Satellite edge caching enables satellites to store and deliver content locally, offering a promising approach to support low-latency and on-demand video services. In this paper, we introduce the concept of Closer is not always better in ultra-dense Low Earth Orbit (LEO) constellations. This refers to scenarios where users are simultaneously covered by multiple satellites, while geographically closer satellites may appear more suitable for service delivery, they often face challenges in inter-satellite cooperation. To address this, we propose a hop-based cooperative satellite caching (HCSC) method. Specifically, we develop a hop-based fast visible satellite selection strategy to identify the top-k satellites for cooperative caching. Furthermore, we design a Dueling Deep Q-Network (Dueling DQN)-based algorithm to optimize caching decisions, aiming to minimize service delay through effective satellite collaboration. Simulation results demonstrate that the proposed HCSC method outperforms other baseline caching algorithms by reducing transmission delay and improving cache hit ratio.

Low Earth Orbit (LEO) Mega-Constellation Net-works (MCNs) have become increasingly popular in recent years due to their ability to offer global coverage, high-throughput data transmission, and low latency for users across the world. While MCNs significantly improve continuous service coverage worldwide, they also pose a challenge in terms of managing a large number of satellites in the system, which increases the complexity of routing and the number of hops required for inter-satellite links (ISLs). In this paper, we propose a software-defined mega-constellation network (SDMCN) architecture that provides flexible network management by enabling dynamic global network monitoring and collecting satellite information. In order to optimize routing in Walker Delta MCNs, we propose an orbit-grid-based dynamic routing (OGDR) algorithm, which begins by developing a path model that selects access satellites based on both the minimum hop metric and the ISLs distribution. Then, leveraging the periodicity of the satellite and the characteristics of the path model, inter-plane and intra-plane links are selected on an orbit grid directly based on the latitude of the satellites to ensure the minimum delay. Finally, we introduce a path failure recovery mechanism to ensure the reliability of the path. Experimental results demonstrate that the algorithm not only achieves performance similar to the shortest path but also minimizes the number of ISL hops and has lower computational complexity.

Satellite-terrestrial integrated network (STIN) possesses the advantage of offering ubiquitous coverage, thereby enabling people to access service freely even in remote areas. There is a growing market for immersive media (IM), and delivering high-quality IM services via stable routes has become a highly active research area. However, existing routing approaches primarily focus on optimizing network performance, the tailored routing strategies for IM services are scarce in the literature. In this paper, we propose an economic routing (EcoR) method for IM services in STIN. Specifically, for the scenario where IM service requests are highly concentrated, utilizing the cost balance method to lighten the burden of maintaining heavy flows on satellite networks. We formulate an optimization problem that aims to direct the traffic to less congested links while ensuring an IM service guarantee and achieving a cost balance for the network. To this end, we propose an advanced Dijkstra algorithm that combines the deep forest searching method and the traditional Dijkstra algorithm to identify all potential routing paths. Additionally, we adopt a greedy approach to select the most appropriate routing path that minimizes the cost. Evaluation results show that the proposed EcoR algorithm achieves cost savings of 32% and a service success rate increase of 64% when compared to other routing methods.

The Mars exploration mission of China named Tianwen-1 is being carried out as scheduled. The Navigation and Terrain Cameras (NaTeCam) equipped on the Zhurong Rover play an essential role in obstacle recognition. The main obstacles on the Martian surface are rocks of different sizes, which influence the path planning of Zhurong Rover in scientific exploration. Most existing semantic segmentation methods are based on the U-Net architecture with ResNet or other backbones, and features extracted by these methods lack long-range dependencies. To fully exploit the context information, we propose the MarsNet framework for the Mars image, which combines transformers with the convolutional neural network (CNN) as the backbone, and hybrid dilated convolution (HDC) is also employed to the decoder path to help detect the huge rocks. Besides, since there are few open-source datasets for rock segmentation for Mars, we establish a segmentation dataset from the Martian surface image, named TWMARS, captured by NaTeCam. Extensive experiments are conducted on the TWMARS dataset, and the experimental results demonstrate that MarsNet achieves accurate rock segmentation and outperforms state-of-the-art methods. The source code is available athttps://github.com/BUPT-ANT-1007/MarsNet.

Light field records the information of the light in space and contributes to regenerate the content effectively, which makes immersive media more promising. In this paper, we propose a geometry-guided compact compression scheme (GCC) for light field image. We regard that the geometry of GCC includes the structure in a single sub-aperture image (SAI) and the relationship among SAIs, which can be used to fully explore the compact representation for light field image. The light field image is grouped into key SAIs and non-key SAIs. The key SAIs are obtained by down-sampling in the angular domain and arranged into the pseudo-sequence that needs to be compressed. We consider the superpixel-based segmentation algorithm to detect the contours and obtain the sketch map for the non-key SAIs. Meanwhile, the graph model is used to establish the relationships among the SAIs by the vertices and edges. On the decoder side, the light field image is reconstructed by the graph convolutional networks, and the sketch map optimizes the details of the recovered images to some extent. Experimental results show the benefit of GCC in terms of rate-distortion performances compare with several state-of-the-art methods for the real-world and synthetic light field datasets. Besides, the proposed GCC is able to generalize over datasets not seen during training.

In China’s Tianwen-1 mission for Mars exploration, Zhurong rover carries the Navigation and Terrain Camera (NaTeCam), and collects a lot of terrains and topographic data. For the scientific popularization of Mars, a virtual reality roaming based on panoramic stitching presents the Martian surface in detail and provides an immersive experience for users. However, the current Mars roaming systems focus on global information instead of details, and the traditional panoramic stitching methods are not suitable for Mars images. This letter proposes a panoramic stitching method and develops a 3DoF+ Mars roaming system. The proposed Stitching-Combines-Features-and-Projection (SCFP) jointly considers the extracted features of images and projection based on intrinsic camera parameters, which improves the matching effect and execution efficiency. Experimental results indicate that SCFP reduces the computational time and improves stitching results than state-of-the-art methods. Further, we design and deploy the 3DoF+ roaming system based on panoramic stitching with datasets obtained from the Tianwen-1 mission.

The light field has excellent application prospects in immersive media because of the abundant information of the light. Due to the sparsity and redundancy in light field images, light field reconstruction based on compressed sensing is used to recover light field images from only a few measurements. And the light field compressed sensing usually optimizes the measurement matrix and the dictionary and processes each of the light field images separately. Since the high similarity of light field images, the different viewpoints of images can be stacked together and formed as a 4D tensor. In this paper, we propose tensor based on compressed sensing (TCS) method to yield measurements with common characteristics. Besides, a better deep learning network is designed for TCS, the measurement matrix optimization and image reconstruction will be performed simultaneously. Experimental results show that the proposed method gets at least 3 dB gain in PSNR and outperforms state-of-the-art in the reconstruction quality.