Discussion and Limitations

A. Satellite Computing

With advancements in the space industry, deploying computational tasks on satellites is becoming feasible. By providing computational capabilities on satellites, mobile satellite networks will be able to offer MEC services to users with lower latency and reduced transmission rates. These services include a variety of applications, such as artificial intelligence (AI), remote sensing data processing, and the Internet of Things (IoT). SkyOctopus can be generalized to satellite computing, thereby enabling the offloading of user traffic to satellite computing nodes via the S-UPF.

随着航天工业的进步，在卫星上部署计算任务正变得可行。通过在卫星上提供计算能力，移动卫星网络将能够以更低的时延和更少的传输开销，为用户提供移动边缘计算（MEC）服务。这些服务涵盖多种应用，如人工智能（AI）、遥感数据处理以及物联网（IoT）。SkyOctopus架构可以被推广至卫星计算领域，从而通过S-UPF实现用户流量向卫星计算节点的卸载。

B. Content Delivery Network

In this paper, we consider that users always access the specific target server. However, the content delivery network (CDN) allows users to be served by the nearest available server. In such cases, SkyOctopus still provides users with the optimal anchor point selection, but the services requested by users may cluster around a specific anchor point, leading to an increase in its utilization. We believe that cooperation among anchor points can better adapt to CDN and leave it for future work.

在本文中，我们考虑的是用户总是访问特定的目标服务器。然而，内容分发网络（CDN） 允许用户由最近的可用服务器提供服务。在此类情况下，SkyOctopus仍然能为用户提供最优的锚点选择，但用户请求的服务可能会聚集于某个特定的锚点，导致其利用率（utilization）上升。我们相信，锚点间的协作能够更好地适应CDN，并将其作为未来的工作。