WXWei Xia

Wei Xia

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Wei Xia has a total of 70 co-authors. Here are the top co-authors by number of shared publications:

  • Zishu He

    School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, China

    5 shared publications
  • Lisha Yu

    School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, China

    4 shared publications
  • Yuhan Zhang

    University of Electronic Science and Technology of China, Chengdu, China

    4 shared publications
  • Deming He

    School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, China

    3 shared publications
  • Guoqing Xia

    School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, China

    3 shared publications
  • Junbin Chen

    University of Electronic Science and Technology of China, Chengdu, China

    3 shared publications

Latest Publications

  • Resilient Distributed Extended Object Tracking Under FDI Attacks

    by Wei Xia, Kaidi Yang, Xuefei Feng, Mengqing Zhou

    We consider tracking an extended object modeled by an ellipse with an adversarial network, each node of which may be compromised by false data injection (FDI) attacks. We develop a resilient distributed extended object tracking algorithm with enhanced tracking accuracy, leveraging the diffusion strategy. Specifically, we develop an attack detection and defense mechanism such that each node could determine whether it is compromised or not, so as to restrain the propagation of malicious data owing to FDI attacks. We further propose a computationally efficient distributed extended object tracking algorithm for secure networks as a by-product. Moreover, we analyze the computational complexity of the proposed algorithms. Illustrative simulation results validate the efficacy of the proposed algorithm for adversarial networks under FDI attacks, especially under potentially time-varying attacks.

  • Resilient Distributed Estimation Against FDI Attacks: A Game-Theoretic Approach

    by Wei Xia, Mengqing Zhou, Yuhan Zhang

    False data injection (FDI) attack is one of the major security threats to distributed estimation for adaptive networks, and could bypass the conventional residual-based detection and thus cause catastrophic performance deterioration. In this work, we develop a resilient distributed game-theoretic estimation algorithm comprising the Bayes-based FDI attack detection, the non-cooperative game based decision-making and the resilient fusion strategy. Specifically, upon the detection of FDI attacks, we develop a distortionless instantaneous reward and distortionless long-term reward so as to facilitate the nodal decision-making on its action of whether sharing its intermediate estimates. We further derive a best response rule and benefit prediction scheme, such that each node could make its own independent decision on its current optimal action. Moreover, we develop a resilient fusion strategy to restrain the propagation of malicious data of each compromised node. Illustrative simulations validate the superior performance of the proposed algorithm for adversarial networks under FDI attacks.


  • 3-D Single Object Tracking With Temporal-Aware and Attention Mechanism

    by Betselot Yewulu Reta, Wei Xia

    3D Single object tracking (SOT) in point clouds is a computer vision task that involves localizing the target object in subsequent frames given the initial state. Most 3D SOT methods have only employed the current feature fusion map while overlooking the significance of temporal knowledge. However, it is crucial to incorporate temporal knowledge of the previous target cues into the current feature fusion map, especially in situations involving occlusion, significant appearance variations, and sparse point cloud representations of the target object. In addition, the performance of object tracking is highly dependent on a discriminative feature fusion map. To address these issues, we propose a 3D SOT by exploiting temporal knowledge and attention mechanisms. First, we design a Cascading Attentionbased Feature Fusion (CAFF) module to integrate the target cues into the search area, taking into account the intra-feature dependency within each feature and the inter-feature dependency between the template and search area features. Second, we present a Temporal Relation Modeling (TRM) module to refine the current feature fusion map by propagating temporal target cues from the previous feature fusion map. To further enhance the performance of the tracker, we present a template update sub-module to address the challenges posed by low-quality templates and appearance variations. The experimental results obtained from the KITTI, nuScenes, and Waymo Open Dataset (WOD) datasets demonstrate that the proposed approach achieves competitive performance compared to state-of-the-art methods. Our method runs at 67.57 FPS on a single RTX 2080 GPU.

  • Resilient Distributed Kalman Filtering Against Malicious Cyber Attacks

    by Wei Xia, Mengqing Zhou

    In this work, we consider the resilient distributed Kalman filtering for adversarial networks in the presence of different malicious cyber attacks, and develop a resilient distributed Kalman filtering algorithm to enhance the network estimation accuracy. Specifically, we develop an attack detection approach such that each node would distinguish its secure neighbor(s) from its compromised counterpart(s), and determine whether it is compromised or not. We further propose a resilient fusion strategy to restrain the propagation of malicious intermediate estimates of each compromised node. We also theoretically analyze the mean and mean-square stability of the proposed resilient distributed Kalman filtering algorithm, and develop an optimal reliable reference neighbor selection scheme, such that the network estimation accuracy could be further enhanced. Illustrative simulations validate the efficacy of the proposed algorithm for adversarial networks under different cyber attacks, even under potentially time-varying attacks.


  • Distributed Collaborative Feedback Particle Filter: A Diffusion Map-Based Feedback Gain Approximation Approach

    by Wei Xia, Zhuoyang Zhou, Wanyue Jiang

    In this paper, we consider the diffusion collaborative feedback particle filter (DFPF) for distributed nonlinear target tracking with an adaptive network consisting of geographically-distributed nodes. In the DFPF, each particle at each node is controlled by the collaborative feedback structure incorporating the innovation process and the feedback gain, which is the solution to the constrained Poisson equation. Utilizing the diffusion map (DM) approximation of the exact semigroup of the Poisson equation in the DFPF, we herein develop the feedback gain approximation, the performance of which is pertinent to the Gaussian kernel bandwidth. Further, based on the mean square error induced by the empirical approximation in the DM feedback gain, we derive the adaptive kernel bandwidth under the positive kernel bandwidth constraint. Moreover, we develop the variable integration step-size, referred to as the pseudo-time step-size in the DFPF, to efficiently solve the ordinary differential equation, which is utilized to describe the gradual transition from the prior to the posterior. Illustrative simulations validate that the proposed DFPF could achieve enhanced tracking performance with preferable computational efficiency.


  • Locational Detection of False Data Injection Attacks in the Edge Space via Hodge Graph Neural Network for Smart Grids

    by Wei Xia, Yan Li, Lisha Yu, Deming He

    Recently, the emerging False Data Injection Attacks (FDIAs), one of the major cybersecurity threats, have been severely threatening smart grids, since FDIAs could bypass conventional bad data detectors to disrupt power system operations. To maintain the security of power systems, it is critical to develop efficient locational detectors for FDIAs. However, designing FDIA detectors that could model the inherent underlying graph structures of smart grids and spatially correlated measurement data residing on both branches and buses such that FDIAs in the edge space could be detected and located, remains an open problem. In this work, we propose an alternative graph representation for smart grids, regarding both the power flows on branches and power injections on buses, such that we could simultaneously process these data in the edge space. We propose a Hodge Aggregation Graph Neural Network (AGNN)-based FDIA detector, leveraging the Hodge theory and exploiting the Hodge Laplacian into the AGNN. We further develop a Hodge Aggregation Graph Attention Network (AGAT)-based FDIA detector to enhance the locational detection performance of the Hodge AGNN-based detector, by utilizing the graph attention mechanism. Illustrative simulation results demonstrate the superior locational detection performance of the proposed detectors, compared to the other state-of-the-art FDIA detectors.


  • Locational Detection of False Data Injection Attacks in Smart Grids: A Graph Convolutional Attention Network Approach

    by Wei Xia, Deming He, Lisha Yu

    As a typical application supported by the Internet of Thing (IoT), smart grids, which are critical complex cyber–physical infrastructures, are facing increasing cybersecurity threats. One major substantial cybersecurity threat to smart grids is false data injection attacks (FDIAs), which could bypass bad data detectors so as to disrupt the operation of power grids. In this work, we consider the locational detection problem for FDIAs, namely, detecting the presence of FDIAs and locating the compromised buses, by utilizing smart grid data (either power injection measurements or system state estimates). Regarding smart grid data residing on the inherent underlying graph structures of power grids, we model smart grid data as non-Euclidean graph signals. We correspondingly develop the FDIA detector based on the graph convolutional attention network (GCAT) for the locational detection of FDIAs, by considering the underlying graph topology of a power grid. The proposed FDIA detector leverages the graph attention mechanism, which could elastically assign the graph shift operators in the GCAT, to enhance the locational detection performance. Integration of the proposed FDIA detector in a noninvasive way could endow existing power systems with the capability of locational detection of FDIAs. Illustrative simulation results demonstrate the superior locational detection performance of the proposed GCAT-based FDIA detector, compared to the other state-of-the-art FDIA detectors.


  • Dynamic UAV Swarm Confrontation: An Imitation Based on Mobile Adaptive Networks

    by Wei Xia, Zhuoyang Zhou, Wanyue Jiang, Yuhan Zhang

    Cooperative unmanned aerial vehicle (UAV) swarms could expand the mission capability of the single UAV and the overall combat effectiveness. We consider the dynamic scenarios where an UAV swarm confronts multiple maneuvering high-value targets (HVTs) and multiple maneuvering attackers. In such hostile scenarios, each UAV of the UAV swarm would generally strive to survive through collaboratively offending their opponents and self-defense from potential attacks. We consider modeling UAV swarms as biologically inspired mobile adaptive networks to imitate the dynamic confrontations of UAV swarms. It is essential to develop an efficient motion control mechanism underpinned by both target tracking and decision making so as to enhance the capability of UAV swarms. We formulate a distributed modularized framework incorporating collaboratively tracking multiple opponents, making successive and prompt decisions, and controlling motion mechanism for mobile adaptive networks. We develop the diffusion multitask interactive Kalman filter to efficiently track multiple preys (HVTs) and multiple predators (attackers). We further develop the collaborative decision making on prey pursuit or predator evasion such that each UAV would acquire early warnings and predictions of the most hypothetically threatening predators to enhance the effectiveness of self-defense. Moreover, we develop a composite motion control mechanism, such that each UAV would be steered toward the prey or away from any approaching predator in a self-organized manner at a variable speed. The modular design of the proposed framework would promote the affordability and adaptability of UAV swarms. Illustrative simulation results validate the efficacy of the proposed distributed framework and the underlying algorithms.


  • Widely Linear Null Broadening GNSS Anti-Interference for Polarization-Sensitive Arrays

    by Wei Xia, Yuhan Zhang, Hao Yu

    In this correspondence, we consider suppressing non-circular interferences for Global Navigation Satellite Systems (GNSS) receivers utilizing polarization sensitive arrays, regarding high dynamic scenarios. We model the widely linear array receive signal corrupted by potentially non-circular interferences. We analyze the second-order statistics of the widely linear receive signal and extend the covariance matrix taper (CMT) methodology to cope with the widely linear receive signal. We further develop widely linear null broadening algorithms to suppress high-dynamic non-circular interferences. Simulation results validate the superior performance of the proposed algorithms in suppressing high-dynamic non-circular interferences.