Frame-Selective Wireless Attack Using Deep-Learning-Based Length Prediction

Abstract

Wireless attack refers to the malicious activity to generate a wireless jamming signal to interfere with the data transmission of legitimate users. If the jamming duration of a wireless attack is long, it can be easily detected; such attacks also consume more energy for generating the jamming signal. We propose a frame-selective jamming to attack shorter frames that are essential to data communication protocols such as media access control (MAC) acknowledgement frames. Once a wireless signal is detected, the proposed jammer predicts the duration of the signal using a deep learning technique and generates a jamming signal selectively if the duration is expected to be shorter than or equal to a certain threshold.