Millions of data points flying in tight formation

By Dhaval Shah, Product Manager, eInfochips

In Part 1 of this article, the impact of Big Data on inflight entertainment was analyzed. Some of the same technology can improve airline and business aviation operations and safety.

For the past several years, much has been written about systems data collection onboard modern airplanes: GE jet engines collect information at 5,000 data points per second; a Boeing 787 generates an average of 500GB of system data a flight; an Airbus A380 is fitted with as many as 25,000 sensors. Much of this data is transmitted or downloaded to plan maintenance, position spare parts, and anticipate component failure. These purpose-built systems represent one of the fore runners of what has grown into today’s Big Data and the Internet of Things (IoT). These systems have provided significant savings to operators for several years. They have also allowed airplane and engine manufacturers to offer new and lucrative data-driven business models such as pay-by-the-flight-hour, so we can expect this evolution to continue.

While data coming from aircraft maintenance sensors is growing, new technology like NoSQL databases, broadband connectivity, and stronger onboard processing capacity provides an opportunity for system integrators to develop new kinds of operation and safety systems. By collecting the unstructured data generated by these sensors right from the source and merging it with additional sources of raw data onboard and from the ground, technology such as MongoDB can provide on-the-flight intelligence to operate the aircraft more efficiently and safely; collect lessons learned and pass them on for further pilot training; as well as provide operators with an overall operational risk management perspective.

Most aircraft operators are now providing their pilots with the latest tablet technology as a convenient and powerful device to access navigational and technical manual data previously only available on paper or dedicated systems. Generally called electronic flight bags (EFB), these commercial-grade tablets offer the opportunity to create new applications derived from Big Data. By analyzing on the fly raw data from the maintenance sensors, geospatial positioning, weather radar, as well as operational data from other aircrafts in the vicinity or on the same route, pilot could soon access a total-awareness application answering questions such as: “What is happening?”, “What action should I take?”, or “What could happen?” Merged data can be superimposed on graphical and video data to give pilots a complete and visual answer to allow for more efficient operations. For example, this kind of application could suggest to the pilot a more optimum engine throttle setting, a more advantageous or comfortable altitude, or a better anticipated route around inclement weather.

In turn, the same data can be collected for hundreds of flights across an entire fleet and provide great operational overview of specific aircraft types or routes. Analyzing this Big Data and combining it with business intelligence and behavioral analytics could produce great training scenarios for ongoing pilot proficiency as well as realistic flight simulator challenges. This would provide airline and business aircraft operators not only with the outmost safety preparedness, but also potential costs savings. For example, data-driven training for improved throttle control, aircraft trimming, and landing round-out could save operators significant costs in fuel consumption and component fatigue.

The same data scenarios could then be reloaded into the EFB and provide an aircraft and route-specific operational baseline to offer the pilot with a live, private, and continuous self-evaluation. Early versions of these kinds of new applications already exist in the military to evaluate a pilot’s proficiency against specific mission criteria. Usually combined with video processing, it is one of the most efficient pilot improvement tools in the U.S. Air Force. Such capability is now in reach of the commercial aviation thanks to off-the-shelf technology and software integration with Big Data.

Most aircraft operators have already established state-of-the-art network operation centers (NOC) that provide an overall flight situational overview of their fleet. Until recently, these centers were relying mostly on air traffic control data or GPS positioning and short messages sent down from the aircraft. But collecting Big Data information described above and combining it with customer data and business intelligence in new database technology like MongoDB, these NOC can now provide instantaneous analysis and risk management at a fleet level to provide improved operations and not just maintenance cost savings.

Finally, the last resort data collection system in aviation is the “black box.” Introduced in the mid-1950s, it collected data from a few sensors on a hardened media that was designed to survive a catastrophic aircraft failure. Over the years, more sensors were connected and more data was added to the black boxes. FAA requires just a minimum of 88 parameters to be collected. Today, these devices amount to some pretty indestructible structured database systems that, when recovered, can provide answers about the last few minutes of a flight. However, many times, like in the recent case of the MH370 disappearing in the South Pacific, it is unlikely that the black box will be useable, if retrieved at all. By combining the data available on board with broadband connectivity, streaming black boxes will soon be designed by specialized system integrators. These new devices will send much more unstructured data to the ground on a constant basis, including maintenance sensors, geospatial poisoning, flight control data, audio recordings, and video images that could all be combined and analyzed instantaneously in case of an emergency. These systems will be hardened to send such information for as long as possible even after a major system failure.

The key to all these improvements is the new approach to store massive amounts of raw data from various sources in the same database without processing it. Operational and safety gains are then obtained by using advanced queries that assemble and analyze the right data just when it is needed, thus requiring much less computing power but providing superior information integration and usability. In addition, because the Big Data is unprocessed, it can be used in the future for yet-to-be-defined applications.