This text is part of the special Research section
Computer hackers (the famous hackers, in English) are deploying increasingly sophisticated strategies for their cyberattacks. A research consortium of which the École de technologie supérieure (ETS) is a member uses artificial intelligence to make next-generation mobile telecommunications networks, such as 5G, more secure.
Led by the University of Waterloo, Ontario, the consortium also includes specialists from the University of Regina, Saskatchewan, Ontario companies BlackBerry and Rockport Networks and Montreal’s NoviFlow. “This three-year project was launched in July 2022 by the Department of National Defence, which is supporting it financially with a $1.5 million grant,” says Rami Langar, professor-researcher in the Department of Software Engineering and ETS Information Technology. The telecom operator Rogers is also taking part in research through the 5G platform it has set up on the Waterloo campus in partnership with the University.
Simulate attacks
Hacker attacks present major risks for critical infrastructures, such as telecommunications networks, hospitals or administrations. Mobile networks from 5G (and the future 6G in 2030) are vulnerable to it. ” Antennas [noeuds] of these networks are all virtualized and “softwareized”, and the data is uploaded to the cloud (the clouds). This addition of a touch of virtualization gives attackers new possibilities to reach their targets, which is why it is very important to protect this type of network,” explains Rami Langar. Consortium industrialists working in the cloud, such as Rockport Network and NoviFlow, are providing equipment for the project.
The researcher’s team and its partners target several attacks while trying to simulate them. “At the ETS, we are working on well-known hacking techniques, such as ‘smart jamming’ or the distributed denial of service attack, called ‘DDoS attack’ (for Distributed Denial of Service)”, explains Rami Langar. From several sources (hence the term “distributed”), this strategy aims to make a service unavailable or to prevent its legitimate users from using it.
“We are simulating these attacks on a small 5G network platform that we have set up at ETS. Attackers are very smart; we try to understand their behavior and deploy certain tools, such as the security strategy by deception”, explains the researcher. The latter aims to deploy digital decoys in network infrastructures to slow down, trap and detect attackers.
Outwit the digital enemy
Artificial intelligence allows the consortium to detect and simulate these attacks and react to them more quickly and dynamically. “We are deploying our machine learning model, which is based on representative datasets of user behaviors in a 5G network. They were not available until then for reasons of confidentiality”, specifies Rami Langar. The platforms installed at the University of Waterloo by Rogers and at the ETS make it possible to collect this data. “Our goal is to learn how to prevent an attack by detecting it before it occurs,” says the researcher.
After a few months, the consortium already has promising preliminary results on the best-known DDoS attack. “It makes it possible to target a specific node in the network by sending what are called successive requests in parallel, so that this node can no longer respond to other requests from users. By simulating it, we were first able to detect it in a few seconds, then once the model was trained, in a few milliseconds,” says Rami Langar. The research team was also able to counter this attack by using the malicious users in a low-capacity virtual network to protect the network.
Its solution based on deep learning, by comparing in particular the time required and the detection accuracy of several models, has already been the subject of a publication noting an accuracy of nearly 97% in the detection of DDoS attacks. Other innovations, by the end of the project in 2025, could further improve the performance, flexibility and reliability of telecommunications networks.
This special content was produced by the Special Publications team of the Duty, relating to marketing. The drafting of Duty did not take part.