Due to the rapid development of the space–air–ground integrated network (SAGIN), a satellite communication system has the advantages of wide coverage and low requirements for a geographical environment and is gradually becoming the main competitive technology for 6G. The low-earth-orbit (LEO) satellite network has the characteristics of low transmission delay, small propagation loss, and global coverage, and its exploration has become the main research object of contemporary satellite communications. However, traditional routing algorithms cannot adapt to the characteristics of the high dynamics and load-balancing requirements of LEO satellite networks. In this paper, a load-balancing routing algorithm for LEO satellites based on Deep Q-Network (DQN-LLRA) is proposed by using deep reinforcement learning. Making use of the model obtained by the DQN training, satellite nodes can select the best routing results according to the delay, bandwidth, and queue utilization of the surrounding satellite nodes. The simulation and analysis show that the path load obtained by the proposed algorithm is low. Compared with the Q-learning-based algorithm, this algorithm reduces the maximum queue utilization rate of the routing path by 5%, reduces the average queue utilization rate of the routing path by 13%, and effectively balances the load in the network.
With the rise of a new generation of low Earth orbit (LEO) satellite constellations and the advancement of the 6G network, satellite–terrestrial integrated Internet of Things (IoT) in the future will achieve global coverage through the integration of LEO constellation, and extend computing to the edge of the network through the deployment of edge computing services in LEO constellation so as to meet the demand of mass connection and low latency data processing. The LEO constellation network of the future will be an edge cloud network combining network and computing. In this paper, we propose a computation offloading strategy for the combined optimization of energy and computational load in a LEO constellation edge cloud network (hereinafter referred to as LEO-ECN). First, we establish the LEO-ECN system model, in which the user task can be offloaded to the satellite through the multi-hop path. Then, a cost model considering energy consumption and load calculation is proposed. Finally, a joint optimization problem to minimize energy consumption and balance the LEO-ECN load is established, which is a convex optimization problem. The simulation result demonstrates that, compared with the benchmark strategy, our proposed strategy has better performance and can improve the computing resource utilization of LEO-ECN.
Digital twin technology provides a reliable paradigm to address the high trial-and-error costs and limited perception capabilities in satellite networking. However, the dynamic constellation topology and real-time twin applications remain significant challenges in satellite network design. This paper proposes a network topology simulation approach that dynamically analyzes the inter-satellite topology based on pre-calculated ephemeris and orbital information. Furthermore, the paper introduces a digital twin algorithm based on network virtualization, cloud platform management, and software-defined networking to validate and analyze the twin requirements at different stages. Finally, a low Earth orbit (LEO) constellation twin validation environment is constructed to verify the networking protocols at various stages. The experimental results demonstrate the performance of the proposed twin systems at different stages.
Large language models (LLMs) have demonstrated powerful capability to solve practical problems through complex step-by-step reasoning. Specifically designed LLMs have begun to be integrated into terrestrial communication networks. However, relevant research in the field of satellite communications remains exceedingly rare. To address this gap, we introduce SCNOC-Agentic, a novel architecture especially designed to integrate the management and control of satellite communication systems in LLMs. SCNOC-Agentic incorporates four components tailored to the characteristics of satellite communications: intent refinement, multi-agent workflow, personalized long-term memory, and graph-based retrieval. Furthermore, we define four typical real-world scenarios that can be effectively addressed by integrating with LLMs: network task planning, carrier and cell optimization, fault analysis of satellites, and satellite management and control. Utilizing the SCNOC-Agentic framework, a series of open-source LLMs have achieved outstanding performance on the four tasks under various baselines, including zero-shot CoT, CoT-5, and self-consistency. For example, qwen2.5-70B with SCNOC-Agentic significantly improves the parameter generation accuracy in the network task planning task from 15.6% to 32.2%, while llama-3.3-70B increases from 16.2% to 29.0%. In addition, ablation studies were conducted to validate the importance of each proposed component within the SCNOC-Agentic framework.
The transition from physical network functions (PNFs) to virtual network functions (VNFs) and cloud-native network functions (CNFs) represents a dynamic advancement in virtualization technologies, addressing limitations in flexibility, prolonged deployment times, and scalability. CNFs encapsulate functionalities in lightweight, containerized units, prioritizing cloud-native principles of agility and resilience. Effective management of CNFs is paramount for modern network infrastructures, and Open Source Management and Orchestration (OSM) frameworks aim to tackle these challenges. However, certain constraints persist, particularly in large-scale applications with complex startup sequencing. These challenges are further complicated by the modular nature of CNFs and the dynamic resource allocation in cloud-native environments. To address these issues, we introduce Homcs, a hierarchical orchestration mechanism for CNF services in 5G Core Networks. Homcs leverages Helm for orchestrated deployment and implements a RESTful Information Exposure Interface within Kubernetes. Real-world experiments have demonstrated that Homcs enhances CNF orchestration by optimizing deployment, scalability, and resource allocation within OSM environments.
Satellite TCP accelerators operate over high-latency and error-prone links, where packet-level faults are commonly investigated through manual capture analysis. We present a packet-level fault-diagnosis framework that deterministically extracts 41 scalar statistics from IPv4, TCP, and ICMP headers and packet timestamps. Twenty-six diagnostic rules map these statistics to candidate fault nodes whose signatures can be inferred from single-point pcap traces within the satellite TCP accelerator fault taxonomy. Knowledge-graph, BM25, and vector retrieval provide evidence for a locally executed quantized large language model, which generates structured diagnostic reports. Rule implementation was verified with 26 positive and 26 boundary-negative cases. On 113 strict single-fault traces, the rule engine achieved 86.73% Top-1 accuracy and a 95.58% candidate-set hit rate. On a 104-trace same-generation comparison, Random Forest achieved 89.42% Top-1 accuracy, compared with 87.50% for the rule engine and 71.15% for RBF-SVM; the difference between Random Forest and the rule engine was not significant (exact McNemar p = 0.8318). On 42 independently generated Linux network-stack fault traces covering seven representative fault classes, the rule engine achieved 95.24% accuracy, compared with 57.14% for Random Forest and 0% for RBF-SVM, and rejected five normal controls. In a 30-case report comparison, the complete configuration preserved all upstream candidate sets and introduced no unsupported fault identifier.