Microsoft Azure Global CTO Bill Chappell confirmed a successful flight test using a Ball Aerospace phased array antenna to enable direct satellite to cloud connectivity during a remote keynote speech he gave as part of the Global Connected Aircraft Cabin Chats web series. Photo: Microsoft Azure Global
Microsoft’s Azure Global is a growing super-computing network of data centers, integrated cloud computing services and algorithms that has the potential to transform the way Internet of Things (IoT) connectivity is managed, processed and routed by the aerospace industry, according to a keynote speech the company’s Chief Technology Officer (CTO) Bill Chappell gave remotely during the recent Global Connected Aircraft Cabin Chats virtual event series.
“We like to call Azure the world’s computer, that’s a bold statement, but fairly true in regards to the hyper-scale nature of the system,” Chappell said.
Azure’s massive global platform includes more than 160 physical data centers, organized into regions where Microsoft has geographically located “latency defined perimeters,” according to their website. The world computer is also powered by more than 100,000 miles of fiber optic and sub-sea scale cabling and 150 edge locations on the ground that currently serve more than 20 million companies globally.
Within these regions, customers have service level agreements to run their individual IT systems, services and applications on Azure’s virtual machines, edge computing, deep learning algorithms and other products that enable digital transformation.
“For people who are wanting to tap into that network in, you can have an ExpressRoute which directly ties into a private connection to the cloud. We also have express route for satellite, that’s a new offering connecting to satellite vendors to directly tap into that network,” Chappell said.
The ExpressRoute service uses Multiprotocol Label Switching - a private network data routing technique for faster connections - connectivity to enable private connections to the Azure network for individual businesses and companies. In October 2019, the ExpressRoute service was adapted to start offering satellite data connections to Azure cloud through SES, Intelsat and Viasat.
Another major aspect of Azure is the collection of products it supplies based on the regional latency perimeter global digital infrastructure that has been established. In France, for example, Azure includes its IoT Hub and Machine Learning products, whereas the North American regional offering features the largest arrangement of Azure products, including a program called “Cognitive Learning” that can build natural language processing into native and third party IT applications. Other built in products include Azure Maps, where users can create location-aware web and mobile applications using geospatial data-powered software development kits.
Microsoft is continuing to expand the number of data centers it manages to enable its world computer's processing power. Photo: Microsoft Azure Global
Inside each Azure user’s ExpressRoute is access to the many virtual machines, edge devices IoT sensors and deep learning algorithms Azure has built into its world computer. The platform’s ability to not only provide faster connectivity but also to take data and applications and train them to improve end user digital workflows could be transformational for the aerospace industry.
“What you can do is figure out, do I want more processing on my satellite – it would apply for airplanes as well – what is the value of an ounce of more processing? Do I send data to the ground or horizontally through my network? Sometimes that’s done with pen and paper, or Excel sometimes,” Chappell said. “Now you have a real environment to be able to model the communication network, model the algorithms and then you can have real algorithms on real hardware exploring your trade space in much much higher fidelity.”
During his keynote, Chappell confirmed a recent successful flight test conducted with the Department of Defense (DoD) where Microsoft worked with Ball Aerospace to prove it could put Azure-enabled satellite connectivity onto an airplane.
“Through a LEO network with one of our partners, we worked with Ball Aerospace antennas, up through the LEO layer, back down and into our cloud and maintained connectivity,” he said. “What was interesting, we brought a suite of AI capabilities onboard a plane that hadn’t been touched since the 1980s in terms of modernization, and now you have this portal into hundreds of cloud services. We’re very excited to be able to do this early demonstration to have shown what an expanded cloud presence is and how it would impact the connectivity and connected aircraft community.”
The demonstration referenced by Chappell was the confirmation of a collaboration between Microsoft, Ball Aerospace and General Dynamics Mission Systems to test the use of Telesat LEO, a low earth satellite constellation which is scheduled to become available in 2022, to enable direct satellite-to-cloud communication and faster data processing for aircraft than what is available today. Testing included transmission of low-latency high-speed links that were “downloaded into the Microsoft Azure cloud and processed using Ball algorithms,” according to the Telesat press release. This demonstration was performed as part of the U.S. Air Force's Space and Missile Systems Center's Commercially Augmented Space Inter Networked Operations (CASINO) program.
Chappell did not provide any details on whether Azure Global would be looking to directly offer in-flight connectivity service plans to airlines or what type of aircraft was used to demonstrate their satellite Azure connection. However, he did say that the company sees itself as more of an enabler than a direct service provider of connectivity to aircraft.
Presagis used Azure's geospatial data and mapping software to decide how to develop image and object detection algorithms for a generic drone's three different payloads. Pictured here is their effort to use Azure to train a drone on flying from one portion of Madrid, Spain to a destination. Photo: Microsoft Azure Global
Azure Global’s ultimate goal in aerospace and all industries is to allow users to take advantage of its massive data processing and cloud computing storage proficiency. One of the other demonstrations shown by Chappell was the use of Azure to autonomously train a drone operated beyond line of sight to fly from a rooftop to another rooftop destination in the city of Madrid. Video of the flight shows how Presagis, a Montreal-based supplier of advanced simulation and graphics software, was able to use Azure’s digital geospatial data and mapping of the city to decide which object detection and identification or waypoint to waypoint navigation algorithms to develop for the drone’s individual cameras and sensors.
Where Chappell sees their world computer network evolving in the near future is coming within the radio frequency and wireless communications development community. The CTO sees Azure unlocking next generation capabilities not only for significantly faster connection speeds, but also for the use of artificial intelligence algorithms to design things that previously were only developed by humans. An article written by a team of engineers from a Virginia-based startup, DeepSig, published by the IEEE’s web magazine in April was highlighted by Chappell to show the type of transformational use cases Azure can support in the future.
“You’re starting to see this impact of AI and ML on the design cycle for things like wireless signals. If you look at connectivity, we typically pre bake in a wave form, work really hard to make that waveform work as good as possible and then that waveform gets distorted by the environment we place it in.” Chappell said. “What they’re demonstrating is the ability to create that waveform on the fly on demand based on present conditions. They’re able to do that working with NASA for satellites and demonstrate that satellites can generate the waveform as needed, but it is generically a capability we’re starting to see pop into the [radio frequency] RF space.”