High performance enabled space object tracking via cloud computing

Space object tracking plays essential roles in communications, navigation, and observation. Many popular nonlinear estimation algorithms have been used to estimate the state of the space object. In this paper, we use improved distributed methods for both uncertainty propagation and space object tracking based on its essential parallel structure. Due to the large scale of space objects, we develop a high performance enabled cloud computing architecture for simultaneously propagating and updating the information uncertainty of space objects. Specifically, a cloud-based spark streaming framework is tailored to implement the space object tracking algorithm. A master node will split the data of the space objects to multiple slaver nodes. Each space object can be processed independently and the identifier of the space object is used as a key for reducer to join all separated data. Benefiting from this observation, we implement two methods to update the propagation and tracking of space objects in parallel with slave works. The simulation results reveal that cloud-based high performance enabled space object tracking is highly scalable and more efficient than the traditional space object tracking using a single CPU/GPU.