As one of the information industry’s future development directions, the Internet of Things (IoT) has been widely used. In order to reduce the pressure on the network caused by the long distance between the processing platform and the terminal, edge computing provides a new paradigm for IoT applications. In many scenarios, the IoT devices are distributed in remote areas or extreme terrain and cannot be accessed directly through the terrestrial network, and data transmission can only be achieved via satellite. However, traditional satellites are highly customized, and on-board resources are designed for specific applications rather than universal computing. Therefore, we propose to transform the traditional satellite into a space edge computing node. It can dynamically load software in orbit, flexibly share on-board resources, and provide services coordinated with the cloud. The corresponding hardware structure and software architecture of the satellite is presented. Through the modeling analysis and simulation experiments of the application scenarios, the results show that the space edge computing system takes less time and consumes less energy than the traditional satellite constellation. The quality of service is mainly related to the number of satellites, satellite performance, and task offloading strategy.

With the continuous development of modern society, the rapid expansion of human civilization squeezes the living space of other organisms, and the extinction of more and more biological species has sounded the alarm for us. Therefore, in order to timely understand the changes of wild animals and other resources in a specific area, and facilitate relevant personnel to formulate effective restoration and protection measures, this paper proposes a wild Animal species recognition network based on deep learning and improved YOLOv5: YOLO-Animal. The application of artificial intelligence and computer vision has covered all aspects of human’s daily life and work. The benchmark YOLOv5 algorithm can quickly and accurately deal with problems associated with images. Through the fusion of weighted Bidirectional Feature Pyramid Network (BiFPN) and Effective Channel Attention (ECA) module, the original YOLOv5s network structure is enhanced, and the detection accuracy of small targets and occluded and fuzzy targets is effectively improved. The YOLO-Animal model outperforms the benchmark YOLOv5s model by 3.2% on mAP and achieves 95.5% accuracy on the test set.

The intelligent satellite, iSAT, is a concept based on software-defined satellites. Earth observation is one of the important applications of intelligent satellites. With the increasing demand for rapid satellite response and observation tasks, intelligent satellite in-orbit task planning has become an inevitable trend. In this paper, a mixed integer programming model for observation tasks is established, and a heuristic search algorithm based on a symmetric recurrent neural network is proposed. The configurable probability of the observation task is obtained by constructing a structural symmetric recurrent neural network, and finally, the optimal task planning scheme is obtained. The experimental results are compared with several typical heuristic search algorithms, which have certain advantages, and the validity of the paper is verified. Finally, future application prospects of the method are discussed.

This work proposes a geometric Covariance Intersection (CI) based algorithm for efficiently solving the decentralized Cooperative Localization (CL) problem with multiple TOA measurements. Unlike existing algorithms that only process one measurement each time, our proposed algorithm processes multiple measurements in a batch style, which improves the precision performance by considering the correlation among these measurements. An efficient way to get the optimal solution is given in our algorithm, which reduces the computational complexity to O(NlogN). Furthermore, the sparsity of the optimal solution is proved, which enables a heuristic state buffering approach to reduce the usage of ranging link resources and improve measurement efficiency. Simulation results show that our algorithm outperforms the conventional methods.