IoT Device Authentication Using Adversarial Machine Learning
Anshul Goel -, Ashwin Sharma -, Deepak Kumar Kejriwal · Journal of Advances in Developmental Research · 2024
The security of IoT networks depends on correctly authenticating devices to allow safe communication and stop unauthorized devices. The enemy of machine learning can break past security measures to enter unauthorized systems. Raising security levels demands the creation of authentication tools that work beyond manipulation from threatening entities specifically through body traits or user actions. This research looks into how incorporating advanced anti-attack methods in machine learning would improve system security from serious hacking threats. Our system creates an authentication tool made of adversarial machine learning methods to find abnormal device actions during verification steps. Our authentication system training processes immune defense by learning from both normal user and adversary actions. Our research tests multiple machine learning methods especially GANs and SVMs to understand their ability in detecting legitimate devices from security threats. Our early findings show that adversarial machine learning makes IoT network security better by better finding unauthorized users. This study explains the safety advantages of integrating adversarial machine learning with present authentication systems. These models show how to spot risks instantly through their flexibility for adjusting to new security dangers. Our study demonstrates that using adaptive computer learning strengthens both IoT device login security and total IoT security defense systems. This research helps develop better IoT security by showing how device authentication should be upgraded to combat new cyber threats.