HomeNewsArticle Display

Innovation during COVID-19 Operations: 771st Test Squadron brings machine learning to data analysis

Sasha programmer Rony Maida,771st Test Squadron, developed the machine learning tool to aid engineers in data analytics. (Air Force photo by Giancarlo Casem)

Sasha programmer Rony Maida,771st Test Squadron, developed the machine learning tool to aid engineers in data analytics. (Air Force photo by Giancarlo Casem)

EDWARDS AIR FORCE BASE, California --

The 771st Test Squadron has discovered new opportunities for innovation while adjusting to the new normal of telework operations during the COVID-19 pandemic. The 771st has recently employed a machine learning tool nicknamed “Sasha” and is showing promise in data analysis, according to engineers at Edwards Air Force Base, California.

The 771st’s Electronic Warfare testers develop tried and true methods of performing data analysis for the test enterprise. This data analysis feeds test reports for program offices which in turn guide the fielding of capabilities for the warfighter and highlight deficiencies that must be addressed in order to ensure the safety of service members.

As EW systems get more sophisticated, the amount of data collected during each mission has rapidly increased. Analyzing and reporting on these larger volumes of data have traditionally been a manual and time consuming process, but these conventional analysis methods could eventually struggle to deliver data at the speed of relevance. Doing quality and timely data analysis on ever-larger volumes of data require new approaches such as big data analytics and machine learning.

One of the first steps in EW data analysis is reconciling data collected by an aircraft sensor with truth data from a test range. This reconciliation process can be manually intensive and often takes hours or days to complete for a single mission. This is where Sasha comes in.

By training a model with previously reconciled data, machine learning can be used to automatically reconcile a new data set in minutes rather than hours. The capability has been demonstrated on multiple data sets from multiple platforms and when fully matured has the potential to save hundreds of man-hours across the squadron.

The arrival of the COVID-19 pandemic and the 771st Test Squadron’s abrupt transition to 100 percent telework has been challenging but has also offered new opportunities for collaboration and innovation across the squadron. The squadron quickly adopted new collaboration tools and setup online development environments at home so that technical work on projects such as Sasha could continue. 

“Members of the 771st have done an amazing job of adapting and thriving in this new telework environment,” said Lt. Col. Chris Rehm, 771st TS Commander. “Sasha is a good case-in-point, the team quickly figured out how to collaborate on Sasha development remotely and the progress they have made recently is remarkable.”

Recently while teleworking, the Sasha team has been working on a new unsupervised machine learning technique called clustering. While supervised machine learning requires training a model with old data sets, clustering can attempt to classify radiofrequency signals with no a priori knowledge. This type of analysis is useful for both refining a previously reconciled data set and for identifying new signals which were not previously known. 

Sasha and tools like it ensure that EW data analysis keeps pace with an ever changing world and ensures that the squadron continues to provide timely, relevant, and accurate information in the fight to keep American warfighting capabilities the best in the world.