最新刊期
卷 42 ， 期 5 ， 2026
- 摘要：The development of communication networks over several decades was systematically reviewed, and the law that communication networks undergo a landmark major revolution approximately every decade was proposed.Starting from the initial circuit switching, there were three revolutions: the IP-based revolution, the IT-based revolution, and the ongoing AI-driven revolution. These three revolutions with a focus on the AI-driven one were summarized and analyzed: the IP-based revolution enabled highly efficient data transmission by evolving from circuit switching to packet switching and from IP-based switching to end-to-end IP networking, the IT-based revolution, driven by technologies such as software defined network (SDN) and network functions virtualization (NFV) etc, upgraded communication network architectures from closed and rigid to open and elastic, the key points of the AI-driven revolution were “network for AI” and “AI for network”. The former aimed to construct a high-performance interconnection base for ultra-large-scale intelligent computing, while the latter integrated AI deeply into network systems, promoting the evolution of network operation, network maintenance, as well as applications and services towards a goal-driven, fully autonomous “intelligence-native” paradigm. Finally, the prospect of the future development of communication networks was prospected, and it was proposed that the future “AI+” era would drive the transformation and upgrading of communication networks from mere connectivity infrastructure to intelligent and service-oriented new information infrastructure.关键词：communication network;IP-based revolution;IT-based revolution;AI-driven revolution75|42|0更新时间：2026-06-17
Expert Views
- 摘要：In extremely large-scale multiple-input multiple-output (XL-MIMO) systems, accurate acquisition of hybrid-field channel state information (CSI) remains one of the key challenges for high-rate transmission in future 6G networks. To overcome the accuracy limitation caused by fixed grid partitioning in conventional hybrid-field channel estimation schemes, a two-stage off-grid hybrid-field channel estimation algorithm was developed. In the first stage, the far-field angular domain and near-field polar domain were jointly sparsely represented. By traversing the ratio between far-field and near-field paths and allocating the path quota, sparse gradient pursuit was applied to a joint dictionary to alternately search for far-field and near-field atoms, while an incremental residual update based on a row-wise least meansquare (LMS) algorithm was employed to obtain a coarse estimate of the hybrid-field channel. On the basis of the initial support, a Newton iteration combining numerical gradients and line search were used to refine continuous parameters such as path angles and distances by the second stage, thereby reconstructing the complete hybrid-field channel. Simulation results demonstrate that, for different signal-to-noise ratios and numbers of user antennas, the normalized mean square error (NMSE) of the proposed scheme is consistently lower than that of conventional hybrid-field channel estimation algorithms, and it achieves a performance gain of 1.5~3 dB compared with an existing off-grid stochastic gradient pursuit (SGP) algorithm.关键词：extremely large-scale multiple-input multiple-output (XL-MIMO);hybrid-field channel;channel estimation;off-grid algorithm;incremental residual6|0|0更新时间：2026-06-17
- 摘要：Dynamic heterogeneous redundancy (DHR) architectures enhance system resilience but majority-voting decisions struggle with heterogeneous source quality and potential conflicts, leading to performance drops or misjudgments in critical security scenarios. The adaptive and robust decision redundancy (ARDR) architecture was presented, which combined adaptive service selection with evidence-based robust decision-making to systematically evaluate and fuse heterogeneous executor outputs. ARDR employed a comprehensive evaluation model considering executor heterogeneity, historical performance, confidence, and efficiency, using the whale optimization Algorithm to select optimal combinations, and a dempster-shafer-based high-conflict-aware framework to handle uncertainty. Prototype and multi-scenario simulations show that ARDR outperforms conventional DHR and improved dynamic heterogeneous redundancy (IDHR) in accuracy, F2-score, and execution time, especially under high redundancy and conflict, with ablation studies confirming the synergy of adaptive selection and robust decision-making. The ablation study provides an interpretable and efficient solution for redundancy system design in heterogeneous, high-uncertainty environments.关键词：dynamic heterogeneous redundancy;mimic defense;adaptive service selection;evidence fusion;uncertainty reasoning;moving target defense3|0|0更新时间：2026-06-17
- 摘要：To address the challenge of collaborative optimization between multi-service quality of service (QoS) guarantee and load balancing in satellite networks, an intelligent multi-service routing algorithm integrating long short-term memory with deep Q-Network was proposed. The algorithm deployed LSTM-DQN (LDQN) as an agent at the current satellite node, where the network state input to the agent includes link latency, bandwidth, packet loss rate, traffic, network topology and service type between the current node and its neighboring nodes. The output action was the next-hop node selection for the current node. The reward function, designed to guide action adjustment, combined the weighted sum of bandwidth, latency, and packet loss rate with the weighted sum of maximum link bandwidth utilization. After the agent’s training convergence, multi-service transmission was performed in the satellite network. Simulation experiments and performance evaluations demonstrate that the proposed algorithm achieves significant improvements across various performance metrics while exhibiting outstanding load balancing capabilities.关键词：satellite network;deep reinforcement learning;load balancing;multi-service routing4|0|0更新时间：2026-06-17
- 摘要：In complex electromagnetic countermeasure scenarios, the time-varying noise interference of wireless channels and multi-domain coupling effects cause the noise distribution characteristics of the test data set to deviate from the prior assumption conditions of the training set, leading to feature mismatch in deep neural network models and subsequently causing performance degradation of modulation recognition systems based on the static channel assumptions in the cross-domain scenarios. To address this challenge, a dynamic noise automatic modulation recognition method based on contrastive learning and real-complex domain fusion was proposed. In the pre-training stage, the bootstrap your own latent (BYOL) contrastive learning framework was utilized to construct a real-complex domain fusion network, forcing the model to deeply understand the intrinsic structure of the data through self-supervised learning, thereby reducing the sensitivity of feature extraction to the changes in noise distribution and enhancing the model's generalization ability under different signal-to-noise ratio conditions. In the fine-tuning stage, the complex time-frequency spectrum generated by short-time Fourier transform was input into the real-complex domain fusion network to extract the multi-dimensional features of signal, enabling the network to learn the essential features that independent of channel noise. The combination of these two strategies enabled the model to effectively cope with the dynamic noise interference under different signal-to-noise ratio conditions. Experimental results show that when the signal-to-noise ratio of the signal to be recognized decreases by 6 dB, the proposed method achieves at least a 23.98% improvement in recognition accuracy compared with the existing methods, such as vision Transformer (ViT).关键词：contrastive learning;feature fusion;dynamic noise;automatic modulation classification4|0|0更新时间：2026-06-17
- 摘要：With the evolution of 6G network architecture toward a distributed autonomous paradigm, traditional centralized security mechanisms face significant challenges, such as boundary dissolution, susceptibility to identity forgery and data leakage. In response, a distributed secure and trustworthy mesh (DSTM) system was proposed. By constructing logical security boundaries and implementing dynamic security policies, the DSTM enables multi-level identity authentication, end-to-end secure connectivity, security isolation, and dynamic enforcement of security policies among distributed subnets. The DSTM can be deeply integrated with the 6G distributed architecture, providing valuable insights for the design of future 6G network security systems.关键词：6G;distributed autonomous network;DSTM;network boundary;security anchor49|28|0更新时间：2026-06-17
- 摘要：Time series prediction was recognized as having significant application value in critical fields such as finance, power, and networks. Deep learning models were demonstrated to possess strong fitting capabilities for this task, but their performance was found to be heavily dependent on structural design and hyperparameter selection. Traditional parameter-tuning methods, such as grid search and manual experience, were criticized for their low efficiency and tendency to fall into local optima. To address these issues, a deep time series prediction model based on the BiTCN-BiGRU-Attention architecture was constructed, and a novel metaheuristic optimization algorithm, RIME, was introduced for optimization. The RIME was designed to simulate the natural growth mechanism of rime ice, combining soft rime search strategies, hard rime piercing mechanisms, and positive greedy selection strategies to achieve an effective balance between global exploration and local exploitation. In the experimental section, the algorithm’s performance was comprehensively evaluated on standard benchmark functions and multiple real-world datasets. The results show that the RIME-optimized prediction model was superior to the unoptimized model in terms of accuracy, convergence speed, and stability. New insights and practical pathways were provided for the efficient and automated optimization of deep time series prediction models.关键词：RIME;deep learning;hyperparameter optimization;time series prediction4|0|0更新时间：2026-06-17
- 摘要：To address the issue of limited communication resources in maritime operation areas, a maritime-assisted Internet of things (IoT) system model based on non-orthogonal multiple access (NOMA) and supported by dual unmanned surface vehicles (USV) was proposed. In this model, the system throughput was maximized through the joint optimization of user communication scheduling and transmission power. Simulation results demonstrate that, compared with orthogonal multiple access (OMA) technology, the NOMA-based dual-USV IoT system achieves higher throughput, and necessary communication support for maintenance personnel on offshore wind turbines is provided through the deployment of USV characterized by low cost and on-demand configurability. Furthermore, the introduction of NOMA technology was intended to enhance spectrum resource utilization, and the bottleneck of scarce spectrum scarcity at sea was expected to be broken through.关键词：IoT;USV;NOMA;resource allocation;throughput maximization50|32|0更新时间：2026-06-17
- 摘要：Intelligent reflecting surface (IRS) can effectively enhance channel quality and improve system capacity by optimizing the phase of the reflection matrix, making the design of high-performance and low-complexity phase optimization algorithms a critical issue. For the problem of maximizing the channel capacity in IRS-assisted multiple-input multiple-output (MIMO) systems, a closed-form update formula for the real-valued phase of each IRS element was derived directly based on the system capacity, which differed from the conventional approaches that targeted the channel gain. On this basis, two low-complexity phase optimization algorithms, namely fixed-step and adaptive-step methods, were proposed. The adaptive-step algorithm further accelerated convergence by dynamically adjusting the phase update magnitude while maintaining the capacity performance of the fixed-step algorithm. Simulation results demonstrate that, under the transmit power of 10 dBm and with 100 IRS elements, the proposed algorithm achieves a capacity improvement of approximately 5 bit/s over the dimensional sine maximization (DSM) method, with the number of iterations and average runtime reducing by 81.4% and 78.3%, respectively. Compared with the fixed-step algorithm, the adaptive-step algorithm further reduces the number of iterations and average runtime by 49.2% and 42.1% without sacrificing the capacity performance, and exhibits more pronounced low-complexity advantages in large-scale IRS deployments.关键词：IRS;MIMO;phase optimization;low complexity4|0|0更新时间：2026-06-17
- 摘要：The RF parameters of devices in two adjacent frequency band systems, namely 5G private networks and vehicle to everything(V2X) direct communication were measured, and the system scenario parameters and channel models based on relevant standards and theoretical models were studied. Through simulations, the analytical conclusions and recommendations regarding the compatibility, coexistence conditions, and interference coordination between V2X systems and 5G private network systems in different application scenarios were provided, it has certain reference value for the collaborative development of the industrial internet and intelligent connected vehicles.关键词：V2X direct communication;5G private network;compatibility and coexistence;interference coordination4|0|0更新时间：2026-06-17
- 摘要：To address the challenges of complex traffic characteristics, insufficient utilization of packet-level semantic information, and the trade-off between detection accuracy and efficiency in software-defined networking (SDN) environments, a hybrid detection model that integrates payload information and flow-level statistical features was proposed. Specifically, an improved Mamba network was employed to perform deep modeling of payload sequences, enabling the extraction of temporal dependencies and contextual semantics within packet-level features. Meanwhile, the conventional convolutional structure was replaced with Ghost convolution to effectively reduce the number of parameters and computational cost while maintaining strong feature representation capability. Finally, a self-attention mechanism was introduced to adaptively weight and fuse multi-dimensional features, which enhanced the representation of critical attack patterns and suppresses irrelevant information. Experimental results demonstrate that the proposed model achieves a detection accuracy of 99.56% on the CICDDoS2019 dataset with an average detection latency of only 0.21 ms, outperforming existing mainstream methods. Moreover, validation on multiple public datasets further confirms the strong generalization capability of the proposed model.关键词：software defined network;DDoS attack detection;deep learning;self-attention mechanism;network security4|0|0更新时间：2026-06-17
- 摘要：An electricity trading risk assessment method based on an adaptive weighting and context-aware mechanism was proposed, which significantly enhanced prediction accuracy and risk management capabilities through a multi-model fusion strategy. The method incorporates an adaptive weight adjustment mechanism that dynamically optimizes the weights of linear regression, random forest, and long short-term memory(LSTM) models by combining historical performance with time decay factors. Additionally, it employs a context-aware switching logic triggered by market volatility, extreme weather, or policy events to dynamically adjust model weights in response to sudden scenarios. A meta-learner optimization strategy using gradient boosting decision tree(GBDT) integrates the predictions of base models with market features to improve robustness in complex scenarios. Experimental results demonstrate that the hybrid model outperforms individual models across key metrics, including mean square error(MSE) of 21.3, mean absolute error(MAE) of 3.1, and R² of 0.89. The adaptive weighting mechanism contributes to a 10% performance improvement, while context-aware switching further reduces errors by 5%. In extreme scenarios such as daily price volatility exceeding 20%, prolonged high temperatures, or sudden policy changes, the model significantly mitigates errors through dynamic weight adjustments. The hybrid model not only enhances returns but also reduces risks. This method provides a scientific risk management tool for electricity markets, enhancing market stability and economic efficiency. It holds potential for extension to financial risk assessment domains and future exploration in federated learning, real-time optimization, and rule automation.关键词：electricity trading risk assessment;machine learning;linear regression;random forest;LSTM network41|34|0更新时间：2026-06-17
- 摘要：In the flying Ad Hoc network (FANET) environment, nodes exhibit high mobility, which easily leads to frequent changes in network topology. To address the optimal path selection problem of routing protocols in flying Ad Hoc networks under high-speed mobility conditions, a routing optimization scheme that applied the Chinese pangolin optimization (CPO) algorithm to FANET was proposed. An intelligent routing decision-making mechanism suitable for FANET scenarios was constructed by simulating the foraging behavior of pangolins. Simulation results demonstrate that compared with traditional routing protocols, the CPO algorithm can still maintain a high packet delivery ratio and low end-to-end delay in scenarios with high-speed node mobility and high-density deployment, exhibiting good adaptability to highly dynamic FANET environments and being suitable for flying Ad Hoc networks in high-dynamic environments.关键词：FANET;metaheuristic algorithm;Chinese pangolin optimization algorithm;NS-36|0|0更新时间：2026-06-17
- 摘要：To address limitations in modeling user interest dynamics, cross-modal semantic alignment, and time-frequency feature extraction in multimodal sequential recommendation, wavelet-enhanced dynamic graph attention recommendation (Wave-DGARec) was proposed. This framework introduced innovations in three dimensions. A multi-scale wavelet transformation module was introduced to reconstruct behavioral sequences in the time-frequency domain, enabling the capture of non-stationary preference fluctuations. A user-item-image tripartite dynamic graph was constructed, wherein a graph attention mechanism is leveraged to enable efficient propagation of structured semantics across different modalities. A cross-modal contrastive learning strategy with a learnable temperature parameter was designed, enhancing both semantic alignment and sample discrimination. Experimental results on four Amazon domain datasets demonstrated the superiority of Wave-DGARec. Ablation studies further validated the effectiveness of both the wavelet module and the dynamic graph modeling. This work introduced a novel paradigm for multimodal recommendation systems by seamlessly integrating time-frequency analysis with structured representation learning.关键词：multimodal sequential recommendation;wavelet transform;dynamic graph attention;cross-modal contrastive learning5|0|0更新时间：2026-06-17
- 摘要：Format-preserving encryption (FPE) is characterized by its ability to maintain data format and length post-encryption while adhering to format constraints, rendering it highly applicable for data anonymization purposes. Current FPE schemes are faced with uncertainties in multiple calls to Cycle-Walking structures or dependencies on mod operations, leading to issues of asymmetry between the original message space and the symmetric ciphertext message space. The dynamic galois field-format-preserving encryption (DGF-FPE) algorithm based on the dynamic galois fields was proposed, which constructed a set of galois fields for multi-field formats that could cover the original message space. Data mapping was performed within the dynamic galois fields to obtain FPE ciphertext that conformed to the format. This approach was addressed to the issues associated with traditional methods that relied on Cycle-Walking or mod operations, and relevant experiments were conducted. Empirical results reveal that datasets encrypted via DGF-FPE exhibit information entropy levels approaching theoretical maxima, thereby substantiating the algorithm’s security profile.关键词：FPE;data anonymization;dynamic galois field;symmetric encryption algorithm;information entropy62|79|0更新时间：2026-06-17
Research and Development
- 摘要：Amid the ongoing advancement of information technology in the digital era, computing paradigms centered on computing power network and edge computing have been widely applied and recognized across numerous fields. By integrating heterogeneous and distributed computing resources, the computing network brain enables efficient resource scheduling and task allocation. The collaborative mechanisms between the computing network brain and edge computing were explored, both current and emerging technical approaches were examined, and an innovative algorithmic system was proposed. This system was characterized by five key features: multi-dimensional subjective and objective metrics, dynamic threshold adjustment and weighting, dual-scale decision-making, and intelligent hierarchical scheduling strategies. The implementation of this algorithm effectively reduces overall service latency, enhances resource utilization, and improves operational reliability.关键词：computing power network;computing network brain;edge computing4|0|0更新时间：2026-06-17
- 摘要：The edge AI agent based on the cloud gateway is an intelligent program or system which integrates edge computing and artificial intelligence technologies. This agent is deployed on edge devices and edge cloud gateways. It possesses the capability of identifying applications based on the cloud gateway, supports adaptive edge strategies, offers high scalability, and has broad application prospects. At the same time, it also can impose certain requirements on its architecture. The architectural solution of the edge AI agent based on the cloud gateway was primarily analyzed, which could be deployed and implemented on a carrier network to realize the adaptive and scalable advantages of the edge AI agent, serving as a reference for providing intelligent agent services in the deployment of next-generation metropolitan area networks.关键词：cloud gateway;edge AI agent;vDPI;converged edge4|0|0更新时间：2026-06-17
Engineering and Application
- 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：1118221 Visits Today：965
0