SkyCastle at the Anchor Level: Core Functionalities for ISTN Mobility Management

Location Management

The goal of location management is to track the location of mobile nodes in a network. The key difference between a user and GS node is that: GS nodes fundamentally belong to network infrastructure nodes (e.g., routers) and run routing protocols to calculate routes, while user nodes are located at the edge of networks and do not participate in routing calculation. In addition, users’ IP addresses are allocated by their anchor points while the IP addresses of GSs are configured by the satellite operator and they are fixed and do not change. In SKYCASTLE, the location information of a node is denoted as a key-value tuple, where the key is the IP address of the node and the value is the IP address of its ingress satellite. We next describe the location management for mobile user nodes and GS nodes respectively, which is also plotted in Figure 6.

Location management for users. The IP address of a node implicitly indicates its location in the network. Once a mobile user node connects to the ISTN, its ingress satellite generates the location information of the user, and registers with the anchor node. If the user moves and changes its ingress satellite, the new ingress satellite updates the location information of the user and notifies the anchor point in the new cluster.

Location management for GSs. When a GS node changes its position in the network, its ingress satellite notifies all anchor points with the latest location information of the GS node. By this method, all anchors in the ISTN can track the position of all GSs as well as their ingress satellites in real time, and thus avoid the routing convergence process when the network topology changes. When a satellite needs to forward traffic to a terrestrial server, data packets are first forwarded to its anchor point, then to the GS, and finally to the destination.

位置管理的目标是跟踪网络中移动节点的位置。用户节点和GS节点之间的关键区别在于：GS节点本质上属于网络基础设施节点（例如路由器），并运行路由协议来计算路由，而用户节点位于网络边缘，不参与路由计算。此外，用户的IP地址由其锚点分配，而GS的IP地址由卫星操作员配置，并且是固定的且不会改变。在SKYCASTLE中，节点的位置信息表示为一个键值对，其中键是节点的IP地址，值是其入口卫星的IP地址。接下来，我们分别描述移动用户节点和GS节点的位置管理，这也在图6中展示。

用户的位置管理。节点的IP地址隐含地指示了其在网络中的位置。一旦移动用户节点连接到ISTN，其入口卫星生成用户的位置信息，并在锚点节点上注册。如果用户移动并改变其入口卫星，新的入口卫星更新用户的位置信息并通知新集群中的锚点。

GS的位置管理。当GS节点在网络中改变其位置时，其入口卫星通知所有锚点GS节点的最新位置信息。通过这种方法，ISTN中的所有锚点都可以实时跟踪所有GS及其入口卫星的位置，从而避免在网络拓扑变化时的路由收敛过程。当卫星需要将流量转发到陆地服务器时，数据包首先转发到其锚点，然后到GS，最后到达目的地。

Handover Management

Intra-cluster handover. For a mobile user, their IP address is allocated by the corresponding anchor when it connects to the cluster and registers with the anchor. As the mobile user moves within a cluster, their anchor point, as well as allocated address, remains unchanged, ensuring the user’s higher-layer connections (e.g., TCP connections) stay active and stable.

Inter-cluster handover. Because satellites are inherently mobile, the user-to-anchor path might lengthen over time and increase latency. To avoid the excessive latency caused by far anchor points, SKYCASTLE triggers a cluster/anchor update to guarantee the user-perceived latency is within an expected threshold. The new anchor in the new cluster allocates a new address to the user. SKYCASTLE’s anchor manager takes care of the anchor assignment and guarantees user-perceived latency while avoiding inter-cluster handovers as much as possible.

集群内切换。对于移动用户，当其连接到集群并在锚点上注册时，其IP地址由对应的锚点分配。在用户在集群内移动时，其锚点以及分配的地址保持不变，确保用户的高层连接（例如TCP连接）保持活跃和稳定。

集群间切换。由于卫星本质上是移动的，用户到锚点的路径可能随时间延长并增加延迟。为了避免由于远距离锚点引起的过高延迟，SKYCASTLE触发集群/锚点更新，以确保用户感知的延迟在预期阈值内。新集群中的新锚点为用户分配一个新地址。SKYCASTLE的锚点管理器负责锚点分配，并在尽可能避免集群间切换的同时保证用户感知延迟。

Convergence-free Route Mechanism

Anchors also intercept and reroute packets based on location information. As described in §V-A, users’ IP addresses are allocated by satellite anchors while the fixed addresses of GSs are configured by the operator. Note that the addresses allocated by different anchors have different prefixes, which help SKYCASTLE to guide routing.

锚点还根据位置信息拦截和重定向数据包。如§V-A所述，用户的IP地址由卫星锚点分配，而GS的固定地址由操作员配置。注意， 由不同锚点分配的地址具有不同的前缀 ，这有助于SKYCASTLE引导路由。

Figure 7a plots the route from a GS to user. When the ingress satellite of GS receives a packet to a user, it determines which satellite anchor manages the user according to the prefix and generates a tunnel to the anchor. When the anchor receives the packet, it finds the ingress satellite of the user (SAT in Figure 7a) and generates a tunnel. Finally, SAT directly forwards the packet to the user by its forwarding information base (FIB) table. Further, Figure 7b plots the route from a user to GS. Traditionally, the topology fluctuates due to handovers in GSs may trigger route recalculation, thus the user-to-GS traffic is interrupted.

图7a显示了从GS到用户的路由。当GS的入口卫星接收到发送给用户的数据包时，它根据前缀确定哪个卫星锚点管理用户，并生成一个到锚点的隧道。当锚点接收到数据包时，它找到用户的入口卫星（图7a中的SAT），并生成一个隧道。最后，SAT直接通过其转发信息库（FIB）表将数据包转发给用户。此外，图7b显示了从用户到GS的路由。传统上，由于GS中的切换可能会触发路由重新计算，因此用户到GS的流量会中断。

In SKYCASTLE, anchors also manage the GSs’ locations to correctly forward packets without routing convergence. Specifically, when the ingress satellite receives a packet, the destination address of which is not allocated from any anchor, the satellite routes the packet to its anchor by default. Then the anchor determines the ingress satellite of the GS (SAT in Figure 7b) and generates a tunnel. When the packet reaches SAT, it is forwarded to the GS by a direct route.

在SKYCASTLE中，锚点还管理GS的位置，以在不需要路由收敛的情况下正确转发数据包。具体来说，当入口卫星接收到数据包时，如果其目的地址不是由任何锚点分配的，则卫星默认将数据包路由到其锚点。然后锚点确定GS的入口卫星（图7b中的SAT），并生成一个隧道。当数据包到达SAT时，它通过直接路由转发到GS。

Tip

感觉这里跟AS、BGP设计的场景一模一样 🔥