Jamming Attack and Defense Based on Deep Learning in Cross-Technology Communication

With the increase of IoT devices, wireless communication protocols occupy a large amount of spectrum. Due to limited spectrum resources, researchers have proposed using software defined radio management, which enables Cross-technology communication (CTC) of different protocols within the same frequency band (such as 2.4GHz) through changes in physical layer signals without a hardware gateway, greatly reducing deployment costs and computational overhead. However, research on CTC security is not comprehensive enough. In order to study the security issues of CTC, we implemented continuous jamming attack, reactive jamming attack and jamming attack based on deep learning in CTC systems, and further studied corresponding jamming attack detection and defense measures. We conducted extensive experiments using USRP, constructed CTC scenarios, and introduced Channel State Infomation (CSI) as signal features for channel prediction, By using deep reinforcement learning algorithms to train CSI data and predict channel states, we can effectively defend against jamming attack signals and improve the performance of CTC.