最新刊期
卷 41 ， 期 7 ， 2025
- 摘要：Firstly, the trends of vehicle intelligence and connectivity were analyzed, along with the new requirements and challenges imposed on in-vehicle communication by the evolution of automotive electrical electronic architecture (EEA). It was pointed out that vehicular optical communication would undergo a development process similar to the “fiber replacing copper” transition in telecom fixed access networks, shifting from “narrowband communication + copper wiring harnesses” to “broadband communication + optical fiber harnesses”. And then, a technical roadmap and standardization proposal for vehicular passive optical network (V-PON) were presented. To address the demands of in-vehicle communication and the challenges posed by its unique operating environment, adaptive improvements were made for existing PON technologies, and key technical hurdles were overcome to establish the V-PON standard framework.This approach fully leverages China’s technological and industrial advantages in optical communication and PON systems, facilitating the development of core intellectual property in vehicular communication. It supports the transformation and upgrading of intelligent connected vehicles (ICV) and enhances the global competitiveness of China’s automotive and telecommunications industries.关键词：intelligent connected vehicle;in-vehicle network;EEA;V-PON;copper wire harness;optical fiber harness7499|7170|0更新时间：2025-08-07
Expert Views
- 摘要：In the context of high-power jamming scenarios, the anti-jamming ability of communication systems can be enhanced by deploying reconfigurable intelligent surface (RIS). In order to maximize the signal to interference-plus-noise ratio(SINR) of the legitimate signals, a joint optimization problem of the transmit beamforming vector and the reflection coefficient matrix of RIS was formulated, considering multiple constraints such as transmit power and reflection phase shift of RIS. Specifically, in the case where line of sight communication is not blocked, an optimization scheme based on alternating iteration to address the complex coupling between high-dimensional variables in optimization problems was proposed. By decomposing the original high-dimensional, multi-variable and strongly coupled non-convex fractional programming problem into multiple sub-problems, and then transforming each non-convex sub-problem into a solvable convex optimization problem through methods such as multi-dimensional complex quadratic transformation and semi-definite relaxation, the solution was obtained efficiently. Extensive simulations analysis validated that the proposed scheme could significantly enhance the communication anti-jamming performance of wireless communication systems and exhibit excellent robustness and adaptability under various system parameters and network node configurations.关键词：reconfigurable intelligent surface;anti-jamming communication;alternate optimization;convex optimization6324|370|0更新时间：2025-08-07
- 摘要：As a new communication paradigm, semantic communication can differentiately extract and transmit task-related semantic information to improve network efficiency. However, semantic communication requires the network to actively participate in information processing, provide computing and storage resources, and at the same time requires the network to have the ability to perceive the importance of information and its own resource conditions. The development path of the integration of network and information processing was discussed. Fisrtly, the evolution of network capabilities and the development history of semantic communication were introduced. Then, how the network transformed from a simple data transmission pipeline to an information processing platform was analyzed from three aspects: firstly, task-oriented semantic communication can extract semantic information according to intelligent task requirements; secondly, the network supports the training, storage and reasoning of semantic communication models by scheduling cloud-edge resources; thirdly, the coordinated evolution of network and semantic communication enables semantic communication to understand the network status, and the network has the ability to intelligently respond to needs of semantic communication. Finally, typical application scenarios of the integration of network and information processing were introduced.关键词：semantic communication;data transmission;information processing329|1210|0更新时间：2025-08-07
- 摘要：In recent years, wireless mobile consumers have also grown exponentially. Due to the openness and broadcasting of the wireless networks, it is easier for unauthorized users to disguise themselves as legitimate users to launch spoofing attacks, resulting in insecurity in the low-altitude industry. Relying on specific cryptographic algorithms alone in the tradition was difficult to cope with password cracking abilities of high-intensity and high-computility. Meanwhile, a secure mechanism for covert communication was urgently needed. A chaotic sequences based watermarking wireless physical layer security authentication algorithm was proposed. This algorithm utilized the transmission of chaotic watermark and carrier information together to achieve covert transmission of service data, and physical layer identity authentication was completed without occupying additional frequency band resources of the system. Through analysis of simulation experiments, this method could embed watermark information by using different modulation methods, and the physical layer security authentication performance was significantly improved via adjusting the watermark power coefficient.关键词：chaotic system;low-altitude economy;wireless physical layer security authentication;spectrum wat-ermarking135|298|0更新时间：2025-08-07
- 摘要：The requirements of the emerging smart grid pose a challenge for power line carrier communication systems in low-voltage networks. However, a large-scale dual-mode communication mesh network, with its advantages, such as high reliability and throughput, holds promise for widespread application in power communication systems. Prior to constructing the new network, it is essential to perform a capacity pre-assessment. Yet, building an accurate analytical model for capacity analysis is complicated because of numerous interrelated factors. To address these challenges, the uplink capacity of dual-mode communication in large-scale mesh networks was investigated. Firstly, a throughput model for multi-hop transmission was established based on the device hierarchy structure within the sub-net mesh. Constraints such as frequency hopping and device half-duplex mechanism were then introduced into the model, and a satisfaction function was proposed to address the issue of unfair resource allocation for information flows, forming a convex optimization problem for throughput in one sub-net. Then, considering the constraints of sustained coefficients between multiple subnets, an integer-constrained optimization problem for throughput in multiple sub-nets was formulated, and the optimal solution represented the uplink capacity of the network. The simulation results show that the network throughput of dual-mode communication is at least 50% higher than that of the single mode communication. The proposed resource allocation algorithm improves throughput by 10% compared with traditional resource allocation algorithms. By improving the network concurrency ability and the receiving ability of the master node, a significant gain in network load capacity can be achieved.关键词：distribution IoT;dual-mode communication;resource allocation;frequency hopping;mesh network126|382|0更新时间：2025-08-07
- 摘要：Conducting research on the factors influencing victimization risks in generative AI (GAI) driven telecom network fraud holds significant theoretical and practical implications for areas such as summarizing patterns of criminal behavior and enhancing technological defense capabilities. For this purpose, simulation experiments were carried out based on real AI fraud case information. The criminal process was deconstructed into three stages: forged information generation, dissemination, and impact of forged information. Latent variables such as GAI, data flow, data packets, network behavior, and network risk were extracted. An analytical framework was then constructed by combining structural equation modeling theory to systematically quantify the influence paths and contribution degrees of different elements on victimization risk. The findings revealed that GAI had a significant direct effect on network risk, and the direct effect played a dominant role in the overall effect. The mediating effects of data flow and packet characteristics were weak, and its role in the influence path was not significant.关键词：generative artificial intelligence;telecom network fraud;structural equation model;quantitative analysis265|839|0更新时间：2025-08-07
- 摘要：With the rapid development of aerospace, aviation and ground communication technologies, it has become an inevitable trend for the development of communication networks to build an integrated space-terrestrial integrated network with global seamless coverage, on-demand access and on-demand services.The realization of the space-terrestrial integrated network based on the standard space optical transport network protocol was studied. Firstly, it is clear that the space optical transport network is the fundamental cornerstone of building the integrated network, the status quo of the standards of space optical communication was described, and the dilemma of the limited satellite information exchange caused by the differences in systems was analyzed. Furthermore, the networking function of the space optical communication network was deeply explored at the protocol level. In view of the characteristics of high-speed satellite movement, frequent inter satellite link switching, and limited on-board resources, a new frame structure and reliable transmission protocol adapted to the space optical transport network were proposed to support the interconnection and communication of multiple constellations, and improve the efficiency of network resource utilization while ensuring accurate and reliable data transmission. Finally, the broad prospects of this network architecture for industrial applications were summarized, future research should focus on the layout of unified space optical transport network standardization, so as to lay a solid foundation for the efficient operation and wide application of the space-terrestrial integrated network.关键词：space optical transport network;frame structure;6G;space-terrestrial integrated network;interconnection186|461|0更新时间：2025-08-07
- 摘要：Contrastive learning has shown excellent performance in semantic embeddings by capturing relationships between data samples to enhance model representation. However, its effectiveness largely depends on constructing positive samples and selecting appropriate objective functions. Positive samples must be carefully designed to ensure the model can identify meaningful similarities while reducing noise. To address this, a novel method that constructed positive samples by splitting, encoding, aggregating, and projecting text was proposed. The text was broken into segments, encoded to extract semantic content, aggregated to highlight relationships, and projected into a semantic space optimized for learning. Additionally, two supervised loss functions were designed, complementing the standard contrastive loss, to enhance the discriminability of the semantic space and thereby improve the model’s discrimination ability. The experimental results show that this method performes well on two public datasets and one private dataset, significantly improving the quality of semantic embedding, solving the core challenges of contrastive learning, and laying the foundation for further applications in the field of natural language processing.关键词：contrastive learning;sentence embedding;semantic textual similarity;text classification;joint loss function151|567|0更新时间：2025-08-07
- 摘要：Network intrusion detection system has gained attention as an effective means of combating various cyber threats. However, there are a lot of redundant information and unbalanced distribution problems in network intrusion data, therefore, deep learning and support vector machine-recursive feature elimination-based network intrusion detection model (DLRF) was proposed. The features were sorted by the support vector machine-recursive feature elimination algorithm and the important features were selected. Moreover, both oversampling and under-sampling techniques were utilized to tackle the unbalance problem of data sample distribution. Three deep learning algorithms were used to build the base learner of the ensemble framework, and deep neural network was used to build the meta-learner, so as to improve the performance of DLRF model to detect network attacks. The proposed framework was experimented with two publicly available and popular network traffic datasets, namely UNSW-NB15 and CICIDS-2017. The accuracy rates of the DLRF model on these two datasets are 0.906 8 and 0.996 8 respectively, and the F1-score are 0.906 8 and 0.996 0.关键词：intrusion detection model;deep learning;recursive feature elimination;ensemble learning176|553|0更新时间：2025-08-07
Research and Development
- 摘要：In order to solve the problems of low computing efficiency utilization, poor stability, high difficulty in training optimization, and imperfect domestic accelerator technology ecology in AI cluster model training with more than 10 000 NPU, a large language model training optimization solution based on a completely domestic AI cluster was proposed. Through automatic distributed strategy recommendation, pipeline parallel optimization, overlap optimization and full-link profiling technology, the model FLOPS utilization (MFU) reached 45.13% when training a 405B large language model on 16 384 domestic NPU, which was more than 10% higher than the baseline performance. At the same time, a set of stability assurance mechanisms was built throughout the entire large language model training process to achieve real-time monitoring of key indicators before and during model training, as well as the ability to quickly diagnose faults after training task were interrupted. The experimental results show that the large language model training optimization solution proposed can effectively improve the utilization of computing efficiency, and has important guiding significance for the future construction of domestic AI cluster and large language model training.关键词：AI cluster with more than 10 000 cards;domestic NPU accelerator card;model training optimization237|1673|0更新时间：2025-08-07
- 摘要：Large language model training is a pivotal scenario in AI development. Under the trend of diversified and heterogeneous computing power, the cross-ecosystem heterogeneous computing power collaboration capability will become the key support for training at the hundred-thousand-card-scale. Based on this background, a heterogeneous AI computing power mixed training system was designed, which could automatically detect and adapt to heterogeneous AI chips, enabling collective communication among heterogeneous computing powers. Based on the prototype system, heterogeneous training was implemented using three types of AI chips in a RoCEv2-interoperable cluster. In the heterogeneous pipeline parallelism (PP) training scenario, peak training efficiency reached 99.77% using NVIDIA and Biren GPU, and 99.03% using NVIDIA, Iluvatar, and Biren GPU. For heterogeneous data parallelism (DP) training, the optimal mixed training efficiency between NVIDIA and Biren GPU reached 92.88%.关键词：large language model;collective communication;heterogeneous parallelism;heterogeneous hybrid training208|741|0更新时间：2025-08-07
- 摘要：A data and knowledge-driven stability assurance scheme for such clusters was proposed to address the issues of frequent hardware failures, persistently high task training failure rates, and difficulties in cross-domain problem localization within ultra-large intelligent computing clusters with over ten thousand computing cards. The cluster performance data was collected by employing heterogeneous resource integrated collection technology and distributed real-time big data ETL techniques. Fault diagnosis was performed using an enhanced SA-BiLSTM deep learning model, improving the explainability of diagnostic model outputs via knowledge graph analysis and matching for the generation of fault diagnosis reports. In the process of extracting time series features with the deep learning model, weighted fusion of features extracted at different scales , thereby improving the accuracy of the fault diagnosis model. In fault diagnosis simulation experiments conducted on an 18 000-card cluster, it was observed that the loss value gradually converged and stabilized at 0.047, achieving an accuracy rate of 98.4%. Practical has shown that the proposed stability assurance scheme can effectively support large-scale model training and enhance the reliability of intelligent computing clusters, providing a solid foundation for the construction of larger-scale intelligent computing clusters and the training of large models in the future.关键词：intelligent computing cluster;fault diagnosis;SA-BiLSTM;knowledge graph214|864|0更新时间：2025-08-07
- 摘要：The intelligent computing center uses distributed file storage for data preprocessing and model training, distributed object storage for the acquisition of raw data and model release, and distributed block storage to provide storage for the resource management platform. Meanwhile, high-performance distributed file storage is used to shorten the read and write time of checkpoint during the training process and improve the training efficiency of the cluster. The entire life cycle of large model training requires data copying and migration between storage systems with different storage protocols and different read-write performances, resulting in duplicate data storage. Additionally, data copying requires computing resources and network bandwidth. To address the above issues and provide a unified namespace for the intelligent computing clusters, a file and object converged storage and a hierarchical file storage scheme were proposed to solve the problem of data transfer between different storage protocols and enable automatic data flow between high-performance file storage (all SSD) and ordinary-performance file storage (SSD and HDD), providing a reference for the optimization of storage systems of ultra-large-scale intelligent computing clusters.关键词：converged storage;hierarchical storage;distributed file storage;distributed object storage145|584|0更新时间：2025-08-07
- 摘要：The existing mainstream post-quantum key encapsulation mechanisms, such as the module-lattice-based key encapsulation mechanism (ML-KEM), rely on the module learning with errors (Module-LWE) problem associated with structured lattices. The algebraic structure of these mechanisms may lead to reduction vulnerabilities. The SCLOUD+ scheme, which was focused on within the unstructured LWE framework, achieved significant compression of public key and ciphertext sizes through high-dimensional lattice coding gain, based on the recursive construction properties of the Barnes-Wall (BW) lattice. Moreover, a dimension-specific full-unfolding recursive elimination technique for BW lattices was proposed. Through compile-time constant optimization, hierarchical hard-coding strategies, and single-instruction, multiple-data (SIMD) friendly memory layouts, the decoding clock cycles of the BW lattice in a 128-dimensional scenario were reduced from 147 798 to 30 107, providing core support for the SCLOUD+ post-quantum key encapsulation mechanism. This research provided a lightweight key encapsulation mechanism (KEM) paradigm that balances efficiency and quantum-resistant security for intelligent computing networks, laying a crucial technical foundation for low-latency scenarios such as distributed federated learning.关键词：intelligent computing network;Barnes-Wall lattice;bounded distance decoding;post-quantum cryptography;key encapsulation250|697|0更新时间：2025-08-07
- 摘要：Addressing key issues such as low efficiency of resource pool orchestration and scheduling, insufficient integration of cloud-network-security technologies in the current development of computing network security services, against the research background of the cloud-network-security collaborative innovation and operation pressure faced by operators in the security service market, a computing network security resource pool orchestration and scheduling technology system based on segment routing over IPv6 (SRv6)+ application responsive network (ARN）was proposed. Through in-depth integration of the SRv6 protocol, the research realizes end-to-end connectivity between the network and the service, and through ARN, it improved the simplicity of data plane identification and dynamic scheduling capabilities, constructing a network data plane with intelligent orchestration and scheduling capabilities to support the flexible combination and rapid deployment of security services. The research results mainly include the SRv6+ARN orchestration and scheduling technology architecture, key technologies, and reliability guarantees, providing technical support for operators to build a cloud-network-security collaborative computing network security resource pool.关键词：security orchestration and scheduling;SRv6;ARN;computing power network;security resource pool144|577|0更新时间：2025-08-07
- 摘要：The construction of artificial intelligence inference centers has become a hotspot in the current development of intelligent computing centers. Evaluating the inference business capability of intelligent computing centers solely based on the scale of intelligent computing power is no longer accurate. A quantitative evaluation method for the inference business capability of intelligent computing centers was proposed by establishing three models: a delay-insensitive business model, a delay-sensitive business model, and a user access business model. This approach aims to achieve alignment between construction and requirements during the construction phase, thereby improving investment efficiency.关键词：intelligent computing center;inference;business capability evaluation121|628|0更新时间：2025-08-07
Topic
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
- Total Visits：1035510 Visits Today：3559
0