A Robust Malware Threat Hunting System and Method based on Deep Neural Networks in IoT environments

今天的工业公司甚至企业都使用物联网设备来实现流程自动化。这些设备暴露于恶意软件威胁，这是工业4.0时代的一个真实的挑战。由于不断发展的恶意软件威胁，目标公司需要为他们的恶意软件威胁狩猎程序支付安全补丁。这些补丁只适用于以前见过的威胁，对新的威胁不可靠，导致高误报率。所提出的系统可以捕获物联网恶意软件威胁，还可以从每个给定的可执行数据集生成adversarialpayload，以创建一个独特的adversarialpayload数据库。通过这种独特的方法，所提出的系统还能够防止伪造样本通过恶意软件威胁狩猎方法。所提出的系统还包括一个对抗性的有效载荷预防方法，以识别旨在绕过狩猎方法的假货（生成的样本）。根据我们最初的市场验证和文献研究，我们发现在现实世界的工业物联网（IIoT）环境中没有等效的机制。因此，所提出的方法可以通过自动化整个恶意软件预防，生成和狩猎过程来帮助IIoT节省时间和运营成本。

Today's industrial companies and even enterprises use IoT devices for automating theirprocesses. These devices are exposed to malware threats and this is a real challenge in theindustrial 4.0 era. Due to evolving malware threats, the targeted companies need to pay forsecurity patches for their malware threat hunting procedures. These patches only work withpreviously seen threats and are not reliable against novel ones, resulting in high false-positiverates. The proposed system can hunt IoT malware threats and can also generate adversarialpayloads from every given executable dataset to create a unique database of adversarialpayloads. With this unique method, the proposed system is also capable of preventing fakesamples from passing through malware threat hunting methods. The proposed system alsoincludes an adversarial payload prevention method to identify fakes (generated sample) that aimto bypass the hunting method. Based on our initial market validation and literature research wefound there is no equivalent mechanism in real-world Industrial Internet of Things (IIoT)environments. Therefore, the proposed method could help the IIoT in terms of saving time andoperational cost by automating the whole malware prevention, generation, and hunting process.