Accurate and Real-time Hardware-assisted Detection of Cyber Attacks

Develop a hardware-assisted real-time accurate detector of cyber-attacks on networked and edge electronic devices. An increasing number of network-connected devices and systems in modern-day life are vulnerable to many attacks. Beyond the traditional computing systems and cloud services, modern Internet-of-Things (IoT) and cyber-physical systems can experience numerous cyber-attacks, such as ransomware, spyware, spoofing, botnets, keyloggers, denial of service (DoS), and distributed denial of service (DDoS), each of which is becoming more prevailing by numbers, as well as more challenging to thwart. There is an on-going need for effective solutions to identify, report and protect against cyber-threats. Current protection techniques are limited in detection efficacy (~70%) and scalability issues. Most techniques are primarily based-upon static software-focused solutions such as code analysis and signature (template) matching. These techniques have proven to be limited in detection efficacy so far, as reflected by the increasing number of threats and compromised cases. This topic is seeking solutions to analyze hardware generated data that would enable real-time, precise detection (>95%) and proactive protection against cyber-threats. The end state of this effort is a device-embedded solution to support highly accurate, real-time (within fraction of seconds) detection of critical cyber-threats, such as crypto-ransomware and DDoS attacks, on networked and edge electronic devices, such as computers, servers, cyber-physical systems, and IoT devices with minimal performance overhead while offering multi-layer and distributed defense, monitoring anomalous behaviors against zero-day attacks, and engaging automatic protection without human intervention.