The existing works on holographic MIMO are mainly based on the time encoding (TE) scheme. Since the continuous aperture of holographic MIMO is able to capture both the temporal and the spatial variation of electromagnetic waves, we propose a space-time encoding (STE) scheme, which relies on the orthogonal basis function representation of the spatial-temporal EM waves. From the perspective of electromagnetic information theory, we derive the achievable information rate of the STE scheme in the narrowband communication systems and prove that the STE scheme achieves a higher information rate than the TE scheme. Firstly, We build the transmission model of the STE scheme based on electromagnetic information theory and investigate the characteristics of the model, including the blocklength of codewords and the signal-to- noise ratio. Specifically, the blocklength is determined through proving the eigenvalue distribution of the space-time-wavenumber-frequency limited operator and the signal-to-noise ratio is obtained based on proposed noise model. Then we derive the achievable information rate of both the STE scheme and the TE scheme by employing the finite blocklength information theory. Closed-form approximations of the rates are further derived, based on which we prove the conclusion that the STE scheme achieves a higher information rate than the TE scheme while utilizing the same spatial and temporal resources. Numerical results verify the accuracy of the approximations and indicate that the STE scheme improves the information rate by 7.96% over the TE scheme.

With the development of 6G immersive communication, video intelligent analytics has garnered significant attention. Video intelligent analytics has diverse requirements in different immersive service scenarios, especially in accuracy and latency. However, as resource-limited terminal devices struggle to accom-plish high-accuracy video intelligent analytics tasks, video frames have to be offloaded to edge nodes with sufficient computational and cache resources for further processing. Therefore, in this paper, we consider device-edge collaboration video intelligent an-alytics tasks to improve trade-off performance between accuracy and latency. Specifically, we propose a joint optimization scheme for video frame scheduling, adaptive video frame compression and Machine Learning (ML) model caching to maximize the minimum of utility among all users. We divide the joint optimization problem into two sub-problems and use convex optimization to solve the adaptive frame compression optimization problem. Furthermore, to avoid the curse of dimensionality, we design an expert-assisted proximal policy optimization (EPPO)-based joint video frame scheduling and resource allocation algorithm. Simulation results demonstrate the superiority of the proposed scheme in improving video intelligent analytics performance.

With the explosive growth of communication requirements for real-time intelligent tasks, mobile communication is shifting from the traditional communication to task-oriented communication, where the transmitted data is shifting from undifferentiated transmission to value-oriented transmission. To maximize the value of transmitted data, it is urgent to match the source decisions with the task demands and wireless channel state. In this article, we focus on the joint source-channel optimization problem in task-oriented communication, and we design the timeliness-accuracy degradation (TAD) metric to measure the value of transmitted source. Moreover, we design a source value-based resource allocation scheme to minimize the TAD through joint optimization of task data generation and compression strategies, bandwidth allocation, and transmit power selection. Furthermore, to avoid the curse of dimensionality, we propose dimension-refined reinforcement learning (DRRL) algorithm to obtain the optimal solution of the problem in a stable and low-complexity manner. Numerical results demonstrate that the designed scheme can effectively improve the task performance and verify the low complexity and stability of the algorithm.

The continuous aperture of Holographic MIMO enables us to encode and transmit information spatially. This paper investigates the achievable rate of linear Holographic MIMO with arbitrary aperture-length using the finite blocklength information theory. Specifically, we first employ the prolate spheroidal wave functions to expand the received wavenumber band-limited electromagnetic field. This orthogonal representation enables two schemes to convey information related to the normal additive white Gaussian noise (AWGN) channel and the non-normal AWGN channel, namely the NA and NNA schemes, respectively. Then we derive the accurate achievable rates and the converse bounds of the two schemes by extending the$\kappa \beta $bound in finite blocklength information theory. Moreover, we derive an approximate closed-form expression of the achievable rate in the large aperture-length regime based on normal approximation. The approximation indicates that for a given space efficiency, the error probability decreases rapidly as the aperture length L increases, with the rate of decline determined by$Q\left ({{O\left ({{\sqrt {L}}}\right)}}\right)$. Finally, we obtain the asymptotic results when the aperture-length tends to infinity. Numerical results demonstrate that the NA scheme outperforms the NNA scheme when the blocklength is small, while the NNA scheme excels in the large blocklength regime. The accuracy of the approximation and the validity of the asymptotic results are verified.

With the commercial application of 5th generation, the coexistence scenario of enhanced Mobile Broadband (eMBB) and ultra-Reliable and Low Latency Communication (uRLLC) is facing significant challenges in utilizing limited resources. The existing scheme of only using puncturing or superposition cannot meet the heterogeneous requirement of eMBB and uRLLC. In this paper, we propose a cooperative multiplexing mode dynamic selection and resource allocation scheme to achieve the trade-off between the transmission quality (the transmission rate and the transmission accuracy ratio) of eMBB and the reliability of uRLLC, which considers power limitation. In the scheme, the multiplexing mode includes puncturing mode and Non-Orthogonal Multiple Access (NOMA) mode. After the multiplexing mode is selected, the resource block and power allocation are carried out. Furthermore, we propose a CoDueling Deep Q-learning Network to obtain the expected long-term benefits of the formulated scheme. Simulation results show that the proposed algorithm reduces the computation time by 27.3% and outperforms the compared scheme. Moreover, the proposed scheme improves the overall transmission quality of the coexistence scenario, where both eMBB and uRLLC services do not occupy the resources selfishly.

Single image dehazing is an important technology in the field of computer vision. In this paper, we propose an image dehazing via dual learning strategy, named dual-task cycle network (DCNet). The core of DCNet is a dual learning framework, which consists of two tasks: the dehazing task and the haze generation task. The dehazing task completes the image dehazing, while the haze generation task achieves the restoration from the dehazed image to the haze image and can form a cycle to provide additional supervision. Our method uses the duality between each task as a constraint to learn and train two tasks jointly, so that the effects of the dehazing model can be improved. Since the haze generation process does not depend on clear images, the DCNet can satisfy the requirements for limited supervision. Extensive experiments demonstrate that our DCNet performs favorably on haze removal.

Using deep learning techniques to process 3D objects has achieved many successes. However, few methods focus on the representation of 3D objects, which could be more effective for specific tasks than traditional representations, such as point clouds, voxels, and multi-view images. In this paper, we propose a Sphere Node Graph (SN-Graph) to represent 3D objects. Specifically, we extract a certain number of internal spheres (as nodes) from the signed distance field (SDF), and then establish connections (as edges) among the sphere nodes to construct a graph, which is seamlessly suitable for 3D analysis using graph neural network (GNN). Experiments conducted on the ModelNet40 dataset show that when there are fewer nodes in the graph or the tested objects are rotated arbitrarily, the classification accuracy of SN-Graph is significantly higher than the state-of-the-art methods.

This paper aims to present a drone swarm light show design platform to support STEAM (science, technology, engineering, art and mathematics) education for K-12 children. With this platform, children can use this platform to design a drone swarm light show easily. To this end, the architecture of this platform contents three layers: UI layer, command layer, and physical layer. The UI layer has an easy-to-use interface for children. Children can feed parameters about the light show by clicking buttons and dragging sliders of four tracks. All actions designed for the swarm in the UI layer will be generated automatically in the form of the drone’s desired trajectories through the command layer. The physical layer includes a router for communication and a drone swarm for the light show. Our experimental results demonstrate that this platform works efficiently and suits for being applied to real STEAM education.

In the smoothed particle hydrodynamics (SPH) fluid simulation method, thesmoothing length affects not only the process of neighbor search but also the calculation accuracy of the pressure solver. Therefore, it plays a crucial role in ensuring the accuracy and stability of SPH. In this study, an adaptive SPH fluid simulation method with a variable smoothing length is designed. In this method, the smoothing length is adaptively adjusted according to the ratio of the particle density to the weighted average of the density of the neighboring particles. Additionally, a neighbor search scheme and kernel function scheme are designed to solve the asymmetry problems caused by the variable smoothing length. The simulation efficiency of the proposed algorithm is comparable to that of some classical methods, and the variance of the number of neighboring particles is reduced. Thus, the visual effect is more similar to the corresponding physical reality. The precision of the interpolation calculation performed in the SPH algorithm is improved using the adaptive-smoothing length scheme; thus, the stability of the algorithm is enhanced, and a larger timestep is possible.

The imperfect material effect is one of the most important themes to obtain photo-realistic results in rendering. Textile material rendering has always been a key area in the field of computer graphics. So far, a great deal of effort has been invested in its unique appearance and physicsbased simulation. The appearance of the dyeing effect commonly found in textiles has received little attention. This paper introduces techniques for simulation of staining effects on textiles. Pulling, wearing, squeezing, tearing, and breaking effects are more common imperfect effects of fabrics, these external forces will cause changes in the fabric structure, thus affecting the diffusion effect of stains. Based on the microstructure of yarn, we handle the effect of the stain on the imperfect textile surface. Our simulation results can achieve a photo-realistic effect.

s: This paper introduces a polarized catadioptric virtual reality optical system. With a focus on the issue of serious ghost image in the system, root causes are analyzed based on design principles and optical structure. The distribution of stray light is simulated using Lighttools, and three major ghost paths are selected using the area of the diffuse spot, Sd and the energy ratio of the stray light, K as evaluation means. A method to restrain the ghost image through optimization of the structure of the optical system by controlling the focal power of the ghost image path is proposed. /Conclusions The results show that the Sd for the ghost image path increases by 40% and K decreases by 40% after optimization. Ghost image is effectively suppressed, which provides the theoretical basis and technical support for ghost suppression in a virtual reality optical system.