| dc.description.abstract | Due to the uneven distribution of networks scattered on the ground, network transmission resources are unevenly allocated in urban and suburban environments, making it difficult to meet the QoS requirements of different network services. Satellite assisted networking technologies are thus employed to enhance ground routing and promote satisfactory QoS requirements. The conceptual approach of this thesis involves utilizing low Earth orbit (LEO) satellites to assist data transmissions along with ground networks. The study in this thesis models network traffic paradigms and takes account of latency, packet loss rate, and bandwidth balance, all aimed at improving data transmission efficiency.
LEO satellites move rapidly along their orbits, and the inter-satellite link (ISL) frequently switches, causing packet loss during transmissions. Traditional packet loss recovery mechanisms lead to constant fragmentation, packet reassembly, and increased latency. To remedy these issues, it is necessary to perform dynamic routing adjustments or enhanced link connections to ensure accurate data transmissions. Hence, this study proposes an SDN-based QoS satellite routing method. The design of this method primarily revolves user demands for different network services, including multi-fold aspects like data flow latency, bandwidth, and packet loss rate, all centrally controlled through SDN. In contrast to conventional centralized routing, we account for satellite network mobility by designing data flows and satellites as dynamically changing topologies over time. Additionally, data streams are dynamically routed based on latency to achieve end-to-end data transmission paths, minimizing transmission delay and packet loss rate.
This study establishes a Mininet wireless environment combined with the Ryu simulation platform to construct an SDN-based satellite topology. Routing decisions are scripted in Python for measure evaluation of latency, packet loss rate, and throughput. Based on experimental results, our approach outperforms DSP, DRA, and calen routing methods, because of effectively reducing transmission delay and packet loss rate, especially under high demand scenarios. | en_US |