Secure multicast routing protocol for FANETs against active sniffing attacks: Deep learning approach

초록

This paper proposes a deep learning-based secure multicast routing (SMRDL) protocol to enhance security in flying ad-hoc networks (FANETs) against active sniffing attacks. In order to improve the route stability and security, a cross-layer cost maximization problem is formulated, incorporating physical and network layer metrics such as hop count, remaining energy, and secrecy spectral efficiency (SSE) at each transmission hop. To support real-time optimization, a deep learning framework is designed to predict the optimal next-hop node based on signal-to-interference-plus-noise ratio (SINR), hop count, and residual energy. By integrating cross-layer information with a deep neural network (DNN), the SMRDL protocol adaptively selects secure and efficient routes in dynamic network conditions. Simulation results demonstrate that the proposed SMRDL protocol significantly outperforms existing protocols in terms of packet delivery ratio (PDR), SSE, control overhead, and routing delay under varying node speeds and network scales. Copyright © 2026. Published by Elsevier B.V.

키워드