Carleton University

Purpose

To deliver packets between non-neighboring satellites using intermediate satellites, efficient routing protocols are required. There exist a high number of routing protocols in the literature, however, they were not designed for the highly dynamic, three-dimensional, fully distributed, and heterogeneous environment of the envisioned satellite networks. As the satellite network changes rapidly, it is necessary to keep learning the topology of the network in order to make the correct routing decisions. In this regard, online machine learning methods will play a significant role in improving the accuracy of routing decisions. In this project the team will use machine learning to determine routing algorithms for LEO satellite networks.