Speaker:Amit Kumar Sikder

Title

Sensing Trouble: Leveraging System Modeling to Safeguard Smart Devices Against Emerging Threats

Abstract

The prevalence of smart devices, including smartphones, smart watches, and smart home devices, has become widespread in our daily lives. These devices, equipped with high-precision sensors, seamless inter-device communication, and artificial intelligence, can interact with each other and the physical world, performing various tasks and making autonomous decisions. Despite the success of existing computer system security tools, there exists a fundamental gap in understanding the dynamics of smart devices, especially concerning sensor data sharing, communication among components and apps, and their collective task execution capabilities. This gap provides an opportunity for attackers to exploit the integration of AI, networking, and sensor technologies in smart devices, easily bypassing deployed security measures. In this talk, I will discuss emerging threats that target sensor channels, insecure device communications, and unauthorized device access to launch different attacks on smart devices. I will share insights from my research, focusing on novel system modeling to address these threats. The key insight is that sensors, devices, and apps within smart systems collectively change their states and data-sharing patterns, unveiling ongoing tasks and dynamic information flow between devices and apps. I will discuss how this inter-component relation can be harnessed to model benign smart device behavior and identify vulnerabilities. Further, I will demonstrate how vulnerabilities in smart device apps and implementation flaws in communication protocols can be reverse-engineered through forensic analysis. This process enables the crafting of effective mitigation strategies to safeguard smart devices and applications against future attacks.

About Amit Kumar Sikder

Dr. Amit Kumar Sikder is a research scientist in the School of Cybersecurity and Privacy at Georgia Institute of Technology. He completed his Ph.D. in Electrical and Computer Engineering from Florida International University (FIU) in 2020. His research interests lie in the intersection of computer system forensics, artificial intelligence, and program analysis. His research has developed novel techniques to model smart device, system, and app behaviors during task execution (AI-assisted, user-initiated, and dynamic tasks) by correlating intra and inter-device data-sharing patterns, program analysis, and system forensics. He has published over 20 papers in top-tier security conferences, including USENIX Security, ACM CCS, NDSS, and ACSAC. His research has received several awards, including UGS Provost Award for Outstanding Creative Project and Dissertation Year Fellowship at FIU. His research has been featured in several national and international media outlets, including ACM Tech News, The Register, PBS News, NBC News, and Science News.