A Real-time Hardware-assisted Endpoint Threat Detection Framework

IoT devices expose greater security risks due to their limited resources, lack of security functions and maintenance. Existing software-based solutions such as static code analysis and signature matching cannot meet the stringent requirements for precision, adaptability, speed, and scalability. Our goal is to design and develop a multi-level hardware-assisted threat detection system. The solution will integrate the real-time hardware performance counter (HPC) sensors with advanced machine learning algorithms to detect various cyber-threats rapidly (in faction of seconds) and accurately (>95%). The key innovation of the solution include: (1) high detection accuracy for various cyber-attacks such as ransomware, botnets, denial of service (DoS), and distributed denial of service (DDoS); (2) real-time performance to rapidly protect mission critical systems; (3) low resource utilization on IoT systems; and (4) high scalability to support IoT systems with heterogeneous platforms and large number of devices. Commercial applications for this solution include IoT system security for smart cameras, smart grid, transportation, telemedicine systems, etc.