A Traffic Scheduling Scheme for Load Balancing in SDN-Based Space-Air-Ground Integrated Networks

Abstract

The deployment of Low-Earth-Orbit (LEO) constellations and Unmanned Aerial Vehicles (UAV) is an emerging trend for future network construction to improve global connectivity. To achieve fast global communication, inter-satellite links (ISL) need to be deployed between LEO satellites. However, unlike terrestrial networks, the deployment of space links requires a series of processes such as alignment, ranging, and tracking. Therefore, it becomes a challenge to design suitable routing schemes in the Space-Air-Ground integrated networks (SAGIN) for global data transmission in order to achieve full utilization of the transmission capacity. In this paper, we consider enhancing the transmission capability of SAGIN with Software Defined Networking (SDN) architecture. A load balancing based traffic scheduling scheme for SAGIN is designed. For each pair of source and destination nodes in SAGIN, the transmission capacity of the possible links is predicted. The traffic scheduling problem is turned into a modified maximum flow problem. Considering the dynamic and complicated SAGIN, the deep reinforcement learning model is utilized to make global optimal traffic scheduling decisions. The simulation results show that the proposed scheme can significantly improve the transmission capacity of the SAGIN while guaranteeing network load balancing.