- Radio Astronomy Observations and Technology
- Orbital Angular Momentum in Optics
- Adversarial Robustness in Machine Learning
- Laser-Plasma Interactions and Diagnostics
- Sparse and Compressive Sensing Techniques
- Mobile Agent-Based Network Management
- Wireless Networks and Protocols
- Anomaly Detection Techniques and Applications
- IPv6, Mobility, Handover, Networks, Security
- Digital Media Forensic Detection
Xi’an Jiaotong-Liverpool University
2021-2022
Chinese People's Armed Police Force Engineering University
2022
Xi’an University
2021
Since the radar echo of space target satisfies sparseness, compressed sensing (CS) theory can be used to reconstruct information with data much lower than Nyquist sampling rate. For vortex electromagnetic wave (VEMW) radar, since azimuth also use CS significantly reduce requirement resolution on amount mode data. Based above analysis, this paper proposes a CS-based VEMW imaging sequence optimization method. Firstly, model is deduced, and beam steering method performed adjust direction...
Abstract Vortex electromagnetic waves (VEMWs) carrying orbital angular momentum (OAM) are promising to benefit the higher degree of freedom for inverse synthetic aperture radar (ISAR) imaging on account their helical wavefront. For high‐speed targets, ‘stop and go’ hypothesis is no longer valid. Thus, intrapulse movement target will cause broadening shift one‐dimensional range profile blurring OAM eigenvalues coupling azimuth. This contribution proposes a method VEMW ISAR. First, an...
With the continuous development of network technology, heterogeneity is one main characteristics future wireless networks. How to provide best connection for end users considered an important point. This paper, we designed a selection optimization algorithm based on AHP and TOPSIS possible mobile devices improve system's performance. Through supplying necessary parameters conditions, can automatically judge which current task. will be helpful increase their efficiency.
When a neural network is trained with data set from an untrusted source, attacker can insert poisoned backdoor trigger into the to make wrong decisions. By using Activation Clustering over convolutional networks, we propose improved method for defensing attacks in process of collection and preparation. Experimental results show that this reliably protect networks interference malicious during training. The essence making learn feature classify toxic separate class. structure existing model...