The SAGIN system is comprised of satellite edge computing nodes, aerial computing nodes, and terrestrial computing nodes, along with the networks that interconnect them. The task scheduling process includes the generation of tasks and task information by users, the creation of task offloading decisions based on scheduling algorithms and task information, the execution of these offloading decisions, executing tasks locally or uploading them to appropriate computing nodes, and returning the results to users. Initially, we define the resources and task models within the task scheduling of the SAGIN. Subsequently, we decompose the task offloading process and define the metrics, and finally, we establish optimization objectives.
Satellite Mobile Edge Computing System
Figure 1. Space-Air-Ground Integrated Network Architecture
In the SAGIN, as illustrated in Figure 1, a multitask MEC system typically comprises terrestrial nodes, aerial nodes, and satellite nodes. The satellite nodes are divided into LEO and high-Earth orbit (HEO) clusters. Terrestrial and aerial nodes are interconnected by ground and air network segments, respectively. In the satellite network, the LEO cluster is positioned in orbits closer to the Earth’s surface, which results in lower communication latencies and provides certain computational capabilities. These nodes fulfill both computational and communication functions. However, they exhibit high mobility, unstable network connections, and limited computational resources. In contrast, the HEO cluster remains relatively fixed in relation to the Earth, featuring a stable network topology and broad coverage.
Satellite computing nodes, aerial computing nodes, and terrestrial computing nodes together constitute the structure of the space-air-ground integrated edge computing system. This structure represents a comprehensive, multilevel collaborative computing system that integrates computing resources from terrestrial, aerial, and space layers to form a cross-domain, efficient data processing network. Within this architecture, cloud computing nodes are at the core of the system, typically deployed in data centers with robust data processing and storage capabilities. They are responsible for executing complex data analysis tasks and long-term data storage, while also managing and deploying strategies across the entire network. Edge computing nodes are positioned closer to the data sources and may include terrestrial base stations, routers, aerial drones, airships, or space-based satellite platforms. These nodes have the capability for data processing and temporary storage, allowing them to preprocess and initially analyze collected data to reduce data transmission volumes, lower network latency, and support real-time or near-real-time application requirements. Terminal computing nodes represent the network’s endpoints, such as smartphones, sensors, and other IoT devices that directly generate data. Some devices are also capable of basic data processing and filtering to further optimize data flow and enhance system responsiveness. This layered design of the space-air-ground integrated edge computing architecture not only achieves global service coverage but also enhances data processing efficiency and speed, meeting the demands for latency sensitivity and computational power across various application scenarios.