread connectivity adoption. A key part of the technology stack for much of the benefit of connected aircraft is the onboard server, with Tier 1 supplier Thales revealing its new FlytEDGE system as an early use case for this stack — and introducing edge caching to aviation.
Using inflight entertainment (IFE) as a proof of concept and as a way to get key technology enablers on the aircraft is a tried and tested path for the connected aircraft. Here, FlytEDGE is essentially edge caching on an aircraft, and a key part of its tech stack is the Onboard Data Center blade server technology that Thales revealed last year.
This server enables edge caching over cloud architecture, as well as advanced algorithms to select the information to be retained into the edge caches located on the blades, freeing up the increasingly bandwidth-heavy streaming traffic for other uses, including live operational data, and providing a key use case for edge caching in the aviation context, marking the first time edge caching has been used within the aircraft.
On the ground, streaming content is delivered to our devices via edge caching content delivery networks. Here, edge caching loads content onto servers at strategic points close (in network terms) to its customers to minimise the amount of data being transferred long-distance over Internet backbones. Key players here include Akamai and, Thales’ partner for FlytEDGE, Netskrt, and they work mostly on an as-a-Service basis, with the notable exclusion of Netflix, which rolls its own, called Open Connect.
Edge caching on the plane works across remarkably similar lines: strategic content that the system’s algorithms estimate to be the most popular is pre-loaded onto a server onboard the aircraft, to minimise the amount of data being transferred over even longer distances — over 70,000 kilometres at a minimum out-and-back if using geostationary earth orbit satellites.
The core assumptions on which the system’s algorithms are based assume that more than 80% of inflight streaming traffic will be accessing less than 20% of the same content: in essence, most people streaming Netflix are catching up on the latest episodes of The Witcher or Bridgerton or having a box-set rewatch of The Golden Girls or Murder, She Wrote.
Where the system knows that this content is popular, or predicts that it will be, it will be preloaded onto the aircraft, with initial loading of many terabytes likely to be via a traditional sneakernet hardwired method of walking data onto the airplane. When the latest episode of Bridgerton drops, it is downloaded onto the aircraft’s edge cache server only once rather than being streamed over inflight connectivity 250 times.
More esoteric long-tail content (the type that relatively few people watch, but which may be cheaper, and which those customers may especially value), meanwhile, is streamed for the first time over the inflight connectivity system. If the algorithms consider that this content is likely to be of interest to other passengers — perhaps it’s a Japanese movie on a flight from London to New York, but the plane’s next flight is to Japan — it will be saved onto the edge cache server.
Thanks to the blade architecture, some 96TB of content can be retained onboard. Netflix estimates high-definition videos use around 3GB per hour of content, with 4K streams around 7GB. With the equivalent of 96,000GB of storage, that’s many thousands of hours of content that can be saved onboard.
Connectivity and APIs offer further integration opportunities
A basic concept of operations for this system is simply to have passengers who already have subscriptions to a certain service using their existing accounts. As a real-world example, the airline is partnered with Netflix, you already have an account, and you just continue watching episodes of The Witcher on your iPad like you would at home.
One expansion — more a contractual one than overcoming a technical restriction — is to enable streaming to the seatback screen of the entire Netflix catalogue (or a subset: just onboard content, or a selection of the content, or some other sort of taster restriction like being able to watch only two episodes of The Witcher but needing a subscription for more).
A further expansion might be to enable some subset of the content to be authenticated via the airline’s app, either for streaming via a player inside the app, as part of an app-seatback screen integration, or as an API authentication into the streamer’s app with partner download. This might be further segmented: base level frequent flyer members might get a restricted subset, while members with status get more.
New low earth orbit satellite networks and hybrid connectivity systems may well change the game, but it’s unlikely that even the lower latency, higher capacity and reduced cost of these networks as they are presently conceived will be superior in performance or pricing to edge caching on the aircraft.
As the connected aircraft and its related systems develop, connectivity, information sharing and established networks can even reduce the burden on inflight connectivity further. Imagine real-time analytics of the content that passengers in the airline’s lounge are consuming enabling an airline to load the next episodes of that content over cheap, fast gate wifi during the aircraft’s turnaround, rather than over inflight connectivity — either seamlessly or with a “preloaded for you by Example Airlines in partnership with Netflix” badge on the IFE.
Key hurdles remain, but these are largely contractual rather than technological: how will geoblocking work when it comes to content that is sold separately into different geographical markets? (For example: the popular BBC-produced Doctor Who is accessed via that broadcaster’s iPlayer streamer in the UK, but is sold to Disney+ in the US, the ABC in Australia, and so on.) How will privacy rights restrictions (Europe’s General Data Protection Regulation is generally the most protective standard to meet here) work, especially since content history and metadata can in some cases be sensitive?
Edge caching also has numerous other use cases once it has been trialled and tested for streaming entertainment. As just one example, with airline apps becoming crucial tokens for the passenger journey, self-processing and ancillary revenue, how much of that experience can be faster and cheaper for the airline to deliver via edge caching rather than over the air?
;
