Abstract:In a low-Earth orbit (LEO) satellite network, the satellite operation speed is high, the operation cycle is short, and intersatellite links change dynamically. To sense the intersatellite link state in time and select the correct route for an intelligent routing decision, a dendritic network-based intelligent-aware routing algorithm for LEO satellites is proposed in this paper. This algorithm divides the intersatellite link routing of an LEO satellite network into situation-aware, quality-aware, and routing-decision stages and establishes a routing policy framework with real-time correction capability from the source node to the destination. This approach overcomes the problems of the limited selection of routing paths from fixed labels of existing deep learning-based routing algorithms and the long convergence time of reinforcement learning-based routing algorithms.In the intersatellite link situational awareness stage, the intersatellite visibility of the entire LEO satellite network is periodically obtained by analyzing the constraint conditions of the intersatellite link establishment. In the intersatellite link quality perception stage, the final output of the probabilistic forwarding matrix based on the ant colony algorithm is used as the label of the training set, and the corresponding intersatellite link quality is evaluated using the probability value of the current node by selecting the next hop node. By changing the weight coefficients in the path cost function under different load states, more effective training set label data can be collected, which can be consequently used to improve the performance of the trained dendritic network. Moreover, the training set can be optimized in real-time through semi-supervised learning. The trained dendritic network is used to analyze and process the link state parameters, perceive the comprehensive service quality of the link, and output the evaluation value matrix of the next hop routing. It is also used to automatically adjust the weight of the global satellite network link. Meanwhile, the traditional Dijkstra algorithm is optimized to realize the quality perception of the intersatellite link. In the routing decision stage, the reciprocal of the evaluation value matrix is used as the adjacency matrix to pass the shortest-path algorithm. Then, the initial routing path between the source and destination nodes is obtained. Finally, the initial path is corrected via periodic monitoring to cope with the failure of the satellite node. The simulation results show that the routing algorithm based on the dendritic network has low computational complexity and fast convergence. The algorithm can determine the status of the intersatellite link establishment in time, assess the quality of the intersatellite link in real-time, and automatically avoid congested satellite nodes. Accordingly, its end-to-end path delay, delay jitter, and packet loss rate are lower than those of the traditional heuristic routing algorithm and Dijkstra routing algorithm.

SDDRL-SR: A High-Reliability Satellite Routing Algorithm Based on Deep Reinforcement Learning

Distributed Routing Algorithm for LEO Satellite Network Based on Deep Reinforcement Learning

A Two-Hops State-Aware Routing Strategy Based on Deep Reinforcement Learning for LEO Satellite Networks

An Intelligent Routing Algorithm for LEO Satellites Based on Deep Reinforcement Learning

Multi-Commodity Flow Routing for Large-Scale LEO Satellite Networks Using Deep Reinforcement Learning

Fast-Convergence Reinforcement Learning for Routing in LEO Satellite Networks

Reinforcement Learning Based Intelligent Routing for Software Defined LEO Satellite Networks

Reinforcement learning based dynamic distributed routing scheme for mega LEO satellite networks

Research on Routing Optimization in Satellite Internet Based on Deep Reinforcement Learning

A Topology Design Method for Satellite Networks Based on Deep Reinforcement Learning

Enabling High-Throughput Routing for LEO Satellite Broadband Networks: A Flow-Centric Deep Reinforcement Learning Approach

LEO Satellite Intelligent-Sensing Routing Algorithm Based on a Dendrite Network

A Routing Optimization Method for Software-Defined SGIN Based on Deep Reinforcement Learning

An Intelligent SDWN Routing Algorithm Based on Network Situational Awareness and Deep Reinforcement Learning

QoS Routing Optimization Based on Deep Reinforcement Learning in SDN

Trustworthy and Load-Balancing Routing Scheme for Satellite Services with Multi-Agent DRL.

Software Defined Routing Algorithm in LEO Satellite Networks

DRL-Based Load-Balancing Routing Scheme for 6G Space–Air–Ground Integrated Networks

Inter-Satellite Routing for LEO Satellite Networks: A GNN and DRL Integrated Approach

A DRL Empowered Multipath Cooperative Routing for Ultra-Dense LEO Satellite Networks

DT-DDQNR: A Digital Twin Assisted Direct-to-Cell Satellite Network Intelligent Routing Algorithm