Programmable & Data Driven Systems

Using Machine Learning to Secure IoT Devices

IoT devices such as wearables, voice assistants, and home appliances are becoming an integral part of our lives. However, these devices still represent a security and privacy risk, with large-scale coordinated attacks often populating the news. For example, the Mirai botnet successfully compromised more than 100K IoT devices and engaged them in coordinated DDoS attacks over the last few years. One approach to securing IoT devices is to fingerprint them, i.e., identifying the device type, manufacturer, or event through different forms of traffic analysis. Ultimately, network administrators can use this information to quickly react to threats and/or take preventive measures. In this project, we leverage emerging machine learning techniques to protect IoT devices against security and privacy attacks. On one hand, we seek developing high-speed device fingerprinting appliances that can identify gadgets on-the-fly even on next-generation multi-hundred gigabit networks. On the other hand, we also consider protecting IoT devices against fingerprinting-based attacks from malicious actors (e.g., a man-in-the-middle) by carefully obfuscating their traffic patterns.


Publications

PoirIoT: Fingerprinting IoT Devices at Tbps Scale
Carson Kuzniar, Israat Haque
IEEE/ACM Transactions on Networking 2024

[Paper]

IoT Device Fingerprinting on Commodity Switches
Carson Kuzniar, Miguel Neves, Vladimir Gurevich, Israat Haque
IEEE/IFIP Network Operations and Management Symposium (NOMS) 2022

[Paper]

People

  • Israat Haque, Dalhousie University, israat at dal dot ca
  • Carson Kuzniar, Dalhousie University, carson.kuzniar at dal dot ca
  • Han Yang, Dalhousie University, hn252486 at dal dot ca
  • Miguel Neves (Past Member), Dalhousie University, mg478789 at dal dot ca
  • Pulkit Garg (Past Member), Dalhousie University, pl201920 at dal dot ca
  • Weiye Liang (Past Member), Dalhousie University, wy875846 at dal dot ca