BUILDING A NETWORK

Given five laser links per satellite and knowledge of orbits, we can now approach the coupled problems of how to build a network, and how to route on that network. The first question is which satellites should we interconnect with lasers?

鉴于每颗卫星配备的五个激光链路及其轨道信息，我们可以开始探讨如何构建网络以及如何在该网络上进行路由的问题。第一个需要解决的问题是：如何选择卫星之间的激光互连方式？

A dense LEO constellation like Starlink has two main advantages over terrestrial networks. First, it can connect almost anywhere, however remote. Second, the speed of light in a vacuum, c, is ≈ 47% higher than in optical fiber. The ability to connect anywhere is important, but we speculate that providing low-latency wide area communication will be where the money to maintain and operate such a network is made, connecting cities that are already well connected using optical fiber, but with lower latency as a premium service. Already there are new private microwave relay links between New York and Chicago[11], London and Frankfurt[1], and London and Paris. These links have relatively low capacity compared to fiber, but are of high enough value to the finance industry to be worth building new low latency links.

低地球轨道（LEO）星座（如Starlink）相比于地面网络具有两个主要优势。首先，它能够连接到地球上几乎任何位置，无论多么偏远。其次，真空中光速（c）比光纤中的传播速度高约47%。尽管连接任何位置的能力极为重要，但我们推测，该网络的主要经济来源将是提供低延迟的广域通信服务。这种服务将面向已经通过光纤良好连接的城市，以低延迟作为高端服务的卖点。目前，已经有针对金融行业需求的低延迟微波中继链路，例如连接纽约和芝加哥[11]，伦敦和法兰克福[1]，以及伦敦和巴黎的链路。尽管这些链路的容量相较光纤较低，但它们对于金融行业的高价值已经足以驱动新的低延迟链路的建设。

商业价值

1. 范围覆盖全球
2. 低延迟（虽然它的链路容量相对较低）

Starlink’s LEO satellites will be in 1,110 to 1,325 km orbits. Although much lower than GEO, this is still too high to provide lower latency than fiber over shorter distances. However, over longer distances the extra latency getting between Earth and the nearest satellite may be more than offset by routing around the world between the satellites at c. The primary goal then, seems to be to connect key population centers with satellite paths that run close to the great circle route.

Starlink的LEO卫星运行在1,110至1,325千米高度的轨道上。虽然比地球同步轨道（GEO）低得多，但仍然过高，以至于在短距离内无法提供比光纤更低的延迟。然而，在更长的距离上，信号从地球到最近卫星的额外延迟（性能拖油瓶）可能会被绕地球的真空路径中较高的光速（性能红利）所抵消。因此，Starlink的主要目标是利用 接近大圆航线的卫星路径 连接 主要人口中心。

Great circle Route

大圆航线是指在球体（如地球）上，两点之间的最短距离。当在地球仪上绘制时，这条路线看起来是弯曲的，但它代表了旅行或通信的最有效路径。

Let us first consider the 1,600 satellites in phase 1. To maintain good paths, most laser links must be up at any time. This constrains the solutions. Except at the extreme north and south extents of their ground track, in any one region, half the satellites are traveling on a northeasterly track and half are on a southeasterly track 1 . All are traveling at ≈7.3km/s. A NEbound satellite will not remain in range of a SE-bound satellite for long, so a laser between the two must track rapidly as the orbits cross, and must rapidly switch to a new satellite as the old one moves away. ESA’s EDRS can bring up its optical link in under a minute[13]. Starlink may be quicker, given the shorter distances, but connections will not be instant.

在第一阶段的1,600颗卫星中，为了维持良好的路径，大部分激光链路必须始终处于连接状态。这限制了潜在的解决方案。在任一区域内，除地面轨迹的极北或极南范围外，约有一半卫星沿东北方向移动，另一半沿东南方向移动。这些卫星的速度约为7.3公里/秒。东北方向移动的卫星与东南方向移动的卫星的交会时间很短，因此它们之间的激光链路需要快速调整以跟踪彼此的相对运动，并在旧卫星离开时迅速切换到新的卫星。

例如，欧洲航天局（ESA）的EDRS系统可以在不到一分钟的时间内建立光学链路[13]。鉴于Starlink的通信距离较短，其切换速度可能会更快，但仍然无法实现即时连接。

From the point of view of any one satellite, two neighbors always remain in the same locations: the next one ahead on the same orbital plane, and the one behind on that orbital plane. Laser links to these neighbors only need to fine-tune（微调） their orientation, so these are the obvious candidates for the first two laser links. To form a network, we also need to link between different orbital planes. There are many options for how to do this. However, only the satellites in the neighboring orbital planes remain consistently in range, so connecting to these is the next priority so as to form a network where most of the links have high uptime.

从任何一颗卫星的角度来看，有两个邻近卫星的位置始终保持不变：轨道平面上的前一颗卫星和后一颗卫星。与这些邻近卫星的激光链路只需进行微小调整即可维持连接，因此它们是第一个两个激光链路的理想候选。

为了形成网络，还需要在不同轨道平面之间建立连接。尽管存在多种连接方式，但只有相邻轨道平面的卫星始终保持在通信范围内，因此与这些卫星的连接是下一个优先级。这种连接方式可以形成一个高可用性的网络。

Routing forwards and backwards along the orbital planes already provides good SW ↔ NE and NW ↔ SE connectivity, so it makes most sense to use the next pair of lasers to connect between the orbital planes in as orthogonal a direction as possible: either north-south or east-west. With a phase offset between orbital planes of 5/32, connecting satellite n on orbital plane p to the nearest satellites (n+1 on orbital plane p+1 and satellite n − 1 on plane p − 1) is not the best solution, as these links nearly parallel those of the crossing orbital plane paths. Rather, connecting satellite n on orbital plane p to satellite n on plane p + 1 and also to satellite n on plane p − 1 provides very good east-west connectivity, while the 5/32 phase offset ensures than the links are slightly offset from running exactly east-west, providing very direct paths in a wider range of nearly east-west directions.

在轨道平面内的前后连接已经能够提供良好的 西南↔东北 和 西北↔东南 方向的连接。因此，最优选择是利用另外两个激光链路在轨道平面之间建立尽可能正交的方向连接，例如东西或南北方向。

然而，由于轨道平面之间具有 \(5/32\) 的相位偏移，连接轨道平面 \(p\) 的第 \(n\) 颗卫星到轨道平面 \(p+1\) 和 \(p-1\) 上的第 \(n+1\) 和第 \(n-1\) 颗卫星并不是最佳选择，因为这些链路几乎平行于交叉轨道路径。而将轨道平面 \(p\) 的第 \(n\) 颗卫星连接到轨道平面 \(p+1\) 和 \(p-1\) 的第 \(n\) 颗卫星，则能够提供非常好的东西向连接。同时，\(5/32\) 的相位偏移确保了这些链路与完全东西向的路径略有偏移，从而在更广泛的东西方向范围内提供了直接路径。

It is also possible to provide reasonable north-south connectivity, but as most of the world’s population in developed nations that are more likely to be willing to pay for latency are clustered in a band from 30°to 55°North, providing east-west connectivity seems to be the higher priority for phase 1.

尽管也可以提供合理的南北方向连接，但由于大多数发达国家的人口主要集中在北纬30°到55°之间，而这些地区更可能为低延迟服务支付费用，因此在第一阶段中，东西向连接被认为是更高的优先级。

The network resulting from this use of each satellite’s first four laser links provides a good mesh network, but in any one region their are two distinct meshes - one moving generally northeast and the other moving southeast, with no local connectivity between the two without going the long way round the planet. Our simulations show that most traffic can route without switching between the two meshes, but using the final laser to provide inter-mesh links improves the routing options significantly, even if such lasers are down frequently, while they re-align from one crossing satellite to another.

通过每颗卫星的前四个激光链路所形成的网络构建了一个良好的网状网络。然而，在任何一个区域内，这种网络实际上由两个彼此独立的子网组成——一个整体上向东北方向移动，另一个向东南方向移动。两个子网之间没有本地连接，若需通信则必须绕行地球。我们的仿真结果表明，大多数通信流量可以无需在两个子网之间切换即可完成路由，但如果利用最后一个激光链路来提供跨网间的连接，则可以显著改善路由的灵活性，即便该链路因需要频繁重新对准交会卫星而导致连接中断的情况较多。

This way of aligning the lasers is shown from the point of view of one satellite traveling northeast in Figure 4. The forward and backwards links remain in a constant orientation; the side links track very slowly as the satellite orbits, but always connect to the same neighboring satellite and always point close to an east-west orientation; the final link tracks crossing satellites very rapidly indeed. Figure 5 shows how the side laser links used this way provide good east-west connectivity; Figure 6 shows all the lasers.

这种激光链路的分配方式如图4所示，从一颗向东北方向移动的卫星的视角来看，其前向和后向的激光链路始终保持恒定的方向；侧向的激光链路随着卫星轨道的运行而缓慢调整，但始终连接到相同的邻近卫星，并且始终指向接近东西方向的方位；而最后一个激光链路则需要非常快速地跟踪交会的卫星。图5展示了通过这种侧向激光链路如何实现良好的东西向连接；图6展示了所有激光链路的完整分布情况。