Abstract

Critical industries like avionics or automotive have the possibility of collecting large sets of data during the functioning of their products. Originally dedicated to vital functions as safety and security, data may be collected to any component level nowadays. This is due to the replacement of analogical and mechanical parts by electrical solutions.  For instance, an automotive braking system may be replaced, completely, by a X by Wire solution, meaning that the hydraulic components are replaced by electrical components. To keep safeness and security as accurate as previous components, the sampling frequency of electrical components (sensors, engines, etc) is as high as possible, producing large sets of data. Under the pressure of other industries that benefit from the arrival of statistical methods and models dealing with large sets of data to accelerate their development cycles, critical industries analyze the possibility to integrate such methods and models within their own development cycle. Within this talk, we present pros and cons on the utilization of existing large sets of data, while focusing on the safety. 

Bio

Liliana Cucu-Grosjean is a Research Director at the French National Institute in Computer Science and Automation (Inria) in Paris, France, where she leads the Kopernic research team. Her research interests include real-time, embedded and cyber-physical systems with a focus on the use of probabilistic and statistical methods for analyzing the schedulability of programs and estimating worst-case execution of those programs. Co-author of several seminal papers on probabilistic and statistical methods for real-time systems, Liliana has published more than 60 papers in top TCRTS conferences and journals. Chair of the first TCRTS diversity sub-committees  (2016 to 2020), she has also co-founded the Inria diversity committee in 2015, that she co-chaired until 2022. Since January 2023, she has been the elected IEEE TCRTS vice-chair. Since December 2019, she has been  co-founder of the start-up StatInf, an Inria spin-off.