FALCON Design

We propose FALCON (Fast, and ScALable Constellation-wide Earth ObservatioN), a multi-path EO delivery framework aiming at enabling fast and scalable EO data delivery for EO constellations.

我们提出了 FALCON (Fast, and ScaLable Constellation-wide Earth ObservatioN)，一个旨在为对地观测（EO）星座实现快速且可扩展数据传输的多路径EO数据传输框架。

A. FALCON Overview

To avoid a long time to wait for visible ground stations as in L2D2, we leverage space routes over inter-satellite links (ISLs) and ground-satellite links (GSLs) to achieve fast and scalable EO data delivery for constellations. However, it involves two challenges in practice. First, EO and broadband satellites are highly dynamic. Routes from the source to the terrestrial destination are likely to be interrupted by inevitable satelliteground handoffs. Thus, it is difficult to make the download process stable and guarantee that on-demand requests can be satisfied on time. Second, as the number of sources with request data in EO constellations expands, it is challenging to properly schedule a large number of download flows from EO satellites to the terrestrial destination under a highly-dynamic network topology to avoid bandwidth contention.

为避免像L2D2那样长时间等待可见地面站，我们利用基于星间链路（ISL）和星地链路（GSL）的空间路由，以实现星座级EO数据的快速、可扩展传输。然而，这在实践中涉及两个挑战。首先，EO卫星和宽带卫星都具有高度动态性。从源端到地面目的地的路由很可能因不可避免的星地切换而中断。因此，难以使下载过程保持稳定，并保证按需请求能够被准时满足。其次，随着EO星座中持有被请求数据的源卫星数量增多，在高度动态的网络拓扑下，要合理调度从EO卫星到地面目的地的大量下载流以避免带宽竞争，是极具挑战性的。

To make the data transmission robust and satisfy EO requests as fast as possible, FALCON explores the path diversity of mega-constellation networks, and exploits a dynamic multi-path download mechanism to aggregate bandwidth and attain stable transmission. Emerging megaconstellations like Starlink and Kuiper follow the Walker Delta constellation design [49], where the ascending nodes of the planes are distributed over the full range of 360 degrees, and satellites are evenly spaced in their orbits. Therefore, a mega-constellation network is essentially an evenly distributed network with high path diversity [26]. In other words, there exists a number of shortest paths between the same source and destination in a mega-constellation. Exploiting multi-path delivery enables two advantages for EO download: (1) the source EO satellite can access more bandwidth by using multiple space paths simultaneously and attain faster data download; (2) if multiple download paths exist, the resilience to the handover events can be improved since EO traffic can quickly switch to an alternate path when a satellite-ground interruption occurs.

为了使数据传输更具鲁棒性并尽快满足EO请求，FALCON探索了巨型星座网络（mega-constellation）的路径分集特性，并利用一种动态多路径下载机制来聚合带宽和实现稳定传输。新兴的巨型星座如 Starlink 和 Kuiper 遵循沃尔克-德尔塔（Walker Delta）星座设计[49]，其中轨道平面的升交点分布在360度的整个范围内，且卫星在其轨道内均匀间隔。因此，巨型星座网络本质上是一个具有高路径分集的均匀分布式网络[26]。换言之，在巨型星座中，同一源和目的地之间存在多条最短路径。

利用多路径传输为EO下载带来两个优势:

（1）源EO卫星可以通过同时使用多条空间路径来获取更多带宽，从而实现更快的数据下载

（2）如果存在多条下载路径，对切换事件的弹性可以得到提升，因为当发生星地链路中断时，EO流量可以迅速切换到备用路径

To handle the circumstances where there are tens or even hundreds of sources all with requested data to transmit, we propose a heuristic EO multipath routing and bandwidth allocation algorithm to judiciously schedule a large number of download flows in the dynamic mega-constellation network, avoiding link congestion and sustaining high throughput for each EO tasks at scale. Notice that, in our framework, instead of waiting for all data collected from space to finish downloading, we only download the requested data of AoIs.

为应对存在数十甚至数百个源卫星都需要传输被请求数据的情况，我们提出了一种启发式的EO多路径路由与带宽分配算法，以便在动态的巨型星座网络中审慎地调度大量下载流，从而避免链路拥塞并为每个EO任务大规模地维持高吞吐量。请注意，在我们的框架中，我们只下载被请求的兴趣区域（AoI）数据，而不是等待从太空收集的所有数据都完成下载。

B. System Architecture and Use Phases

Fig. 2 plots the high-level system architecture of FALCON. Collectively, FALCON contains an EO mission center which is operated by the EO service provider, and a hybrid constellation integrating EO and broadband satellites to cooperatively perform the download tasks. The mission center processes the user requests, like the government needs information for disaster emergency response, which contains the AoI locations and the request time duration, assigns download tasks to the EO constellation, and delivers required contents back to the users. EO satellites in the hybrid constellation are equipped with high-volume ISLs.

图2展示了FALCON的顶层系统架构。总体而言，FALCON包含一个由EO服务提供商运营的EO任务中心，以及一个集成了EO卫星和宽带卫星的混合星座，以协同执行下载任务。

1. 任务中心负责处理用户请求（例如政府需要灾害应急响应信息），这些请求包含兴趣区域（AoI）的位置和请求的时间范围
2. 而后，任务中心将下载任务分配给EO星座，并将所需内容交付给用户

混合星座中的EO卫星装备有大容量的星间链路

At runtime, EO satellites continuously gather information from various regions around the world and save EO data in their own storage. Every image is tagged with the location information and the timestamp when the source takes the photo. When the request for information on a specific area is received by the mission center, the mission center calculates which EO satellites contain the required data by checking which has passed over the AoI in the request duration. This can be achieved by reverse extrapolation of the satellite’s trajectory using the public two-line-element (TLE) information. After that, the mission center assigns download tasks to all EO satellites carrying the required information via ground station networks or LEO networks, and invokes the multipath routing and bandwidth allocation mechanism to establish high-throughput download sessions while avoiding severe link congestions. In practice, the routing mechanism can be achieved by using source routing based on the routing protocols in the network and the bandwidth allocation part can be implemented by Traffic Control (TC) tools. Finally, when all related data have been gathered by the mission center, it combines and sends them back to the user. In the following sections, we introduce the details of our multipath routing and bandwidth allocation algorithm.

在运行时，EO卫星持续从全球不同区域收集信息，并将EO数据保存在各自的存储中。每张图像都标记有位置信息和拍摄时的时间戳。

当任务中心收到针对特定区域的信息请求时，它会通过检查哪些卫星在请求的时间段内经过了该兴趣区域，来 计算出哪些EO卫星存有所需数据 。这可以通过使用公开的 TLE 信息对卫星轨道进行反向推算来实现。

之后，任务中心通过地面站网络或LEO网络向所有携带所需信息的EO卫星 分配下载任务 ，并 调用多路径路由与带宽分配机制来建立高吞-吐量的下载会话 ，同时避免严重的链路拥塞。

在实践中，路由机制可以通过基于网络中路由协议的源路由来实现，而带宽分配部分则可以通过流量控制（TC）工具来实施。最后，当任务中心收集完所有相关数据后，会将其整合并发送给用户。

在接下来的章节中，我们将详细介绍我们的多路径路由与带宽分配算法。