Switching-Aware Dynamic Control Path Planning for Software Defined Large-Scale LEO Satellite Networks with GEO Controllers

Recently, to acquire the programmability, flexibility and re-configurability of network, the technologies of software-defined networking (SDN) are utilized to design new architectures for LEO satellite networks, where control plane is realized by several GEO satellites. But in the previous work, it is assumed that each LEO satellites could directly connect with the GEO controllers. This is unreasonable when the LEO data plane is a large-scale LEO satellite constellation, as the number of GEO controllers and antennas of each GEO controller are limited. In this work, we propose a design of software defined large-scale LEO satellite networks with limited GEO controllers, where the communication between GEO control plane and LEO data plane is achieved by limited cross-layer links. In this model, the selection of cross-layer links directly affects the load balance among GEO controllers and the latencies of control paths, which will deeply influence the latency of routing response. Thus, we propose a switching-aware dynamic control path planning scheme to handle this, where we propose a switching-time selection scheme to handle the impact of switching and a multi-objective optimization problem to deal with the selection of cross-layer links, and finally present a particle swarm optimization based algorithm to solve it. Simulation results demonstrate the better performance on reducing the bad influence of switching, optimizing the load balance among controllers and the average latencies of control paths.