Efficient usage of available network resources is a crucial factor for broadband services in interconnected satellite constellations. To meet required quality of service standards under heavy network loads, it is essential to optimize traffic distribution among the inter-satellite links. To address this challenge, we propose an adaptive traffic engineering framework based on deep reinforcement learning. Our approach employs a path-based decision-making strategy, using a centralized agent to distribute incoming flow requests on a set of candidate paths. This method approximates optimal solutions to the multi-commodity flow problem with relatively low computational complexity, making it suitable for in-space network control despite on-board processing limitations. The performance of the proposed scheme is evaluated against state-of-the-art rule-based benchmarks in various scenarios. We quantify the impact on performance of different candidate-path sets and traffic patterns. Overall, the proposed solution presents a viable approach for optimizing flow distribution in satellite constellation networks, suitable for the integration into the controller logic of software-defined networks.

Routing in satellite constellation networks with inter-satellite links has become an important aspect to enable broadband Internet access and to integrate into terrestrial networks. However, their dynamic characteristics and large physical size require specifically tailored solutions. To address these challenges, we propose and investigate a load-balanced routing protocol based on distributed software-defined networking. The approach relies on independent space-borne clusters with on-board controllers. Reduced signaling overhead is achieved by geographical inter-cluster routing algorithms. We evaluate the performance of the protocol in a custom-built system-level simulator, considering different architectures, design choices, and scenarios. Comprehensive comparisons with source-routed schemes and an upper benchmark demonstrate the viability of the solution. Notably, for the given scenario, the protocol handles network loads 97.4% higher than source-routing before quality of service compliance falls below 95%, while maintaining an average routing convergence of 117.338 ms . The work provides valuable in-depth insights into the design of optimized routing protocols for satellite constellation networks.