Satellite-Based Computing Networks with Federated Learning

Driven by the ever increasing penetration and proliferation of data-driven applications, a new generation wireless communication, sixth (6G) mobile system enhanced artificial intelligence, has attracted substantial research interests. Among various candidate technologies 6G, low Earth orbit (LEO) satellites have appealing characteristics ubiquitous access. However, costs satellite communication (SatCom) are still high, relative to their counterparts ground networks. To support massively interconnected devices with intelligent adaptive learning reduce expensive traffic in SatCom, we propose federated (FL) LEO-based We first review state-of-the-art SatCom related machine (ML) techniques, then analyze four possible ways combining ML The performance proposed strategies is evaluated simulation results reveal that FL-based computing networks improve overheads latency. Finally, discuss future topics along this direction.