Critical nodes in complex networks are of fundamental importance, as they play essential roles in maintaining the network’s functionality, performance, stability, robustness, and dynamic behavior. Identifying critical nodes in complex networks is a significant research topic with broad applications across diverse domains, including biology, sociology, transportation, information science, computer networks, and communication systems. Most existing algorithms for identifying critical nodes rely on predefined mathematical models or heuristic rules to determine target strategies. However, these manually designed, rule-based strategies are constrained by fixed-dimensional representations, which significantly limit their ability to extract meaningful features. To address this issue, this paper proposes a framework named GCRA: Graph Contrastive Reinforcement learning based on Adaptive augmentation. The framework introduces an adaptive graph contrastive learning algorithm that effectively captures both the structural topology and node attributes, thereby representing the overall state of nodes and the network. Subsequently, reinforcement learning is integrated to identify critical nodes by gradually disintegrating the network. The model, trained on randomly generated BA networks, can be directly applied to real-world networks. In experiments, the algorithm is evaluated on synthetic networks of varying scales and real-world networks from diverse domains, and compared against eight benchmark methods. The proposed GCRA achieves performance improvements of 4.91% (unweighted networks) and 14.26% (weighted networks) in critical node identification, demonstrating its effectiveness for complex network analysis.

Network key nodes are very important in a network, after identifying them, network optimization can be carried out for these nodes. For example, increasing bandwidth or resource allocation, enhancing fault tolerance, protecting and strengthening key nodes. Or destroy key nodes to quickly break down the network, just like spreading rumors and preventing diseases. Different network optimization measures may vary in different network scenarios, but the core is to enhance or break key nodes, thereby affecting the overall network performance and effectiveness. However, most current network key node identification is based on fixed mathematical models or formulas, limiting the possibility of deep information mining in the network. To address this issue, this paper proposes a deep learning network architecture based on resource allocation to identify key nodes in the network. This allows for more in-depth feature extraction of network semantics and structures while combining the resource allocation situation of each node. In the process of gradually disintegrating the network through deep reinforcement learning, finding the strategy with the fastest decrease in cumulative connectivity rate is considered optimal. Finally, this paper compares the performance of six classical algorithms and the optimal algorithm on six real networks. The effectiveness improvement is 8.11% without sacrificing performance, validating the effectiveness of the method.

Existing multi-view 3D human pose estimation methods heavily rely on precise extrinsic calibration, which significantly restricts their practical deployment in uncontrolled environments. To address this limitation, we propose an Extrinsic Parameter-free Multi-view 3D Human Skeleton Estimation (EFMK) framework with three technical contributions. First, a Local-Global Pose Embedding scheme is proposed to simultaneously capture the fine-grained joint dependencies while establishing cross-view correspondences. Second, a Spatial-View Joint Transformer architecture is developed with three dedicated components: (1) Feature Transformation Modulation generates adaptive modulation vectors for distinct tokens to model heterogeneous relationship patterns; (2) Prior Knowledge Enhancement systematically integrates human kinematic constraints and multi-view geometric priors into attention computation through structural topology encoding; (3) Spatial-View Joint Attention implements decoupled spatial-view attention computation followed by joint distribution modeling to capture hierarchical spatial-view dependencies. Third, a Bone-wise Reprojection-based Multi-view Aggregation mechanism is introduced to consolidate multiple 3D outputs into a single, higher-quality 3D pose for practical applications. Extensive experiments on three benchmarks demonstrate that our method achieves state-of-the-art performance while maintaining a compact model size. Code and results are available at https://github.com/Z-Z-J/EFMK.

Identifying and safeguarding key nodes is crucial for maintaining the reliability and stability of business functions. Currently, most key node identification algorithms are based on manual or deductive models, which are complex and suboptimal. While deep reinforcement learning algorithms have demonstrated promising practical results, they often exhibit excessive randomness in network feature extraction. To address this challenge, this paper proposes a deep reinforcement learning framework based on multi attributes attention mechanism. This framework extends node attributes to encompass general business scenarios and employs graph convolutional networks combined with an attention mechanism to learn adaptive weights for different attributes of various nodes. Subsequently, reinforcement learning is utilized to determine the sequence of key nodes in the network. This algorithm is compared with six algorithms across six types of networks, consistently achieving optimal results, thereby validating the effectiveness of the proposed approach.

Unmanned aerial vehicles (UAVs) have emerged as a promising solution to support ground networks. In UAV-enabled wireless powered communication networks (WPCNs), the UAV provides data collection (DC) and energy transmission (ET) services for ground energy harvesting (EH) IoT devices. In view of the limited working efficiency of a single UAV, we propose a multi-UAV-enabled WPCN system. In order to further improve the efficiency of the system, we jointly optimize the data transmission performance of the network and the average harvested energy of IoT devices by designing the trajectories of the UAVs. The high mobility of UAVs causes the rapid change of channels, which leads to limitations of the centralized management scheme. Therefore, the problem is modeled as a multi-agent deep reinforcement learning (MADRL) problem. The scenario of multiple UAVs working collaboratively is formulated as a decentralized partially observable Markov decision process (Dec-POMDP). We develop a multi-agent deep deterministic policy gradient (MADDPG)-based approach to design UAVs' trajectories by adjusting the flying velocity and direction. Simulation results prove that the proposed algorithm far exceeds the benchmark schemes.

Abstract With the development of the internet of things, people pay more and more attention to wireless sensor networks. Designing the energy efficient routing is an essential objective for wireless sensor networks. Cluster routing is one of the most popular routing protocols to enhance the network lifetime. However, hotspot problem always exists in cluster-based routing protocol. The task of this study is designing a cluster routing protocol with mobile base station which aims at balancing the energy consumption and prolonging the network lifetime. In this article, we design a particle swarm optimization and artificial bee colony algorithm for clustering and mobile based software-defined wireless sensor networks. The software defined network architecture is used to reduce the energy overhead and computation overhead in sensor nodes. Particle swarm optimization-based cluster routing algorithm is used to calculate the cluster heads and the sojourn locations of base station. Artificial bee colony algorithm-based traversal path algorithm is used to design the move path of the base station. Comparing with relevant protocols, the proposed protocol reduces the energy consumption, enhances the network lifetime and reduces the control overhead.

Recently, wireless sensor networks play an important role in our life. Cluster routing has gained more attention in wireless sensor networks. However, hotpot problem always exists. One way to solve this problem is the unequal cluster routing. In most of the unequal routing protocols, nodes which closer to base station have the smaller cluster size than others. It will reduce the relay pressure of the node which near to the base station. In this paper, we propose the fuzzy-based multi-hop unequal cluster routing. In the cluster head election phase, relative inter cluster cost and relative intra cluster cost are proposed innovatively. Fuzzy system is used for unequal clustering which reduces the energy consumption for cluster members. In the cluster formation phase, a novel probability mechanism is proposed to let the cluster members decide which cluster to join. In the multi-hop routing phase, relative relay cluster cost is proposed innovatively for inter cluster routing. Two factors are considered for multi-hop routing which reduces the energy consumption for cluster heads. Self-adaptive rotation mechanism is proposed in the data transmission phase. It reduces the frequency for re-clustering self-adaptively which reduces the control overhead of the entire network. According to the simulation results, the proposed protocol balances and reduces the energy consumption as well as extends the network lifetime of the whole network.

The Content-Centric Network (CCN) is a very important structure in the future network, in which every node has caching ability. Caching strategy has a decisive influence on the performance of the CCN. In this paper, an Autonomous System Collaboration Caching Strategy (ASCCS) is proposed to achieve the explicit collaboration, reduce the cache redundancy, improve the cache utilization rate and increase the cache hit rate. Firstly, the network is divided into several Autonomous Systems (AS) before carrying out the centralized control. Then we select the control nodes according to the betweenness and the cache replacement rate of the nodes. Each AS is centralized controlled by the control node. At the same time, the selected control node computes the popularity statistics of different contents to determine the suitable cache policy for them. Under this strategy, nodes in the same AS cooperate with each other, which improves the transparency of the network cache, and thereby achieve the cache performance gain of the CCN. It makes the optimal performance of the cache become possible.

Content-centric Networking (CCN) is considered as a promising architecture to achieve reliable content distribution at large scale. One of the key research items of CCN is cache strategy, and most of the existing approaches consider little of the dynamicity of user interests. In this paper, we present a new cache policy, named as the betweenness and content popularity prediction (BEACON). Betweenness measures the importance of nodes in the whole network, and content popularity represents the user preference for service contents. By taking into account both network topology characteristics and flow distribution, the load of network and server is optimized. Moreover, we use the gray model to predict the content popularity, tracking the trend of user interest. The simulation results demonstrate that the BEACON scheme can effectively improve the cache hit rate, shorten access distance and reduce the delay of transmission.

With the rapid development of high speed railway (HSR), propagation characteristics of channels in HSR scenarios are therefore in urgent demand. We conducted numerous single input single output (SISO) measurements at 2.6 GHz with a bandwidth of 20 MHz along the Harbin-Dalian passenger dedicated railway line. Here, first analytical results in hilly terrains are provided. A double-slope path loss model fits measured data well and shadow fading is extracted to be log-normal distributed. Statistical results of small-scale fading are presented and compared in near regions and far regions relative to the transmitter, including the mean excess delay, root-mean-square (RMS) delay spread and the number of paths. Meanwhile, the delay Doppler spectrum is given out and verified. Finally, tapped-delay-line (TDL) channel model is established in detail based on the measured data. It is supposed that these results and models have a promotion for the further evaluation, simulation and design of the wireless communication system in HSR.

To obtain more accurate features of the time-varying wireless channel in the viaduct districts along Harbin-Dalian passenger dedicated railway line, a newly proposed 2 × 2 measurement, in which m-sequence optimum pairs are modulated and transmitted, is performed at 2.6 GHz with a bandwidth of 20 MHz. In this paper, the counted statistical correlation coefficients provide sufficient proof that the lower the relevance between subchannels is, the larger users the channel can accommodate. In addition, the derived path loss model is presented and the shadow fading is fitted to lognormal distribution. Small-scale parameters such as the delay spread, the number of the resolvable multipath and K-factor are also calculated. Fading properties are speculated under viaduct scenarios in high speed railway with the extensive analytical results. This paper promotes evaluation, simulation and design of the wireless communication system based on LTE-R applicable to viaduct scenes in high speed railway.

For multiuser multiple-input multiple-output (MU-MIMO) systems with limited feedback in the downlink scenario, an end-to-end robust Tomlinson-Harashima Precoding (THP) transceiver design that incorporates receiver combining and transmit precoder is proposed based on sum mean squared error (SMSE) minimization. The channel state information (CSI) available at the transmitter for precoding is the quantized channel direction information (CDI) relayed back from each receiver using feedback bits in such system. The proposed transceiver design method is robust to the channel uncertainties arising from the quantization error and the lack of channel magnitude information (CMI). As an aside, the authors analyse the SMSE performance of the proposed algorithm. Our simulation results show that the new THP scheme outperforms the conventional precoding in limited feedback systems with respect to Bit Error Ratio (BER).

The increasing interest that has been drawn to LTE system used in High-Speed Railway therefore requires measurement and analysis of broadband mobile channel for train-ground communications. By the approval of the Ministry of Railways of China, extensive and practical measurements are performed on Harbin- Dalian passenger dedicated line with the maximum running speed of 370 km/h. Based on the hierarchical two-hop network structure, this paper first presents comprehensive analyzed results of channel characteristics in viaduct scenario, including both the large-scale and small-scale fading parameters. The proposed path loss model benefits LTE system link budget and feasible transceiver-range determination. In particular, channel impulse response, power delay profile, the delay spread, the number of paths and delay-Doppler spectrum are extracted, evaluated and reported. All these informative results promote the evaluation and verification of broadband communication system on High-Speed Railway.

The increasing interest that has been drawn to the LTE system used in High-Speed Railway therefore requires measurements and analysis of wideband mobile channel for train-ground communications. Based on the hierarchical two-hop network structure, extensive and practical measurements are performed on Harbin-Dalian passenger dedicated line with a speed up to 370 km/h by the approval of the Ministry of Railways of China. This paper first presents comprehensive analyzed results of the statistical channel properties, including both the large-scale and small-scale fading parameters. The proposed path loss model is applicable to high-speed railway plain scenarios and benefits the link budget of the LTE system. In particular, channel impulse response, power delay profile, the delay spread, the number of paths and delay-Doppler spectrum are extracted, evaluated and reported. All these results promote the evaluation and verification of wireless wideband communications on the high-peed railway.