Deep Learning-Based Dynamic Watermarking for Secure Signal Authentication in the Internet of Things

摘要: Securing the Internet of Things (IoT) is a necessary milestone toward expediting deployment its applications and services. In particular, functionality IoT devices extremely dependent on reliability their message transmission. Cyber attacks such as data injection, eavesdropping, man-in-the-middle threats present major security challenges. against requires accounting for stringent computational power need low-latency operations. this paper, novel deep learning method proposed to detect cyber via dynamic watermarking signals. The framework, based long short-term memory (LSTM) structure, enables extract set stochastic features from generated signal dynamically watermark these into signal. This IoT's cloud center, which collects signals devices, effectively authenticate Furthermore, prevents complicated attack scenarios eavesdropping in attacker aims break algorithm. Simulation results show that, with an detection delay under 1 second, messages can be transmitted almost 100% reliability.