Speaker:Chenglin Miao

Exploring the Security Vulnerabilities of Intelligent IoT Systems

Abstract:

The Internet of Things (IoT) has emerged as a new networking paradigm that connects humans and the physical world through ubiquitous sensing, computing, and communicating devices. With numerous such connected devices having a large variety of on-board sensors, IoT systems can provide rich information about the surrounding world. To extract useful knowledge from such big sensory data and help people with decision making, researchers have developed various intelligent IoT systems using artificial intelligence (AI). Although these intelligent IoT systems can achieve outstanding performance, they still face many security challenges when they are applied in real world. In this talk, I would like to introduce my research efforts on exploring the security vulnerabilities of intelligent IoT systems under different types of malicious attacks. Specifically, I will first present a novel adversarial attack method based on which the attacker can use arbitrary objects to fool the deep learning-based LiDAR perception systems in autonomous driving. Then, I will introduce an intelligent data poisoning attack framework that can take down a crowd sensing system even with the reliability-aware data aggregation mechanism empowered.

Bio:

Dr. Chenglin Miao is an Assistant Professor in the Department of Computer Science at the University of Georgia. He received his Ph.D. from the State University of New York at Buffalo. His research interests include security and privacy, Internet of Things (IoT), and machine learning. He is especially interested in developing novel techniques for the security, privacy, and safety of emerging IoT systems and machine learning algorithms. His research work has been published in various top venues such as CCS, MobiCom, SenSys, MobiSys, MobiHoc, NeurIPS, KDD, AAAI, and IJCAI.