An integrated fuzzing and PPO-based deep reinforcement learning framework for firmware vulnerability detection in embedded IoT devices
The rapid expansion of Internet of Things (IoT)-enabled embedded devices and network-connected firmware environments has introduced significant cybersecurity challenges due to increasing firmware vulnerabilities and complex attack surfaces. Existing fuzzing approaches, such as American Fuzzy Lop (AFL++), primarily depend on predefined mutation scheduling and limited exploration strategies, which can reduce execution-path discovery efficiency and vulnerability detection…