SkyCastle Overview

Key Ideas Behind SKYCASTLE

We present SKYCASTLE, a novel global mobility management mechanism with the key ideas as follows.

• (i) Judiciously deploying distributed anchors at LEO satellites to efficiently manage the mobility of both users and network infrastructures. Taking the satellite constellation as the frame of reference, users and GSs continuously change their topological locations in the ISTN. SKYCASTLE leverages anchors at satellites to track where a user or GS is and cope with their handovers. Specifically, when a user or GS connects to a new ingress satellite, its new ingress satellite sends an MM message to its satellite anchor to notify its current location. The ingress-to-anchor path is stable since the network topology of the space segment in the same orbital altitude is stable.

• (ii) Dynamically assigning satellite anchors to users to avoid triangular routing and attain low latency. Because satellite anchors are moving, the user-to-anchor path can be prolonged over time. To avoid the latency increase, S KYC ASTLE dynamically updates the anchor assignments and always chooses anchors satisfying the latency requirements while avoiding frequent anchor changes for users.

(i) 在LEO卫星上合理部署分布式锚点 ，以高效管理用户和网络基础设施的移动性。以卫星星座为参考框架，用户和GSs在ISTN中不断改变其拓扑位置。SKYCASTLE利用卫星上的锚点来跟踪用户或GS的位置并处理其切换。具体来说， 当用户或GS连接到新的入口卫星时，其新的入口卫星向其卫星锚点发送移动性管理消息以通知其当前位置 。入口到锚点的路径是稳定的，因为同一轨道高度的空间段网络拓扑是稳定的。

(ii) 动态分配卫星锚点给用户，以避免三角形路由并实现低延迟 。由于卫星锚点移动，用户到锚点的路径可能随时间延长。为了避免延迟增加，SKYCASTLE动态更新锚点分配，并始终选择满足延迟要求的锚点，同时避免频繁更改用户的锚点。

SKYCASTLE Architecture

Figure 5 plots the high-level overview of our SKYCASTLE design. A mobile user (e.g., an airplane) visits its destination content server via a path built upon a sequence of LEO satellites and GSs. Based on this typical architecture, SKYCASTLE incorporates two core components: (i) a collection of satellite anchors which are deployed at a subset of LEO satellites in the ISTN to handle the mobility; and (ii) an anchor manager deployed at the satellite operator’s control center which runs a series of algorithms to dynamically decide which satellites should carry the anchor functionality, and decide how geodistributed mobile users should be assigned to different anchors.

图5显示了我们SKYCASTLE设计的高层次概述。一个移动用户（例如飞机）通过一系列LEO卫星和GSs构建的路径访问其目标内容服务器。基于这种典型架构，SKYCASTLE包含两个核心组件：

（i）一组 部署在ISTN中部分LEO卫星上的卫星锚点 ，用于处理移动性

（ii）部署在卫星操作员控制中心的锚点管理器 ，它运行一系列算法，以动态决定哪些卫星应该承担锚点功能，并决定如何将地理分布的移动用户分配给不同的锚点

Satellite anchor and cluster division (§V). SKYCASTLE divides all satellites into multiple disjoint clusters. Each cluster is a satellite set which contains a collection of geographically adjacent satellites. One satellite in a cluster is selected as the anchor point of this cluster. A satellite anchor manages the mobility of all GSs and mobile users that connect to satellites in this cluster. In addition to managing the location (§V-A) and handover (§V-B) for users and GSs, an anchor also adopts a convergence-free route mechanism (§V-C) to deal with routing fluctuations and accomplish fast route recovery during LEO dynamics to reduce the connection interruption time.

卫星锚点和集群划分（§V） SKYCASTLE将所有卫星划分为多个不相交的集群。每个集群是一个包含一组地理上相邻卫星的卫星集合。 集群中的一个卫星被选为该集群的锚点。一个卫星锚点管理连接到该集群中的所有GSs和移动用户的移动性 。除了管理用户和GSs的位置（§V-A）和切换（§V-B）外，锚点还采用收敛自由的路由机制（§V-C）来处理路由波动，并在LEO动态过程中实现快速路由恢复，以减少连接中断时间。

集群 and 锚点

1. 划分成互不相交的集群
2. 每个集群一个Anchor
3. 这个Anchor负责此集群中所有对“GS 和 Mobile User”的处理
4. 额外操作 (routing):
   • 处理路由波动
   • 快速路由回复

Anchor manager (§VI). Since both users and satellites are mobile, a user may change their current cluster as well as their corresponding satellite anchor. Thus, the goal of SKYCASTLE’s anchor manager is to judiciously place anchor functionalities at different satellites and decide proper anchor assignments for geo-distributed users to guarantee that the end-to-end latency would not exceed a performance threshold specified by the satellite operator, while ensuring infrequent anchor/cluster change to reduce the interruption time for mobile users.

锚点管理器（§VI） 由于用户和卫星都在移动，用户可能会改变其当前集群以及对应的卫星锚点。因此，SKYCASTLE的锚点管理器的目标是合理地在不同卫星上放置锚点功能，并为地理分布的用户决定适当的锚点分配，以确保端到端延迟不会超过由卫星操作员指定的性能阈值，同时确保锚点/集群更改频率较低，以减少移动用户的中断时间。

锚点管理器

相当于是brain，专门做决策的，经由GS转发出去

1. 如何划分集群
2. 决定这个集群的锚点是谁
3. 为不同地理位置的用户选择合适的对应锚点
4. 尽可能保持“锚点/集群”的更改频率较低

Working stages. We describe an example to illustrate the working stages of SKYCASTLE. First, as shown in Figure 5, the operator invokes the anchor manager to calculate and decide how to divide all satellites into multiple clusters and which satellite is selected to deploy anchor functionalities. These decisions are delivered to satellites via distributed ground stations. When a user connects to the ISTN, it registers the current location information with the anchor. The anchor then assigns a unique IP address to the user. Interactive traffic between the user and their target server is forwarded via the anchor. Due to the mobility of both satellites and users, a user may change to another cluster, which triggers an anchor update. After the cluster handover, interactive traffic between the user and server is carried over a new path via the new anchor.

工作阶段 我们描述一个例子来说明SKYCASTLE的工作阶段。首先，如图5所示，操作员调用锚点管理器来计算和决定如何将所有卫星划分为多个集群，以及哪些卫星被选中部署锚点功能。 这些决定通过分布式地面站传递给卫星 。当用户连接到ISTN时，它在锚点上注册当前位置信息。 锚点然后为用户分配一个唯一的IP地址。用户与其目标服务器之间的交互流量通过锚点转发 。由于卫星和用户的移动性，用户可能会切换到另一个集群，这会触发锚点更新。在集群切换后，用户与服务器之间的交互流量通过新的锚点传递。