While thousands of satellites photograph Earth every day, most of that data never makes it to the ground because downlink bandwidth simply cannot keep up. Processing data in the Low Earth Orbit (LEO) zone offers promising capabilities to overcome this limitation. We propose SpaceCoMP, a MapReduce-inspired processing model for LEO satellite mesh networks. Ground stations submit queries over an area of interest; satellites collect sensor data, process it cooperatively at light-speed using inter-satellite laser links, and return only the results. Our compute model leverages space physics to accelerate computations on LEO megaconstellations. Our distance-aware routing protocol exploits orbital geometry. In addition, our bipartite match scheduling strategy places map and reduce tasks within orbital regions while minimizing aggregation costs. We have simulated constellations of 1,000-10,000 satellites showcasing 61-79% improvement in map placement efficiency over baselines, 18-28% over greedy allocation, and 67-72% reduction in aggregation cost. SpaceCoMP demonstrates that the orbital mesh is not merely useful as a communication relay, as seen today, but can provide the foundations for faster data processing above the skies.

翻译：尽管每天有数千颗卫星拍摄地球图像，但由于下行链路带宽无法跟上，大部分数据从未传回地面。在近地轨道区域处理数据为克服这一限制提供了前景广阔的能力。我们提出SpaceCoMP——一种受MapReduce启发的近地轨道卫星网状网络处理模型。地面站向目标区域提交查询；卫星采集传感器数据，通过星间激光链路以光速协同处理，并仅返回处理结果。我们的计算模型利用空间物理原理加速近地轨道巨型星座的计算。我们的距离感知路由协议充分利用轨道几何特性。此外，我们的二分图匹配调度策略将映射和归约任务部署在轨道区域内，同时最小化聚合成本。我们模拟了包含1000-10000颗卫星的星座，结果显示：相较于基线方法，映射任务部署效率提升61-79%；相较于贪心分配策略提升18-28%；聚合成本降低67-72%。SpaceCoMP证明轨道网状网络不仅可作为当今所见的通信中继，更能为构建天基高速数据处理体系奠定基础。