LQLiangang Qi

Liangang Qi

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Liangang Qi has a total of 41 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Dynamic Satellite Communication Optimization via Multi-Satellite Cooperative SNR Prediction

    by Jingxiang Gao, Zhuo Wang, Siyu Lai, Liangang Qi

    In dynamic satellite communication scenarios, terminals without antenna servo systems suffer attitude disturbances that induce severe non-stationary fluctuations in the satellite-to-ground signal-to-noise ratio (SNR). The instantaneous SNR of the target communication satellite is often unobservable to the terminal. To address this problem, this paper proposes a supervised SNR prediction framework based on multi-satellite cooperative sensing. By designating one observable navigation satellite as the surrogate target, the unobservable link-quality prediction is recast as a supervised learning task; a random-forest scoring model selects the optimal observation combination, and a CNN-LSTM-Attention predictor with a lightweight unidirectional long short-term memory (LSTM) backbone forecasts the future SNR of the target link. On this basis, a joint strategy integrating prediction-driven communication-window selection, adaptive modulation, and hybrid automatic repeat request (HARQ) is developed with an online adaptive update mechanism. Numerical results show that, compared with direct communication, the joint strategy reduces the average bit error rate (BER) by more than five orders of magnitude, while nearly quadrupling the effective throughput at a communication-time proportion of only 59.25%. The proposed framework offers a feasible path for small low-power terminals in dynamic environments.

  • Open Access

    SLC-Domain SAR RFI Suppression via Sliding-Window Local Tensorization and Energy-Guided CUR Projection

    by Qiang Guo, Yuhang Tian, Shuai Huang, Liangang Qi, Sergiy Shulga

    Synthetic aperture radar (SAR) imaging is highly vulnerable to radio-frequency interference (RFI) in complex electromagnetic environments, which can introduce structured artifacts and obscure targets in single-look complex (SLC) products. Most existing suppression methods rely on separability along a single dimension or require interference-specific parameter tuning, limiting robustness under multidimensional coupling and strong scatterers. We propose a range-domain sliding-window local tensorization that rearranges SLC data into localized range–azimuth–block-index tensors to better expose multi-mode correlations. On this representation, an energy-guided tensor CUR low-rank projector is embedded into an alternating-projection scheme that alternates complex-valued soft-thresholding for the sparse scene-plus-noise term and CUR-based projection for the structured RFI term. The cleaned SLC image is obtained by de-tensorizing the estimated RFI component and subtracting it from the input SLC. Experiments on semi-synthetic data, where controlled RFI is superimposed on real SLC scenes, and on real Sentinel-1 SLC data containing RFI demonstrate improved Pearson correlation coefficient (PCC) and perceptual image quality while preserving target signatures and scene textures, particularly under strong interference and strong coupling. The proposed approach provides a practical SLC-domain RFI mitigation tool for post-focusing SAR products without requiring explicit interference parameterization.

  • A Training-Free Subspace Suppression Method Based on Periodic Rearrangement for LFM Interference in SAR Imaging

    by Qiang Guo, Yuhang Tian, Shuai Huang, Liangang Qi, Sergiy Shulga

    To suppress striping artifacts and ghost targets induced by periodic Linear Frequency Modulation (LFM) interference in Synthetic Aperture Radar (SAR) imaging, this letter proposes a training-free four-stage subspace suppression scheme based on periodic rearrangement. By reorganizing the echo data according to the estimated interference period, a low-dimensional interference subspace with near–zero-phase dominance is constructed in the matrix domain, enabling the design of an orthogonal projection operator for selective interference removal. Owing to its subspace-based formulation, the proposed approach requires no data-driven learning or offline training, and it is well suited for practical implementation. Under extremely severe interference (SIR = -20 dB), the proposed method achieves up to a +11.86 dB gain in IRR compared with the reference methods, while reducing the EVM to 1.13%, effectively suppressing residual striping and preserving the target mainlobe response and fine structural details. These results demonstrate that the proposed scheme provides a favorable balance between robustness, target fidelity, and computational efficiency for SAR imaging in complex electromagnetic environments.

  • Cross-Scale Context-Aware Ship Detection in SAR Images Using CSCF-Net

    by Liangang Qi, Chen Huang, Qiang Guo

    Synthetic aperture radar (SAR) ship detection faces challenges such as scale diversity, weak target features, and strong background interference. To address these issues, we propose a cross-scale context fusion network (CSCF-Net), integrating a multiscale feature fusion (MSFF) module for cross-scale contextual aggregation and a multitask interactive detection head (MTIDH) for task-specific optimization through dynamic deformable convolution. Extensive experiments on three SAR datasets demonstrate superior performance; CSCF-Net achieves 98.5% mean average precision (mAP) on SSDD, 91.5% on HRSID, and 98.1% on SAR-Ship-Dataset, with 1.0%, 2.2%, and 0.9% improvements over baseline, respectively, outperforming state-of-the-art methods and validating the effectiveness of our proposed method.

  • UWB LOS/NLOS Identification Using Spatio-Temporal Deep Learning

    by Liangang Qi, Yun Zhao, Qiang Guo, Mykola Kaliuzhnyi

    Reliable identification of line-of-sight (LOS) and non-line-of-sight (NLOS) conditions is critical for accurate localization in ultra-wideband (UWB) systems. In this letter, we propose a Voronoi+CNN-BiLSTM architecture for LOS/NLOS classification of UWB signals. By integrating Voronoi-based spatial propagation features, the model captures geometric characteristics of the environment to enhance classification performance. Experimental results on the publicly available UPT dataset demonstrate that the proposed method achieves an overall accuracy of 97.43%. Cross-scene evaluations demonstrate robust generalization. Moreover, the model achieves an average inference time of 2.08 ms per sample on a standard CPU, making it suitable for real-time applications.


  • A Multi-View Attention Hypergraph Neural Network for Radar Emitter Signal Sorting

    by Hongzhuo Chen, Liangang Qi, Qiang Guo, Mykola Kaliuzhnyi

    In complex electromagnetic environments, the deinterleaving of dense, interwoven radar pulse signals poses a formidable challenge. To address the propensity of conventional graph models for confusion and misclassification in such scenarios, this letter proposes a novel method for radar emitter signal deinterleaving: the Multi-view Attention Hypergraph Neural Network (MVA-HGNN). This model maps radar pulses to nodes in a hypergraph, leveraging hyperedges to capture higher-order correlations among pulses. This overcomes the limitation of traditional graphs, which can only describe pairwise relationships. To fully exploit the heterogeneous information within Pulse Descriptor Words (PDWs), we construct two distinct hypergraph views: ”spatial” and ”intrinsic.” In the MVA - HGNN model, parallel hypergraph network branches learn node representations from different views. An advanced attention - based fusion mechanism is introduced to dynamically integrate these feature representations. Simulation results demonstrate that our method achieves superior performance, particularly in small data scenarios, showing great potential for engineering applications.


  • Deep Learning-Based NLOS Identification of UWB Signals

    by Liangang Qi, Yun Zhao, Qiang Guo, Mykola Kaliuzhnyi, Enqiang Wang

    Ultra-wideband (UWB)-basedindoor positioning systems (IPS) have superior positioning performance. In indoor complex environments, non-line-of-sight (NLOS) reception due to multipath and signal fading can significantly affect the ranging accuracy of UWB signals. In this case, conventional machine learning (ML) and deep learning (DL) methods often fail to achieve the required accuracy and precision. To address this challenge, we propose an ADCNN-BiLSTM-Transformer model that incorporates adaptive dilated convolution, adaptive weight decay, and an improved transformer-like attention mechanism to effectively classify NLOS signals in UWB localization. Experimental results show that the model achieves 92.19% accuracy on the public eWINE dataset, outperforming existing methods by 7.81%. It also achieves a 3.74% improvement on the public UPT dataset.


  • Image encryption based on four-dimensional multi-parameter robust chaotic system and dynamic spiral block transformation

    by Jianhong Xiang, Shubei Liang, Liangang Qi, Yu Zhong

    At present, image encryption methods generally suffer from the problems of insufficient randomness and weak attack resistance, particularly as traditional encryption methods often employ fixed image scrambling methods, making the encryption results easy to predict. To address this issue, this paper proposes an image encryption algorithm based on a four-dimensional multi-parameter robust chaotic system and dynamic spiral block transformation. First, a four-dimensional multi-parameter robust chaotic (4D-MPRC) system is proposed, which cross-couples the low-dimensional chaos into a high-dimensional chaotic system and incorporates multi-parameter control to better regulate chaotic behavior and ensure the stability of the system dynamics. Next, a dynamic spiral block transformation is proposed for image scrambling. The starting point and direction of scrambling are dynamically selected based on the chaotic system, which breaks the limitation of fixed image scrambling methods in the traditional image scrambling method, and makes the image scrambling more complex and difficult to decipher in the reverse direction. Finally, an image encryption model consisting of chaotic system, dynamic spiral block scrambling and DNA diffusion is constructed by combining DNA coding. The experimental results show that the proposed encryption model effectively improves the security and anti-attack capability of image encryption.


  • Efficient and Accurate Single-Snapshot DOA Estimation for Massive ULA via Enhanced ApFFT

    by Liangang Qi, Jiakang He, Qiang Guo, Lichao Liu, Mykola Kaliuzhnyi

  • Distributed Fusion of Highly Maneuvering Multitarget Under Limited Field of View Sensors

    by Qiang Guo, Long Teng, Liangang Qi

    The problem of tracking multiple highly maneuverable targets in a distributed sensor network is addressed under constraints of limited field of view, computational capacity, and communication resources. First, a hybrid-driven labelled multi-Bernoulli (HDLMB) filter, driven by Gaussian processes and motion models, is proposed to track multiple highly maneuverable targets. Second the local state estimates, rather than the local multi-target posterior densities, are fused by each node. This fusion strategy decouples distributed fusion from local estimates at individual nodes, aligning better with modular applications and reducing both fusion time and communication bandwidth. Finally, a suboptimal distributed fusion algorithm based on local track matching is developed. It is designed without the prerequisite of a known sensor field of view and effectively mitigates the NP-Hard problem associated with optimal matching while tracking multiple targets by multiple sensors. Numerical experiments have demonstrated that compared to advanced distributed fusion methods, the proposed approach achieves superior tracking accuracy and incurs lower fusion costs.