Grant: Discovery Launch Supplement
École Polytechnique de Montréal
$12,500.00
Robust, Resilient, and Trustworthy Federated Learning for Intrusion Detection in the Internet of Things
FY 2024-25
Summary
The Internet of Things (IoT) has revolutionized our interaction with technology by fundamentally changing how we collect and process data in real-time. This monumental leap forward has profoundly influenced various industries, including those involving connected medical devices, self-driving vehicles, and the development of smart city applications. The large amount of data generated by IoT devices, their limited resources, and their heterogeneous nature have increased their vulnerability to attacks. Intrusion Detection Systems (IDSs) play an important role in detecting malicious activities in the IoT environment. Most existing IDSs are powered by Machine Learning (ML) models to support the detection of unseen or zero-day attacks. ML models are trained on data shared by IoT devices. However, the current trend in IoT solutions aims to minimize data sharing, primarily for two reasons: safeguarding data privacy and reducing the overhead costs associated with transmitting large data volumes. Consequently, the adoption of Federated Learning (FL) has emerged as a novel paradigm to address this issue. In FL, ML models are trained locally on IoT devices, with only model weights aggregated via a central server to create a tailored global model for a specific application. Therefore, aligning with this prevailing trend has become imperative, leading to the design of IDS for IoT powered by FL.
Despite the importance of FL in IoT systems, three challenges must be addressed to enhance its performance, ensure security, and promote its acceptance and practical suitability. The first challenge is noisy IoT data, arising from disturbances incurred within the IoT system, which can degrade the performance of the local ML model, thereby undermining the effectiveness of the global detection system. The second challenge involves the poisoning of data, which can be introduced by malicious or compromised IoT devices during the FL training process. The third challenge is establishing trust in FL-powered IDS, where they should provide interpretable results to instill trust in users.
My long-term research objective is to design and develop a robust, resilient, and trustworthy IDS for IoT systems. To attain this goal, the proposed research program will address the following three challenges: First, we will design and develop robust FL for detecting intrusions under a noisy IoT environment. Second, we will design and develop a resilient FL system that can operate effectively in the presence of compromised or malicious IoT devices. Third, we will create an explainable FL approach that fosters intrusion detection with clear and correct interpretations.
The proposed research program is of paramount importance for the economic development of Information Technology societies within Canada. This holds true for citizens/users, governments, public and private sectors, and infrastructure providers, all of whom benefit from more secure IoT applications.
Program
Discovery Launch Supplement
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
DGECR-2024-00177
Name nom
Abusitta, Adel
Department département
Computer Science
Organization id
64
Institution établissement
École Polytechnique de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
12500
Program id
DGECR
Program name en
Discovery Launch Supplement
Program name fr
Supplément Tremplin vers la découverte
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1507
Committee name en
Computer Science
Committee name fr
Sciences informatiques
Area of application code
1207
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Information, computer and communication technologies
Area of application fr
Technologie de l'information, des ordinateurs et des réseaux de communications
Research subject code
2706
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Software engineering
Research subject fr
Génie logiciel
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Robust, Resilient, and Trustworthy Federated Learning for Intrusion Detection in the Internet of Things
Keywords
network security; cybersecurity; internet of things security; resilient applied machine learning; poisoning attacks in federated learning; explainable applied machine learning; robust applied machine learning; federated learning; algorithmic game theory; data securty
Application summary
The Internet of Things (IoT) has revolutionized our interaction with technology by fundamentally changing how we collect and process data in real-time. This monumental leap forward has profoundly influenced various industries, including those involving connected medical devices, self-driving vehicles, and the development of smart city applications. The large amount of data generated by IoT devices, their limited resources, and their heterogeneous nature have increased their vulnerability to attacks. Intrusion Detection Systems (IDSs) play an important role in detecting malicious activities in the IoT environment. Most existing IDSs are powered by Machine Learning (ML) models to support the detection of unseen or zero-day attacks. ML models are trained on data shared by IoT devices. However, the current trend in IoT solutions aims to minimize data sharing, primarily for two reasons: safeguarding data privacy and reducing the overhead costs associated with transmitting large data volumes. Consequently, the adoption of Federated Learning (FL) has emerged as a novel paradigm to address this issue. In FL, ML models are trained locally on IoT devices, with only model weights aggregated via a central server to create a tailored global model for a specific application. Therefore, aligning with this prevailing trend has become imperative, leading to the design of IDS for IoT powered by FL.
Despite the importance of FL in IoT systems, three challenges must be addressed to enhance its performance, ensure security, and promote its acceptance and practical suitability. The first challenge is noisy IoT data, arising from disturbances incurred within the IoT system, which can degrade the performance of the local ML model, thereby undermining the effectiveness of the global detection system. The second challenge involves the poisoning of data, which can be introduced by malicious or compromised IoT devices during the FL training process. The third challenge is establishing trust in FL-powered IDS, where they should provide interpretable results to instill trust in users.
My long-term research objective is to design and develop a robust, resilient, and trustworthy IDS for IoT systems. To attain this goal, the proposed research program will address the following three challenges: First, we will design and develop robust FL for detecting intrusions under a noisy IoT environment. Second, we will design and develop a resilient FL system that can operate effectively in the presence of compromised or malicious IoT devices. Third, we will create an explainable FL approach that fosters intrusion detection with clear and correct interpretations.
The proposed research program is of paramount importance for the economic development of Information Technology societies within Canada. This holds true for citizens/users, governments, public and private sectors, and infrastructure providers, all of whom benefit from more secure IoT applications.