Innovation never stops
Inventors’ Awards are a celebration of those at Atos who bring inspiration to life, diligently driving ideas through development cycles and patent processes. Learn about the inventions and inventors here.
“The 2022 Inventors’ Award winners have been selected given their impact on people and the planet. They’re saving lives, securing private data, preventing fuel leaks and theft, and saving electricity. Achieving such goals is fully in line with our raison d’être. Congrats to all our gifted Atos colleagues who contribute to our success!”
Inventors competed in 4 categories
And the winners are …
Winner: Asthma management solution
Dr. Natalia Jimenez Lozano
Breathe easy and take control
Asthma is a chronic, potentially dangerous, condition that affects millions of people globally and is also a significant burden to patients and society. While today’s healthcare facilities treat patients by managing symptoms and reducing risks, the treatment is highly individualized and difficult to scale. Can technology be used to treat asthma more effectively and efficiently for a large population? How can technology help in better self-management?
The answer to these questions is the new asthma management solution, winner of the 2022 Atos Inventors’ Award in the Newcomers category for first-time patent filers. This innovative offering combines IoT, AI and big data analytics to manage and prevent asthma outbreaks for patients, helping them take charge of their self-care.
The personalized mobile solution leverages Atos’s digital expertise to access and manage data from different sources — electronic medical records (EMRs) from healthcare providers, real-time meteorological and environmental reports, smart devices and spirometer readings. This data is fed into an artificial intelligence system which integrates and personalizes it. The platform can provide warnings of exacerbated asthma symptoms, accompanied by doctor recommendations, as well as advice such as wearing a mask, outdoor excursions and any additional tests needed. The solution goes a long way in advancing self-care management and empowering patients to make informed decisions about their travel plans, outdoor activities and more, all available in a convenient mobile app.
The story behind this solution is closer to the heart that we can imagine; the inventor is an asthma patient herself. She has worked with Atos for ten fulfilling years and is proud of the way Atos delivers interoperability, merging technology and healthcare to deliver benefits for both, and always championing the cause of the end-user — the patient.
Winner: Method for detection of lateral movement of malware
Vinod Vasudevan, Harshvardhan Parmar and Ravi Raman
How to respond when things go sideways
Cyberattacks have grown significantly in recent years, and as businesses digitize their operations, attackers are developing increasingly sophisticated ways to attack your assets. The most common tactic used by adversaries today is lateral movement, which enables them to access high-value assets and sensitive data networks.
While tools such as security incident and event management (SIEM) can normalize and correlate data, they are better suited to detecting clear cyberattacks based on rules and signatures. Today’s deeper attacks involve attackers moving laterally after compromising an endpoint, which can often go undetected and requires a different response.
In just six months, a group of Atos cybersecurity professionals tackled the issue and created a methodology for spotting anomalies related to lateral movement using data science and behavioral analytics. Their approach has since been incorporated into Atos’s AIsaac platform, which combines our award-winning artificial intelligence (AI) for cybersecurity and innovations in edge AI to provide a comprehensive managed detection and response (MDR) solution.
The core logic of this method analyzes large volumes of network data logs to profile the normal patterns and flow of network traffic. This profile is then used as a threshold to detect abnormal behavior — for example, the sudden communication between two previously non-communicating servers under the influence of malware. Model thresholds or baselines are created by continuously training an AI model. Any deviation from the baseline triggers a rapid investigation to determine if an actual attack is taking place and takes action to counter it. Atos has used this method to successfully detect several lateral attacks between the development environment and a production website, which traditional security technologies like anti-malware and next-generation firewalls have missed.
Today, the AIsaac platform is helping more than 150 clients around the world orchestrate swift and effective responses to malicious and complex threats. As the threat landscape evolves, integrated platforms like AIsaac will become even more critical to ensuring total digital security.
Winner: Solution to suspect pilferage of fuel at fuel stations
Redefining “consumer comes first” with this zero-pilferage loss solution
Fuel stations are critical drivers in fueling the growth of any economy. There are more than 85,000 fuel stations across India and they need to deliver consistent and on-demand consumer-centric services. They need to ensure accurate deliveries and transactions.
Shortfall in deliveries may be caused by the wear and tear of delivery nozzles used to dispense the fuel, and/or malpractices at fuel stations. The reason can only be ascertained by physical inspections conducted by licensed professionals using calibrated jars at the fuel stations. These inspections are costly, labor intensive, and few and far between. Illegal manual tampering can not only cause great pilferage losses to the end-consumers but also jeopardize established brands and cost them dearly by way of penalty charges.
In recent years, many leading fuel stations have collaborated with digital partners to help them minimize pilferage losses and deliver seamless consumer transactions. However, they lack the precision of the Solution to Suspect Pilferage of Fuel at Fuel Stations — an Atos-powered pilferage loss solution.
The proposed solution attempts to eliminate three key variables that may impact the variance in fuel levels and cause differences during an inspection — the effervescent nature of the fuel, the wear and tear of the nozzles used, and the physical nature of the underground storage tanks. It proposes the use of third-party IoT sensors and monitoring apparatus attached to nozzles and storage tanks to track dispensation and inventory levels respectively.
The data from these IoT sensors is recorded by the Forecourt Controller (FCC), an Atos product, which shares this with the Atos team for analysis of potential shortfall or excess dispensations. In addition to the dispensation reports, the solution also shares half-hourly reports of the stock inventory. Aided by these reports, the oil and gas company can track fuel levels at all fuel stations efficiently and centrally. Leveraging this information, they can reduce the possibility of incorrect deliveries, ensuring fewer complaints from consumers and clamping down on possible penalties and litigation charges. Not only does this help manage our client’s brand image but also highlights and reports inaccurate deliveries for their review and follow-up actions.
The solution has already been successfully deployed at a leading oil marketing company in India.
Winner: Method for managing data life cycle and job scheduling in an HPC system
Ensuring data is in the right place (at the right time)
Data management is becoming increasingly critical in high-performance computing (HPC) infrastructures because data movement consumes time and a great deal of energy. Additionally, fast storage solutions are more expensive than their slower counterparts. Thus, most supercomputing centers implement multiple storage solutions, often broken down into tiers such as ultra-fast local storage, very fast buffer storage, fast parallel-access storage, slower high-capacity storage and finally, archiving or backup.
In most cases, these tiers are managed by rules that move data from one to another based on whether or not the data has been accessed for a certain period of time. An example of such a rule might be to move unused data to a slower tier after one month to free up space on the faster tier. While this scheme optimizes data placement, moving data back from a slower level to a faster level when it is needed again takes a significant amount of time and energy. In addition, it doesn’t consider job execution dynamics.
To handle this issue, an Atos software architect designed a scheduling logic that considers the data required for each HPC job, and in which tier this data should be stored for the job to begin. By defining an additional resource constraint, the jobs can be prioritized based on the data being used. This constraint helps to reduce data movements by indicating the current state of data use to the data movement solution.
By using this method, HPC users can not only avoid data movement if the desired data is already in the right place at the right time, but they can also anticipate data movement from one storage level to another in order to prioritize and schedule future jobs. The result is a highly efficient HPC infrastructure that operates more efficiently — dramatically reducing energy consumption and costly data movement without sacrificing job performance.
Jean-Olivier Gerphagnon (JOG)
Software Principal Architect, HPC, AI & Quantum
With more than 20 years of experience in technology, high performance computing (HPC), Linux, and security, Jean-Olivier is a distinguished expert at Atos. As a principal software architect, he manages the relationship between engineering, product management, and operations in terms of priorities and technological choices. He is deeply involved in the design of our next-generation software solution, including its sustainability and security.
In addition to supporting our customer relationships through BUX, he also promotes Atos at conferences and workshops. Several of his patents pertain to performance optimization, decarbonation and new HPC design approaches.