Background on Nanosatellite Constellations

Momentum away from exquisite [89], monolithic satellites towards small, cheap nanosatellites reduces the cost of remote sensing in space by orders of magnitude. A nanosatellite has a mass between 1 kg and 10 kg, often adhering to the “CubeSat” standard [69] to enable use of commercial, off-the-shelf (COTS) components and avoid custom deployers [75]. A cubesat is physically constrained to 10 cm×10 cm×10 cm (“1U”) volumes, with mass limited to 1.33 kg per 1U. This volume must house all sensors, actuators, and communication subsystems. Computers onboard existing nanosatellites are simple, low-performance systems for command and data handling (C&DH), guidance navigation and control (GNC), buffering sensor data, and communication. Virtually all nanosatellites today rely on a ground control segment to manage data.

A nanosatellite electrical power system (EPS) collects, stores, and distributes energy. Many low-cost cubesats avoid higher-risk, deployable solar arrays and instead rely on surfacemounted solar panels. As a result, the small size of the satellite constrains collected power to a few watts. Batteries must be small due to limited cubesat volume and mass. To prevent damage, batteries are heated in the cold of space (e.g. -40°C), incurring a power cost overhead [27]. Supercapacitor storage is less energy-dense, but has less mass, less volume, and avoids thermal issues; we focus on supercapacitor energy storage.

Figure 1 illustrates the magnitude of the shift from large, monolithic satellites to nanosatellites. Monolithic satellites are meters in size, thousands of kilograms in mass, collect kilowatts of power, and can cost over half a billion USD [21]. A nanosatellite is four orders of magnitude smaller (cubic centimeters), has three orders of magnitude less mass (a few kilograms), collects three orders of magnitude less power (a few watts), and has four orders of magnitude lower cost.

A nanosatellite constellation is a collection of nanosatellites that share a purpose. Existing nanosatellite constellations are coordinated from the ground, often to accomplish a remote sensing task (e.g. imaging the Earth). Today, commercial ventures leverage the relatively low per-device cost of nanosatellites to operate large constellations [12, 59]. In the future, public and private organizations expect to launch constellations of thousands of devices, each with high-datarate sensors and the capacity for more capable onboard computers.

A constellation consists of a ground segment and a space segment. In bent-pipe architectures, the ground segment consists of geographically-distributed, manually-controlled transceivers, and the space segment consists of remote--controlled satellites in one or more orbits. As we show quantitatively in Section 3.3 and Section 7.1, bent pipes break down as the amount of edge-sensed data increases. Further, limited link availability and bitrate bottlenecks can cause reconfiguration of a constellation to take days, weeks, or months [22]. These growing limitations of bent-pipe architectures motivate the OEC techniques presented in this work.

从“精致”[89]的单体式卫星向小型、廉价的纳米卫星发展的趋势，已将空间遥感的成本降低了数个数量级。一颗纳米卫星（nanosatellite）的质量介于1公斤至10公斤之间，通常遵循“立方体卫星”（CubeSat）标准[69]，以便使用商用现成品（COTS）组件并避免定制化的部署器[75]。

一颗立方体卫星的物理尺寸被限制在10厘米×10厘米×10厘米（即“1U”）的体积内，每1U的质量限制为1.33公斤。所有传感器、执行器和通信子系统都必须容纳在这个体积内。现有纳米卫星上的计算机是简单的低性能系统，用于指令与数据处理（C&DH）、制导导航与控制（GNC）、缓冲传感器数据以及通信。如今几乎所有的纳米卫星都依赖地面控制部分来管理数据。

纳米卫星的电子电源系统（EPS）负责收集、存储和分配能量。许多低成本的立方体卫星为规避风险，不采用可展开式太阳能帆板，而是依赖表面贴装的太阳能电池板。因此，卫星的小尺寸将可收集的功率限制在几瓦。由于立方体卫星的体积和质量有限，电池也必须很小。为防止损坏，电池在太空的寒冷环境（例如-40°C）中需要被加热，这会产生额外的功率开销[27]。超级电容器的储能密度较低，但质量更轻、体积更小，并能避免热管理问题；我们的研究重点关注使用超级电容器的储能方案。

图1展示了从大型单体卫星到纳米卫星的转变幅度。单体卫星尺寸达数米，质量为数千公斤，可收集数千瓦的电力，成本可能超过五亿美元[21]。而一颗纳米卫星在尺寸上小了四个数量级（立方厘米），质量上少了三个数量级（几公斤），收集的功率少了三个数量级（几瓦），成本则低了四个数量级。

纳米卫星星座（nanosatellite constellation）是为实现共同目的而协作的一组纳米卫星。现有的纳米卫星星座从地面进行协调，通常是为了完成遥感任务（例如地球成像）。如今， 商业公司正利用纳米卫星相对较低的单位设备成本来运营大型星座[12, 59]。 未来，公共和私营组织期望发射由数千个设备组成的星座，每个设备都将配备高数据率的传感器和能力更强的星上计算机。

一个星座由地面部分和空间部分组成。在“弯管”（bent-pipe）架构中，地面部分由地理上分散、手动控制的收发器组成，而空间部分则由一个或多个轨道上受远程控制的卫星组成。正如我们在第3.3节和第7.1节中定量展示的，随着边缘感知数据量的增加，“弯管”架构会逐渐失效。此外，有限的链路可用性和比特率瓶颈可能导致星座的重新配置耗费数天、数周甚至数月[22]。正是“弯管”架构这些日益增长的局限性，催生了本研究中提出的OEC（轨道边缘计算）技术。