Wence Zhang

Reflective visible light positioning (VLP) systems localize targets utilizing transceivers mounted on walls or ceilings for signal transmission and require only simple passive reflectors on targets. This design offers passivity and portability, making it suitable for mobile computing applications like logistics robot tracking and augmented reality, where wide reflection ranges are critical for maintaining connectivity. Among various reflectors, convex mirrors (CMs) are notable due to the excellent reflection range. This paper derives the Cramér-Rao Lower Bound (CRLB) for received signal strength (RSS)-based VLP systems that employ CMs as reflectors, and reveals the mutual coupling effects between CM position and orientation estimation. To address limited localization coverage of CM, this paper derives coverage probability formulas from LED and PD perspectives, which guide the optimization of the integrated LED-PD module layout for enhancing VLP coverage. To address the challenge of coupled position and orientation estimation, we propose a simultaneous CM position and orientation (SCPAO) estimation algorithm. It strategically decouples the joint estimation task, allowing dedicated Levenberg-Marquardt (LM) solvers to alternately refine the position and orientation estimates. Simulation and experimental results validate the feasibility of the SCPAO algorithm for real-time tracking and localization.

Visible light positioning (VLP) enables high-precision indoor localization, but single-sensor systems do not scale well to multi-user settings: camera-based methods suffer from motion jitter, while photodiode (PD)-based approaches are vulnerable to ambient light and multipath effects. This paper proposes a camera-photodiode fusion-based cooperative localization algorithm (C-CPD): cameras extract geometric features of a circular luminaire for coarse positioning, and PD-light-emitting diode (LED) modules measure distances to cooperative visible light communication (VLC) units via received signal strength (RSS) to refine estimates, mitigating motion distortions. We derive the Cram´er-Rao lower bound (CRLB) to characterize performance, revealing impacts of focal length, signal-to-noise ratio (SNR), and cooperative VLC units. Simulations show 0.41 cm average error, and experiments demonstrate 3.24 cm real-world error, outperforming baselines in dynamic scenarios. This work highlights multi-sensor fusion and geometric feature exploitation for robust, high-accuracy VLP, with CRLB insights guiding system design.

The channel knowledge map (CKM) is an emerging technology for enabling future integrated sensing and communication (ISAC) services that have attracted significant research interests in recent years. Current methods for constructing CKM primarily include interpolation-based, model-based, and machine learning algorithms. However, these methods are often limited by their low-estimation accuracy or high-computational complexity. To address these challenges, we propose a novel approach to construct CKM based on environmental partitioning and interpolation weight learning (EPIWL). The proposed method leverages channel state information (CSI) from partially known locations to interpolate and estimate the channel state in the target region, thus completing the CKM construction. To reduce the computational complexity, we proposed a graph-based feature aggregation and community detection-based partitioning (GFA-CD-P) algorithm, which selects representative anchor points through environmental partitioning, thereby decreasing the computational load. Furthermore, we propose an interpolation weight learning scheme based on Kolmogorov–Arnold network (KAN) and multihead cross attention (MHCA), namely, KM-IWL algorithm, which automatically learn the weight between anchor points and target points, enhancing both the efficiency and accuracy of CKM construction. The experimental results demonstrate that the proposed EPIWL approach achieves a 10-dB improvement in the normalized mean squared error (NMSE) and reduces the computational complexity by over 60% compared with existing schemes, showcasing robust performance and excellent generalization capability.

This paper proposes a design method for broadband power amplifier (PA) using irregular matching structure based on an improved multi‐objective optimization algorithm, aiming at expanding its operating bandwidth. Initially, the gravitational search algorithm is enhanced by reconstructing the gravitational constant function for improving its global optimization capability. Subsequently, a fast non‐dominated sorting mechanism combined with a crowding distance strategy is introduced for multi‐objective optimization problems. Furthermore, to effectively overcome the bandwidth limitations that are inherent in conventional PA, the proposed multi‐objective gravitational search algorithm is employed for optimizing the irregular structure matching network. For verification, a broadband high‐efficiency PA is designed and fabricated, covering a frequency range of 0.4–4.0 GHz with a relative bandwidth of 163%. Measurement results show that the PA with the irregular matching structure maintains an efficiency of 59.1%–64.4% across the entire operating band, with a saturated output power of 40.3–41.8 dBm.

This letter proposes a multiload dual-state matching optimization method for load-mismatch-resilient Doherty power amplifiers (DPAs). The combination impedances under different external loads are analyzed, along with the corresponding load characteristics of the carrier and peaking amplifiers in both the saturation and back-off states. Then, the carrier and peaking output matching networks are optimized using the dual-state optimization under external load variations within 2:1 VSWR. For verification, a 2.8-GHz load-mismatch-resilient DPA is designed and measured. The results show that, under a 50- $\Omega $ load and the loads for a 2:1 VSWR, the proposed DPA achieves a saturated output power of 43.4–43.7 dBm with a corresponding drain efficiency (DE) of 56.3%–65.9% and a DE of 50.2%–55.1% at 6-dB back-off output power. In addition, with a 20-MHz-modulated signal and digital predistortion (DPD), the DPA achieves an adjacent channel leakage ratio (ACLR) better than −47.2 dBc.

To address the challenges of low sampling efficiency in obstructed environments, high fingerprint database construction costs, and insufficient positioning accuracy in indoor visible light positioning (VLP) systems, this article proposes an integrated intelligent framework termed the obstructed environment-based intelligent fingerprint positioning (OEIFP). This framework combines adaptive sampling, sparse expansion, and optimized positioning algorithms. The framework first employs the obstructed environment adaptive sampling (OEAS) algorithm to achieve highly representative sparse sampling by incorporating obstacle information, thereby significantly reducing the cost of fingerprint data collection. Subsequently, the variational autoencoder with convolutional encoding (VACE) model accomplishes high-quality reconstruction from sparse to dense fingerprint databases. On this basis, the firefly algorithm-based cross-variation optimized extreme learning machine (FA-CVO-ELM) positioning model is designed, integrating the firefly algorithm (FA) with cross-variation operation (CVO) to perform dual optimization of the input weights and biases of the extreme learning machine (ELM), which enhances positioning accuracy and generalization capability. Simulation and experimental results demonstrate superior performance across diverse obstacle scenarios. The simulation achieves a 99.75% reduction in database construction cost compared to the initial sampled fingerprint database, while maintaining an average positioning error (APE) as low as 0.0423 m. Experimental results show a 97.6% reduction in construction cost with a minimum APE of 0.0503 m. The framework realizes synergistic optimization between adaptive sampling for low-cost dense database construction and high-precision positioning, offering a viable solution for deploying visible light fingerprint positioning systems in obstructed environments.

In this letter, we propose a camera-photodiode fusion-Based 6-degree-of-freedom (DoF) state estimation (i.e., 3D location and 3D pose angles) for cooperative multi-User visible light positioning (CPF-VLP). In the considered model, the camera extracts geometric features of a circular luminaire for 6-DoF state estimation. Subsequently, the orientations of PD–LED modules can be derived from the estimated camera pose, thereby eliminating the need for inertial inputs that are prone to cumulative drift. In addition, PD-LED modules measure the distances between cooperative VLC units using received signal strength (RSS) to refine the estimates and mitigate motion-induced distortions. Simulation results demonstrate that the CPF-VLP algorithm achieves an average positioning error of 1.47 cm.