WWWeidong Wang

Weidong Wang

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Weidong Wang has a total of 91 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Efficient DOA Estimation via Finite-Precision MUSIC Algorithm

    by Yiming Fang, Li Chen, Ang Chen, Weidong Wang

    Efficient estimation of direction-of-arrival (DOA) is vital in wireless communication and sensing systems. In this paper, we propose a DOA estimation method derived from multiple signal classification (MUSIC), which incorporates finite-precision arithmetic. First, we analyze the computational bottlenecks of the classic low-complexity randomized unitary based MUSIC, formulating this computational issue as an inner product problem. Then, a mixed-precision method is introduced to address this problem. Specifically, this method partitions summations in inner products into blocks, where intra-block computations use low-precision arithmetic and inter-block sums use high-precision arithmetic. To further improve computational accuracy, we develop an adaptive-precision method that supports adaptive block sizes and multiple precision levels. Finally, simulation results show that the proposed finite-precision MUSIC design achieves DOA estimation performance similar to that using full-precision arithmetic while reducing more than 50% computational cost.

  • Model-Driven Radio Frequency Fingerprint Identification Based on Hardware Impairments

    by Tianyu Yang, Yijie Guo, Li Chen, Weidong Wang

    The proliferation of the Internet of Things (IoT) necessitates advanced physical layer authentication (PLA) techniques. While radio frequency fingerprint identification (RFFI) leverages hardware impairments for device authentication, existing data-driven approaches face challenges such as heavy data dependency, limited interpretability, and sensitivity to channel variations. To address these issues, this paper proposes a model-driven RFFI system based on hardware impairments, which integrates a radio frequency fingerprint (RFF) extractor with a support vector machine (SVM) classifier. Inspired by the LS parameter estimation technique, the RFF extractor associates the physical model with hardware characteristics, constructing an interpretable RFF database. Compared to the benchmark scheme, our proposed scheme achieves higher classification accuracy with limited training samples and exhibits strong robustness against channel effects. Experimental results validate the superiority of system performance in device classification and anomaly detection. Simulation results demonstrate that the proposed RFFI system significantly reduces data dependency while maintaining reliability in dynamic IoT environments.

  • Open Access

    Generalization-Enhanced Channel Estimation Through Adaptive Interpolation and Multi-Task Learning-Based Denoising Network

    by Bolin Wang, Li Chen, Xiaohui Chen, Weidong Wang

    Accurate CSI estimation with low pilots is desirable for the multiple-input multiple-output orthogonal frequency-division multiplexing (MIMO-OFDM) system. In the existing channel estimation methods, both interpolation and denoising suffer from the problem of generalization. In this paper, we propose an adaptive interpolation and multi-task learning denoising network for generalization-enhanced CSI estimation. First, we model the wireless channel based on Gaussian process (GP) and use Bayesian optimization (BO) to find the optimal parameters of the Mat´ern kernel for interpolation. For each matrix, we can find the most suitable parameters of the kernel to achieve precise interpolation adaptively. Then, we design the multi-task residual network (MT-Net) based on multi-task learning. In MT-Net, shared layers are employed to utilize the relevant information between multiple tasks. And task-specific layers are also designed to extract the characteristics of each task. Compared to single-task learning, MT-Net can achieve information sharing between multiple tasks to enhance the scenario generalization of the model. Simulation results show that when the application scenario changes, our method exhibits a stronger generalization ability compared to other neural network-assisted methods.

  • CLT-NN MUSIC for Robust DoA Estimation: Distinguishing Eigenvalue Differences Between Signal and Noise

    by Pengkun Song, Weidong Wang, Li Chen

    The estimation of the signal source number presents a key challenge for the multiple signal classification (MUSIC) algorithm under low signal-to-noise ratio (SNR) conditions. However, previous methods have difficulty addressing low SNR scenarios. To solve this problem, this work proposes a CLT-NN MUSIC method, which provides a new strategy to distinguish between signal and noise eigenvalues. Specifically, based on the Central Limit Theorem (CLT), we establish the eigenvalue perturbation distribution model for the sample covariance matrix and derive the K-estimation criterion. Subsequently, we adopt neural networks (NN) to fit the aforementioned perturbation model. Through a dynamic decision mechanism, this approach enables the adaptive estimation of the number of sources under low SNR. Simulation results show that CLT-NN MUSIC significantly improves the estimation performance even under low SNR.


  • Depth and Penetration Imaging Based on Multipath IR-UWB Signals

    by Xinzhao Zhou, Li Chen, Huarui Yin, Xiaohui Chen, Weidong Wang

    Radio frequency (RF)-based imaging plays a pivotal role in sensing applications due to its robustness against visibility constraints and environmental factors. Among various RF imaging technologies, impulseradio ultra-wideband (IR-UWB) stands out by offering both fine temporal resolution and penetration capabilities. Existing work is limited to planar or depth imaging with multipath suppression. However, achieving depth imaging exploiting multipath IR-UWB signals and further realizing penetration imaging remains a challenge. To address it, this paper proposes a novel neural network architecture that uses IR-UWB signals to reconstruct depth and penetration images. First, we give a feasibility analysis of the dual imaging tasks using IR-UWB signals in an enclosed environment with multipath effects. Then, a theoretical spatial geometric inverse model is presented to formulate depth and penetration imaging as a joint optimization problem. Next, to address the ill-posed and intractable nature of the optimization problem, we propose a neural network architecture named the attention-enhanced depth and penetration imaging network (AEDPI-Net) as an approximate solver. AEDPI-Net employs a shared encoder to extract common features from IR-UWB signals, which are subsequently mapped by two parallel decoders into depth and penetration images. Finally, extensive experimental studies on an IR-UWB imaging platform verify the effectiveness of AEDPI-Net. The results reveal that AEDPI-Net achieves superior depth and penetration imaging performance compared to the baseline methods.

  • Open Access

    Communication Efficient Federated Learning With Quantization-Aware Training Design

    by Xiang Fang, Li Chen, Huarui Yin, Xiaohui Chen, Weidong Wang

    Model quantization is an effective method that can improve communication efficiency in federated learning (FL). The existing FL quantization protocols almost stay at the level of post-training quantization (PTQ), which comes at the cost of large quantization loss, especially in the setting of low-bits quantization. In this work, we propose a FL quantization training strategy to reduce the impact of quantization on model quality. Specifically, we first apply quantization-aware training (QAT) to FL (QAT-FL), which reduces quantization distortion by adding a fake-quantization module to the model so that the model could perceive future quantization during training. The convergence guarantee of the QAT-FL algorithm is established under certain assumptions. On the basis of the QAT-FL algorithm, we extend the discussion of non-uniform quantization and the adaptive algorithm, so that the model can adaptively adjust the parametric distribution and the number of quantization bits to reduce the amount of traffic in training. Experimental results based on MNIST, CIFAR-10 and FEMNIST datasets show that QAT-FL has advantages in terms of training loss and model inference accuracy, and adaptive-bits quantization of QAT-FL also greatly improves communication efficiency.


  • A Data-Driven Adaptive Impedance Matching Method Robust to Parasitic Effects

    by Wendong Cheng, Li Chen, Weidong Wang

    Adaptive impedance matching between antennas and radio frequency front-end (RFFE) power modules is essential for mobile communication systems. To address the matching performance degradation caused by parasitic effects in practical tunable matching networks (TMNs), this paper proposes a data-driven adaptive impedance matching method that avoids physical adjustment. First, we propose the residual enhanced circuit behavior modeling network (RECBM-Net), a deep learning model that maps TMN operating states to their scattering parameters (S-parameters). Then, we formulate the matching process based on the trained surrogate model as a mathematical optimization problem. We employ two classic numerical methods with different online computational overhead, namely simulated annealing particle swarm optimization (SAPSO) and adaptive moment estimation with automatic differentiation (AD-Adam), to search for the matching solution. To further reduce the online inference overhead caused by repeated forward propagation through RECBM-Net, we train an inverse mapping solver network (IMS-Net) to directly predict the optimal solution. Simulation results show that RECBM-Net accurately predicts S-parameters, achieving a mean absolute error of 6.98 × 10−5. Across 9000 mismatched scenarios, the compliance rate after tuning increases from 0.97% with the analytical solution of the ideal L-network to 95.92% with SAPSO, 93.42% with AD-Adam, and 95% with IMS-Net. While AD-Adam significantly reduces computational overhead, lowering the average number of RECBM-Net inferences from 2097 with SAPSO to 285, it sacrifices some accuracy. IMS-Net requires only a single inference to obtain the matching solution, resulting in minimal online overhead while maintaining excellent matching accuracy.


  • Gyroscope-Free UE Angular Velocity and Attitude Estimation With Channel State Information Variation

    by Wanting Xie, Li Chen, Xiaohui Chen, Weidong Wang, Huarui Yin

    Attitude and angular velocity estimation is essential for various applications of the user equipment (UE). In gyroscope-free scenarios, how to perform attitude and angular velocity estimation is a critical issue. In this letter, we propose a UE attitude and angular velocity estimation framework with the assistance of channel state information (CSI) variation. We first present a detailed proof of the existence of the mapping from CSI variation to angular velocity. Furthermore, we employ a neural network to learn this mapping, leveraging its capacity to model complex relationships. Then the estimated angular velocity is incorporated into the kinematic model of the gyroscope to derive the current attitude estimate of the UE. Simulation results further substantiate the significant potential of our method as a viable alternative to gyroscopes. Moreover, results also show that more propagation paths leads to better angular velocity and attitude estimation.

  • Mixed-Precision Quantization: Make the Best Use of Bits Where They Matter Most

    by Yiming Fang, Li Chen, Yunfei Chen, Weidong Wang, Changsheng You

    Mixed-precision quantization offers superior performance to fixed-precision quantization. It has been widely used in signal processing, communication systems, and machine learning. In mixed-precision quantization, bit allocation is essential. Hence, in this paper, we propose a new bit allocation framework for mixed-precision quantization from a search perspective. First, we formulate a general bit allocation problem for mixed-precision quantization. Then we introduce the penalized particle swarm optimization (PPSO) algorithm to address the integer consumption constraint. To improve efficiency and avoid iterations on infeasible solutions within the PPSO algorithm, a greedy criterion particle swarm optimization (GC-PSO) algorithm is proposed. The corresponding convergence analysis is derived based on dynamical system theory. Furthermore, we apply the above framework to some specific classic fields, i.e., finite impulse response (FIR) filters, receivers, and gradient descent. Numerical examples in each application underscore the superiority of the proposed framework to the existing algorithms.


  • Beyond Traditional Coherence Time: An Electromagnetic Perspective for Mobile Channels

    by Zihan Zhou, Li Chen, Ang Chen, Weidong Wang

    Channel coherence time has been widely regarded as a critical parameter in the design of mobile systems. However, a prominent challenge lies in integrating electromagnetic (EM) polarization effects into the derivation of the channel coherence time. In this paper, we develop a framework to analyze the impact of polarization mismatch on the channel coherence time. Specifically, we first establish an EM channel model to capture the essence of EM wave propagation. Based on this model, we then derive the EM temporal correlation function, incorporating the effects of polarization mismatch and beam misalignment. Further, considering the random orientation of the mobile user equipment (UE), we derive a closed-form solution for the EM coherence time in the turning scenario. When the trajectory degenerates into a straight line, we also provide a closed-form lower bound on the EM coherence time. The simulation results validate our theoretical analysis and reveal that neglecting the EM polarization effects leads to overly optimistic estimates of the EM coherence time.