Guangyao Han

Although recent multi-object tracking (MOT) methods have achieved considerable progress, tracking from a moving vehicle remains challenging because ego-motion, scale variation, occlusion, and uncertain detections degrade motion prediction and data association. To address these issues, we propose an adaptive motion association framework consisting of a task-aligned enhancement head (TAE-Head), adaptive motion estimation (AME), and extended shape-constrained IoU (ES-IoU). TAE-Head improves the consistency between classification and localization features. AME adjusts process and measurement covariances, estimates a confidence-weighted common displacement from provisional matches, corrects predicted positions, and recomputes association before the final state update. ES-IoU increases geometric tolerance while suppressing shape-incompatible matches. On BDD100K, the proposed framework achieves 56.7% HOTA and 65.0% IDF1, exceeding the baseline by 9.0 and 7.7 percentage points, respectively, while operating at 32.5 FPS. On MOT17, it achieves 68.2% HOTA and 80.5% IDF1.

To meet the design and evaluation requirements of underwater vision-based docking localization, a Webots-based simulation platform for Autonomous Underwater Vehicle (AUV) visual docking localization was designed and implemented to address the high cost of real sea trials, uncontrollable operating conditions, and the difficulty of systematically covering extreme scenarios. An end-to-end simulation of the docking localization pipeline was provided. Visual components—including fiducial markers, underwater illumination and imaging, and occlusion—were modeled in relatively fine detail, while non-vision-dominant factors such as propulsion and hydrodynamics were treated with approximate models to balance visual realism and simulation efficiency. The platform supported multiple types of visual markers, parameterized configuration of underwater lighting and turbidity, and the generation of diverse occlusion scenarios, enabling unified integration and benchmarking of docking localization algorithms. The results showed that the platform offered tunable scene parameters, repeatable conditions, and broad algorithm compatibility, and it effectively revealed performance differences across algorithms for complex combinations of illumination, turbidity, and occlusion. These capabilities reduced the risk and cost of real underwater docking experiments and supported faster iterative improvement of vision-based localization methods.

Ocean buoys play a critical role in marine hydrological, water quality, and meteorological monitoring, with applications in navigation, environmental observation, and communication. However, accurately modeling and deploying a multi-buoy system in the complex marine environment presents significant challenges. To address these challenges, this study proposes an enhanced deployment strategy using the tuna swarm optimizer enhanced with the fractional-order calculus method for marine observation. The proposed method first introduces a detailed observation model that precisely captures the performance of buoys in terms of coverage and communication efficiency. By integrating the observation coverage ratio and communication energy consumption, we establish an optimal multi-buoy deployment model. The proposed method leverages tent chaotic mapping to improve the diversity of initial solution generation and incorporates fractional-order calculus to strengthen its search capabilities. Simulation experiments and statistical analysis verify the effectiveness of the proposed deployment model, with the proposed method achieving the best performance in deploying the multi-buoy system, reaching a final fitness value of 0.190052 at iteration 449, outperforming TSA, PSO, GWO, and WOA. These results highlight the potential of the proposed method in optimizing multi-buoy system deployment in marine observation.

The Internet of Underwater Things (IoUT) technology has been gaining importance in smart ocean investigation in recent years. The integrated system for underwater detection and communication (ISUDC) is considered a crucial technique for IoUT due to its potential to decrease equipment size and power consumption. The linear frequency modulation (LFM) signal has the characteristics of large time-bandwidth product and has great advantages in detection. However, it has high complexity and low robustness of communication in ISUDC. To address this issue, this article proposes a sector-modulated LFM (SM-LFM) signal for ISUDC. In terms of communication, it divides Fresnel domain into equal length sectors. Then, data is loaded by activating different sector. At the communication receiver, by establishing a relationship between the Dirac function in the Fresnel domain and the channel impulse response (CIR) in the time domain, data is demodulated by finding CIR’s main path without channel estimation or equalization. In terms of detection, SM-LFM has an improvement in the maximum unambiguous velocity by calculating relative radial velocity in the Fresnel domain instead of the frequency domain. Numerical simulations and lake experiments have confirmed that the SM-LFM signal can deliver superior robustness and easy-to-use advantages for underwater active detection and communication.

In underwater Internet of Things (IoT) applications, efficient and reliable transmission is demanding. Recently, the non-orthogonal multicarrier (MC) modulation is utilized as a promising spectrum-efficient solution in the UWA channel, where generally, the constellation mapping (CM) methods (such as 8-PSK and 16-QAM) are utilized to further increase the transmission rate. In this paper, we focus on the transmission of sonar images among IoT nodes and a constellation diagram learning-based adaptive sparse non-orthogonal wavelet division multiplexing (CDLA-SNOWDM) system is proposed to combat complex channel states between IoT nodes. We found that different CM methods contribute to unique time-frequency characteristics in constellation diagrams. We first utilize the time-frequency characteristics in designing the non-orthogonal subcarriers. The complex-valued CM signals are regarded as a new type of image. We introduce dictionary learning (DL) to non-orthogonal subcarrier designing, with the idea that designing CM signal adaptive non-orthogonal subcarriers can be regarded as designing an adaptive subcarrier dictionary. The CDLA-SNOWDM modulation is performed as a projection of the CM image to the subcarriers with an adaptive dictionary. The learned adaptive subcarriers can represent constellation diagrams more efficiently and improve transmission reliability. Both simulations and experiments show that the proposed scheme improved the reliability in transmitting sonar images over other orthogonal and non-orthogonal modulation schemes in UWA scenarios.

Multicarrier modulation shows the advantages of improving bandwidth efficiency in underwater acoustic (UWA) communications, whereas the more spectrum-efficient nonorthogonal frequency division multiplexing (NOFDM) is seldom investigated in the underwater scenario. The main idea of the NOFDM is to use overlapped subcarriers for increasing spectral efficiency. However, the NOFDM suffers from severe distortions both in the time and frequency domains because of the UWA channel's multipath propagation and Doppler spread. Considering these problems, we propose a novel time-reversal convolutional neural network (TR-CNN) based sparse NOFDM (S-NOFDM) scheme for UWA communication. The S-NOFDM waveform is determined by solving the sparse representation of input symbols under a set of nonorthogonal subcarriers, which can allocate symbols sparser to subcarriers and reduce distortion caused by multipath propagation and Doppler spread. The proposed TR-CNN is constructed with a time-reversal (T-R) convolutional layer where the T-R process is used to deal with multipath propagation. After the received signal has passed through the T-R layer, the main path of the UWA channel strength is enhanced and the other paths are weakened. The TR-CNN is also used to learn the inherent information between S-NOFDM symbols, improving UWA channel adaptability. After the training, the proposed scheme can be applied in the real-time UWA communication implementation. The simulation and experimental results prove that the proposed scheme outperforms the traditional scheme in signal detection and recovery. It is also robust with general capabilities in various UWA channel environments.

Underwater acoustic (UWA) communication has been developing rapidly over the past decades for its crucial position in resource exploration, environmental monitoring, and scientific research. However, the transmission data rate of UWA communication is limited by the narrow bandwidth of the underwater acoustic channel. Here, the generation of quadrature acoustic frequency combs (AFCs) is first reported. Massive parallel channels achieved with multiplexing of AFCs for UWA communication are demonstrated. The generated AFCs have unique characteristics which have a stable and precise division in frequency and time domain simultaneously with carrier spacing of 1 Hz with the stability of 10⁻⁸. The different frequency spacing in combs leads to different teeth spacings in time, which is orthogonal and separable in the time domain. The orthogonality among AFCs provides a new dimension for multiplexing and can drastically increase channel efficiency. The underwater acoustic communication experiments demonstrate that it is drastically increasing the transmission rate to 45.33 kb s⁻¹ and high spectral efficiency of 1.51 (channel s⁻¹) Hz⁻¹ without utilizing other modulation techniques by a single transducer. Since AFCs multiplexing is a completely independent degree of freedom that can be readily integrated with other high order modulation techniques, such as quadrature amplitude modulation and phase‐shift keying, in a single channel, AFCs multiplexing opens a new dimension for acoustic communication.

In this paper, compressed sensing (CS) theory is utilized in a medium-access control (MAC) scheme for wireless sensor networks (WSNs). We propose a new, cross-layer compressed sensing medium-access control (CL CS-MAC) scheme, combining the physical layer and data link layer, where the wireless transmission in physical layer is considered as a compress process of requested packets in a data link layer according to compressed sensing (CS) theory. We first introduced using compressive complex requests to identify the exact active sensor nodes, which makes the scheme more efficient. Moreover, because the reconstruction process is executed in a complex field of a physical layer, where no bit and frame synchronizations are needed, the asynchronous and random requests scheme can be implemented without synchronization payload. We set up a testbed based on software-defined radio (SDR) to implement the proposed CL CS-MAC scheme practically and to demonstrate the validation. For large-scale WSNs, the simulation results show that the proposed CL CS-MAC scheme provides higher throughput and robustness than the carrier sense multiple access (CSMA) and compressed sensing medium-access control (CS-MAC) schemes.