Defense and Offence Strategies for Tactical Wireless Networks Using Recurrent Neural Networks

Abstract

Network forces attempt to dominate the spectrum in tactical wireless networks while preventing their opponents from holding any communication. In this context, both the defensive and the offensive capabilities of a network force are critical. Therefore, in this paper, we study the communication competition between two network forces, denoted as blue force (BF) and red force (RF), where each force is equipped with a transmitter, a receiver, and a friendly jammer. We propose defense and offense strategies for the BF using recurrent neural networks (RNN)s where the occupied channels by the RF nodes in the future time slots are predicted. Then the entity of each occupied channel in terms of RF users and jammer is identified. As a result, the BF transmitter finds unoccupied channels, and the BF jammer jams the RF transmitter's channels. The performance of the proposed method is evaluated by calculating both the ergodic rate (ER) of the BF receiver and RF receiver as well as the prediction accuracy of RF future channels. Simulation results show that the BF receiver attains ER of at least <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$90\%$</tex-math></inline-formula> of the maximum achievable ER and a detection accuracy higher than <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$85\%$</tex-math></inline-formula> for all the considered scenarios. Furthermore, the BF jammer can target the RF transmitter frequency channels and can reduce the RF receiver ER to less than <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$10\%$</tex-math></inline-formula> of its maximum achievable value.