With the rapid development of the Internet of Things (IoT), device identification has become a critical research direction for IoT security. Traditional two-stage device identification methods based on Internet fingerprints suffer from low identification accuracy, high model redundancy, and significant time complexity. To address these issues, this paper proposes a method for IoT device identification based on probability adaptive boosting, aimed at improving identification accuracy and reducing time complexity. This method constructs new fingerprint features by analyzing network protocol fields and traffic behavior patterns, employing the adaptive boosting algorithm to learn traffic behavior patterns, and incorporating traffic perturbations to enhance performance. Experimental results demonstrate that the proposed method significantly improves the accuracy and efficiency of device identification, holding practical application value and better meeting the device detection demands in the current network environment.
The integrated sensing and communication (ISAC) system is a key research direction for 6G, where it faces multiple challenges in both communication and radar functionalities. In this paper, we investigate the near-far effect (FNE) problem in an orthogonal frequency division multiplexing (OFDM)-based ISAC system. Firstly, a cooperative ISAC prototype platform is built to confirm the specific impact of FNE. Based on the experimental results, it is found that the FNE is caused by the main and side lobes of a target with strong power in 2D-fast Fourier transform (2D-FFT) matrix when the received power is not saturated. It can be suppressed using environmental clutter elimination and moving target detection (MTD). In addition, we model the FNE problem as a 2D sinc function and derive a theoretical FNE threshold. The ideal theoretical FNE threshold is 13 dB when considering only the first side lobe, and the measured FNE threshold is about 8 dB. This difference is attributed to the received bottom noise, signal randomness and environmental factors.
With the rapid development of internet of things (IoT) technology, its applications in smart transportation and other scenarios are becoming increasingly widespread. Traditional anomaly detection algorithms have not deeply explored IoT traffic information and have certain limitations. Therefore, this paper proposes an anomaly detection and attack type identification algorithm based on traffic characteristics for IoT. The algorithm first analyzes traffic patterns on real IoT data and extracts key information for feature construction. In the anomaly detection phase, four features are constructed, utilizing the density-based spatial clustering of applications with noise (DBSCAN) clustering algorithm with Manhattan distance as a metric. Finally, in the attack type identification phase, multiple features are generated based on the key information, employing a joint algorithm of Principal Component Analysis and the C4.5 decision tree. Experimental results indicate that the features extracted from both phases have a certain degree of accuracy in the designed algorithm, providing practical technical support for anomaly detection and attack identification, and possessing theoretical significance and reference value.
With the rapid advancement of unmanned aerial vehicle (UAV) technology, UAVs are widely used in surveying and mapping, geographic information systems, agricultural protection, security, emergency response, logistics, and transportation. They play an increasingly significant role, particularly in unmanned combat operations. The increasing technical requirements for autonomy, intelligence, efficiency, and collaboration in UAV missions have accelerated the shift toward multi-UAV operations. As one of the key technologies enabling collaborative missions, distributed data link adaptive networking technology must be enhanced to improve the capability of intelligent collaborative operations in manned/unmanned systems. Therefore, to address the sudden changes in information volume during multi-UAV missions and meet the demands of dynamically varying service data during transmission and relay among nodes in a collaborative network data link, this paper proposes a resource allocation method based on service requirements. This approach ensures quality of service (QoS) for multi-node and multi-service concurrency under limited bandwidth resources. The proposed method significantly improves bandwidth utilization while maintaining low algorithmic complexity, making it suitable for practical engineering applications.
Unmanned aerial vehicle and intelligent reflecting surface (UAV-IRS)-assisted device-to-device (D2D) communication enhances coverage and improves channel reconfigurability, but it still faces significant challenges in maintaining covert transmission in the presence of multiple randomly distributed eavesdroppers. This paper studies covert communication in spectrum-sharing D2D networks assisted by UAV-IRS. Unlike existing works that focus on single or location-known eavesdroppers, this paper addresses practical scenarios with multiple spatially random eavesdroppers. We propose a novel system model, where the UAV-IRS dynamically reconfigures the wireless environment and a full-duplex base station generates artificial noise to enhance covertness. The minimum detection error probability at the eavesdroppers is adopted as the covertness metric, and a comprehensive joint optimization problem is formulated to maximize the covert transmission rate while ensuring the reliability of the cellular link. To tackle the non-convexity, a hierarchical alternating optimization (AO) algorithm is developed to obtain closed-form solutions for IRS phase shifts and UAV positioning, and to efficiently allocate power under covertness and outage constraints. Numerous analyses demonstrate that the proposed scheme outperforms baseline methods, validating the effectiveness of joint UAV and IRS optimization in enhancing both covertness and covert rate under multiple eavesdroppers environments.
