Modern network attacks are becoming stealthier and smarter. Attackers use encryption to cover up malicious traffic, which makes it really hard to detect. To solve this problem, this paper introduces a new model called Graph Convolutional Network with Multi-Head Attention (GCN-MHA). The goal of this model is to improve how we find and sort encrypted malicious traffic. First, we turn network traffic into a “graph”—this helps capture its structural and time-related features. Then, our GCN-MHA framework uses graph convolutional layers to learn spatial information. A multi-head attention mechanism helps it focus on the most important features. When tested on the ISCX-VPN2016 dataset, the model achieved an overall high accuracy of 98.79% and a recall rate of 99.24% under six categories of malicious traffic. We also performed cross-validation on two other datasets: USTC-TFC2016 and CIC-Darknet2020. These tests showed that the model has strong generalization ability on different data.

This paper proposes a Graph Attention based Convolutional Neural Network (GAT-CNN) for encrypted malicious traffic classification and detection, which improves the performance of network encrypted malicious traffic detection, recognition, and classification. By establishing a connectivity diagram between nodes, the spatiotemporal features in network traffic records are strengthened, providing a feature basis for intrusion detection. Using graph convolutional neural networks to dynamically aggregate and extract node behaviors in the graph, generate feature representations with category discrimination ability, and allocate different weights through multi head attention mechanism to achieve accurate classification of network intrusion behavior. To verify the effectiveness of the method, experiments were conducted on the ISCX-VPN encrypted traffic dataset, with a detection accuracy of 98.79% and a recall rate of 99.24%, demonstrating that the proposed method exhibits good performance in detecting malicious traffic.

Unmanned Aerial Vehicles (UAVs) in Flying Ad-Hoc Networks (FANETs) are widely used in both civilian and military fields, but they face severe security, trust, and privacy vulnerabilities due to their high mobility, dynamic topology, and open wireless channels. Existing security protocols for Mobile Ad-Hoc Networks (MANETs) cannot be directly applied to FANETs, as FANETs require lightweight, high real-time performance, and strong anonymity. The current FANETs security protocol cannot simultaneously meet the requirements of strong anonymity, high security, and low overhead in high dynamic and resource-constrained scenarios. To address these challenges, this paper proposes an Anonymous Authentication and Key Exchange Protocol (AAKE-OWA) for UAVs in FANETs based on One-Way Accumulators (OWA). During the UAV registration phase, the Key Management Center (KMC) generates an identity ticket for each UAV using OWA and transmits it securely to the UAV’s on-board tamper-proof module. In the key exchange phase, UAVs generate temporary authentication tickets with random numbers and compute the same session key leveraging the quasi-commutativity of OWA. For mutual anonymous authentication, UAVs encrypt random numbers with the session key and verify identities by comparing computed values with authentication values. Formal analysis using the Scyther tool confirms that the protocol resists identity spoofing, man-in-the-middle, and replay attacks. Through Burrows Abadi Needham (BAN) logic proof, it achieves mutual anonymity, prevents simulation and physical capture attacks, and ensures secure connectivity of 1. Experimental comparisons with existing protocols prove that the AAKE-OWA protocol has lower computational overhead, communication overhead, and storage overhead, making it more suitable for resource-constrained FANET scenarios. Performance comparison experiments show that, compared with other schemes, this scheme only requires 8 one-way accumulator operations and 4 symmetric encryption/decryption operations, with a total computational overhead as low as 2.3504 ms, a communication overhead of merely 1216 bits, and a storage overhead of 768 bits. We have achieved a reduction in computational costs from 6.3% to 90.3%, communication costs from 5.0% to 69.1%, and overall storage costs from 33% to 68% compared to existing solutions. It can meet the performance requirements of lightweight, real-time, and anonymity for unmanned aerial vehicles (UAVs) networks.