Title: Defeating Super-Reactive Jammers With Deception Strategy: Modeling, Signal Detection, and Performance Analysis

Abstract: This paper develops a novel framework to defeat a super-reactive jammer, one of the most difficult jamming attacks to deal with in practice. Specifically, the jammer has an unlimited power budget and is equipped with the self-interference suppression capability to simultaneously attack and listen to the transmitter's activities. Consequently, dealing with super-reactive jammers is very challenging. Thus, we introduce a smart deception mechanism to attract the jammer to continuously attack the channel and then leverage jamming signals to transmit data based on the ambient backscatter communication technology. To detect the backscattered signals, the maximum likelihood detector can be adopted. However, this method is notorious for its high computational complexity and requires the model of the current propagation environment as well as channel state information. Hence, we propose a deep learning-based detector that can dynamically adapt to any channels and noise distributions. With a Long Short-Term Memory network, our detector can learn the received signals' dependencies to achieve a performance close to that of the optimal maximum likelihood detector. Through simulation and theoretical results, we demonstrate that with our approaches, the more power the jammer uses to attack the channel, the better bit error rate performance the transmitter can achieve.