跳转至

FALCON: Towards Fast and Scalable Data Delivery for Emerging Earth Observation Constellations

Exploiting a constellation of small satellites to realize continuous earth observations (EO) is gaining popularity. Largevolume EO data acquired from space needs to be transferred to the ground. However, existing EO delivery approaches are either: (a) efficiency-limited, suffering from long delivery completion time due to the intermittent ground-space communication, or (b) scalability-limited since they fail to support concurrent delivery for multiple satellites in an EO constellation.

To make big data delivery for emerging EO constellations fast and scalable, we propose FALCON, a multi-path EO delivery framework that wisely exploits diverse paths in broadband constellations to collaboratively deliver EO data effectively. In particular, we formulate the constellation-wide EO data multipath download (CEOMD) problem, which aims at minimizing the delivery completion time of requested data for all EO sources. We prove the hardness of solving CEOMD, and further present a heuristic multipath routing and bandwidth allocation mechanism to tackle the technical challenges caused by time-varying satellite dynamics and flow contention, and solve the CEOMD problem efficiently. Evaluation results based on public orbital data of real EO constellations show that as compared to other state-of-the-art approaches, FALCON can reduce at least 51% delivery completion time for various data requests in large EO constellations.

利用小卫星星座实现连续对地观测（Earth Observations, EO）的技术正获得广泛关注。从太空获取的海量对地观测数据需传输至地面。然而，现有的数据传输方法存在两类局限:

（a）效率受限，间歇性的地空通信导致数据传输完成时间过长

（b）可扩展性受限，无法支持对地观测星座中多颗卫星的并发传输

为实现新兴对地观测星座中大数据的快速、可扩展传输，我们提出了一个名为FALCON的 多路径对地观测数据传输框架 。该框架巧妙地利用宽带星座中的多样化传输路径，以实现高效的协同数据传输

具体而言，我们构建了星座级对地观测数据多路径下载（Constellation-wide EO data Multipath Download, CEOMD）问题，其优化目标是最小化星座内所有对地观测源卫星的数据传输完成时间。我们证明了CEOMD问题是NP-难的，并进而提出了一种启发式多路径路由与带宽分配机制，以应对卫星时变动态特性与网络流量竞争所带来的技术挑战，从而高效地求解该问题

基于真实对地观测星座公开轨道数据的评估结果表明，与当前最先进的其他方法相比，在大型对地观测星座中，FALCON能针对不同数据请求将传输完成时间缩短至少51%

Introduction

Thanks to the recent technique breakthrough in the sensing and aerospace industry, earth observation (EO) technologies are evolving rapidly in the past decade. It is estimated that the revenues of EO data and services are forecast to double from roughly C2.8 billion to over C5.5 billion over the next decade [44].

Two critical trends can be observed from the recent evolution of the EO ecosystems. First, emerging EO satellites are equipped with multiple high-resolution sensors to capture EO data from space [12], [13], [32] for various missions, e.g., finegranularity environment monitoring and disaster prediction. Second, many EO service providers tend to leverage a large number of EO satellites (i.e., a constellation) to cooperatively execute EO missions. The revisit time can be significantly reduced by cooperatively using EO satellites in a constellation for observation. This is because EO satellites typically work in low earth orbit (LEO) close to the earth surface and move at a high velocity, and the revisit time of a single EO satellite could be very long (e.g., several hours or days) [11], [50].

Taking the above two trends together, emerging EO constellations are generating a large volume of data every day [21]. In urgent cases like disaster response, EO data acquired in space needs to be downloaded to the ground mission center as fast as possible. Therefore, optimizing the data delivery process for EO constellations is critical for the EO industry.

得益于传感和航空航天工业近期的技术突破，对地观测（Earth Observation, EO）技术在过去十年中发展迅速。据估计，在未来十年内，对地观测数据和服务的市场营收将从约28亿加元翻倍增长至超过55亿加元[44]。

从近期对地观测生态系统的演进中，可以观察到两个关键趋势。首先，新兴的对地观测卫星装备了多种高分辨率传感器，用于从太空捕获数据以执行多样化的任务[12], [13], [32]，例如细粒度的环境监测和灾害预测。其次，许多对地观测服务提供商倾向于利用大量的对地观测卫星（即星座）来协同执行观测任务。通过协同使用星座中的卫星，可以显著缩短重访时间。这是因为对地观测卫星通常在靠近地球表面的低地球轨道（LEO）上高速运行，单颗卫星的重访时间可能非常长（例如，数小时或数天）[11], [50]。

综合以上两个趋势，新兴的对地观测星座每天都在产生海量数据[21]。在灾害响应等紧急情况下，太空中获取的对地观测数据需要尽快下载到地面任务中心。因此，优化对地观测星座的数据传输过程对于整个行业至关重要。

Many prior works have studied the methods for delivering data from EO satellites. At a high level, existing methods can be divided into three categories: (a) downloading via ground station networks [48]; (b) downloading via geostationary (GEO) relays [40]; and (c) downloading via LEO satellite routes [25]. Specifically, the first category exploits a collection of distributed ground stations to download EO data when the satellite carrying data moves into the transmission range of a certain ground station. However, this method fails to sustain long-duration data download and suffers from high delivery completion time due to the limited deployment of available ground stations and intermittent ground-space communication. The second category leverages geostationary (GEO) satellite relays to download EO data to the ground via persistent and geostationary forwarding paths. While robust, this method has very limited scalability since it can not support large-scale EO constellations due to the limited amount of transmission links in GEO relays. Although the third category inter-connects satellites by inter-satellite links (ISL) to establish high-speed download paths from the EO satellite to its ground destinations, this method is not fast and scalable enough due to: (a) the path contention when the number of EO sources increases; and (b) frequent network disruptions caused by the high-velocity movement of satellites. Therefore, we ask a pragmatic question: is there a viable path to deliver big EO data collected from emerging EO constellations to the ground in a fast and scalable manner?

许多先前的工作已经研究了从对地观测卫星传输数据的方法。

从宏观层面看，现有方法可分为三类：

（a）通过地面站网络下载[48]

（b）通过地球静止轨道（GEO）中继下载[40]

（c）通过低地球轨道（LEO）卫星路由下载[25]

具体而言，第一类方法利用一组分布式地面站，在携带数据的卫星进入某个地面站的传输范围时进行数据下载。然而，由于可用地面站的部署有限以及地空通信的间歇性，该方法无法维持长时间的数据下载，并导致较长的传输完成时间。

第二类方法利用地球静止轨道（GEO）卫星中继，通过持久且固定的转发路径将数据下载到地面。这种方法虽然稳健，但由于GEO中继的传输链路数量有限，其可扩展性非常有限，无法支持大规模的对地观测星座。

第三类方法通过星间链路（ISL）将卫星互连，以建立从观测卫星到其地面目的地的高速下载路径，但该方法仍然不够快速和可扩展，原因在于：（a）随着观测源数量的增加，路径竞争问题加剧；（b）卫星的高速移动导致频繁的网络中断。

因此，我们提出了一个实际问题：是否存在一种可行路径，能够以快速且可扩展的方式，将从新兴对地观测星座收集的海量数据传输到地面？

In this paper, we affirmatively answer the question above by presenting FALCON, a multipath EO delivery framework. Falcon mainly adopts two key ideas to enable fast and scalable big data delivery for EO constellations: (a) download big EO data in an on-demand way. That is, it is unnecessary to wait for all the satellites to finish downloading all the data they’ve collected, we only download the requested EO data that is related to the area of interest. This can greatly reduce the amount of data we need to transfer. (b) exploit the high multipath diversity in broadband mega-constellation networks to concurrently establish multiple delivery paths. In this way, the download throughput can be improved while enhancing the ability to resist the impact of frequent handover between satellites and ground stations.

However, the dynamic fluctuations in the core infrastructure of mega-constellation networks involve new challenges on applying multipath download for EO tasks: the time-varying topology fluctuations accordingly result in route and traffic fluctuations. Moreover, carrying a large number of EO delivery flows can lead to excessive congestion with shared bottleneck links. Hence, multiple routes should be dynamically and wisely re-calculated in a time-varying manner, and flows should be adaptively re-scheduled in each route, to avoid link congestion and obtain transmission efficiency.

在本文中，我们通过提出FALCON —— 一个多路径对地观测数据传输框架，对上述问题给出了肯定的回答。FALCON主要采用两个核心思想，为对地观测星座实现快速、可扩展的大数据传输：

（a）以按需方式下载大数据。也就是说， 无需等待所有卫星完成其全部采集数据的下载，我们只下载与目标区域相关的被请求数据。这可以极大地减少需要传输的数据量

（b）利用 宽带巨型星座网络中的高多路径分集特性来并发地建立多条传输路径 。通过这种方式，可以在提高下载吞吐量的同时，增强抵抗卫星与地面站之间频繁切换所带来影响的能力

然而，巨型星座网络核心基础设施的动态性给应用多路径下载于对地观测任务带来了新的挑战：时变的拓扑结构会导致路由和流量的相应波动。此外，承载大量对地观测传输流可能导致共享瓶颈链路的过度拥塞。因此，必须以时变的方式动态且智能地重新计算多条路由，并自适应地在每条路由上重新调度流量，以避免链路拥塞并获得高传输效率。

Our mechanism addresses the above challenges in two steps. First, we model the dynamic and hybrid constellation topology, as well as its time-varying network capacity and availability. Combining EO constellations and broadband mega-constellations, we formulate the Constellation-wide EO data Multipath Download (CEOMD) problem, which aims at minimizing the time when all the sources complete the transmission of requested data. We also demonstrate the hardness of solving CEOMD problem under representative EO scenarios.

Second, to solve the CEOMD problem efficiently, we propose a Heuristic Multipath Routing and Bandwidth Allocation (HMRBA) algorithm. The key principle behind our algorithm is to select paths with low inter-path link overlap to avoid severe network congestion in a greedy order where the source with more data comes first, and preferably choose those paths that have longer stable time to transmit data. When allocating bandwidth for each path, we proportionally distribute the bandwidth resources according to the data volume of each source avoiding the ones with much data from costing too much time to transfer.

我们的机制分两步应对上述挑战。首先，我们对动态的混合星座拓扑及其时变的网络容量和可用性进行建模。结合对地观测星座和宽带巨型星座，我们构建了星座级对地观测数据多路径下载（Constellation-wide EO data Multipath Download, CEOMD）问题，其目标是最小化所有源卫星完成请求数据传输的时间。我们还证明了在典型的对地观测场景下求解CEOMD问题的困难性。

其次，为高效求解CEOMD问题，我们提出了一种启发式多路径路由与带宽分配（Heuristic Multipath Routing and Bandwidth Allocation, HMRBA）算法。该算法的核心原则是：按照数据量从大到小的贪心顺序，优先选择路径间链路重叠度低且稳定时间更长的路径以避免严重的网络拥塞。在为每条路径分配带宽时，我们根据每个源的数据量按比例分配带宽资源，以避免数据量大的源花费过多的传输时间。

We evaluate FALCON by simulations based on real and public constellation information. Extensive evaluation results demonstrate that by dynamically constructing multiple download paths and judiciously allocating EO traffic upon them, FALCON can reduce at least 51% of the delivery completion time when serving various data requests in large EO constellations, as compared to other state-of-the-art approaches,.

我们基于真实且公开的星座信息，通过仿真对FALCON进行了评估。大量的评估结果表明，通过动态构建多条下载路径并审慎地分配流量，与当前最先进的其他方法相比，FALCON在为大型对地观测星座处理各种数据请求时，能够将传输完成时间缩短至少51%

In conclusion, this paper makes three contributions.

• (a) Exposing the importance and challenges of optimizing data delivery for EO constellations, with a formulation of the Constellation-wide EO data Multipath Download (CEOMD) problem, which is NP-hard.

• (b) Presenting a novel framework called FALCON to achieve fast and scalable constellation-wide EO data download, by adopting a Heuristic Multipath Routing and Bandwidth Allocation (HMRBA) algorithm.

• (c) Demonstrating the effectiveness of FALCON by extensive simulations based on real and public information obtained from the satellite ecosystem.

总而言之，本文做出了三点贡献:

（a）揭示了优化对地观测星座数据传输的重要性与挑战，并构建了 NP-Hard 的星座级对地观测数据多路径下载（CEOMD）问题

（b）提出了一个名为FALCON的新颖框架，通过采用启发式多路径路由与带宽分配（HMRBA）算法，实现了快速且可扩展的星座级对地观测数据下载

（c）通过基于从卫星生态系统获取的真实公开信息进行的大量仿真，验证了FALCON的有效性

EO 卫星

绝大多数的EO卫星都运行在 LEO (Low Earth Orbit)，即低地球轨道

  1. LEO轨道通常指距离地面 200 - 2000km 的空间
  2. 为获得高分辨率的图像，EO卫星的轨道高度会更低一些，普遍在 500-800km 之间
LEO vs. MEO vs. GEO
  1. LEO: 200 - 2000km
  2. MEO: 2000 - 35786km
  3. GEO: 35786km