The output of Microsoft Kinect is a multimodal signal, which provides RGB videos, depth sequences and skeleton information at the same time, opening up a new opportunity for the research of human action recognition. However, for different single modalities of the signals, how to exploit and fuse useful features of these various sources remains a very challenging problem. Most of the methods based on RGB-D action recognition simply fuse the multimodal features, ignoring the potential semantic relationship between different models. In this paper, we propose a multi-modal action recognition model based on Bilinear Pooling and Attention Network (BPAN), which could effectively fuse multi-modal for RGB-D action recognition. Firstly, we adopt the efficient data preprocessing methods for RGB and skeleton data. Then, we propose a multimodal fusion network combining RGB video and skeleton sequences. The proposed BPAN module could effectively compress the features of RGB and skeleton, and project them into latent subspace to get the fusion features. In the end, a fully connected three-layer perceptron is adopted to obtain the final classification decision. Experimental results on three public datasets demonstrate that our proposed method leads to a more favorable performance compared with the state-of-the-art methods.
Power communication network (PCN) is the cornerstone of the stable operation of smart grid. The evaluation of the essential vulnerability of the PCN based on the node importance is advantageous to the protection of critical nodes and the improvement of the reliability and controllability of the system. So, firstly, this paper establishes a multi-index evaluation system for the analysis of important nodes. Based on the site operation level, site load capacity, network topology, service flow, service quality requirements and security operation requirements, the network characteristics of nodes are divided into four layers to describe. Meanwhile, the characteristics of node topology are subdivided into three dimensions to analyze. Then, fully considering the coupling between power network and communication network, a multi-dimensional and multi-layer node importance (MMNI) model based on entropy weight method is proposed. Combined the long-term and short-term vulnerability loss, the node cross-layer vulnerability loss (CVL) function and network cross-layer loss entropy (CVE) model are proposed. Finally, based on the actual power communication network, under different attacks and fault strategies, the changes of maximum network connectivity and network cross layer loss entropy are compared. It shows the effectiveness and accuracy of the MMNI model for the analysis and evaluation of important nodes and it provides a theoretical basis for key weak point identification and the network optimization.
Software-defined networks (SDNs) are improving the controllability and flexibility of networks as an innovative network architecture paradigm. Segment routing (SR) exploits an end-to-end logical path and is composed of a sequence of segments as an effective routing strategy. Each segment is represented by a middle point. The combination of SR and SDN can meet the differentiated business needs of users and can quickly deploy applications. In this paper, we propose two routing algorithms based on SR in SDN. The algorithms aim to save the cost of the path, alleviate the congestion of networks, and formulate the selection strategy by comprehensively evaluating the value of paths. The simulation results show that compared with existing algorithms, the two proposed algorithms can effectively reduce the consumption of paths and better balance the load of the network. Furthermore, the proposed algorithms take into account the preferences of users, actualize differentiated business networks, and achieve a larger comprehensive evaluation value of the path compared with other algorithms.
The new power system is an energy interconnection network based on renewable energy generation. Information interconnection, data security, and reliability are the basis for the digital transformation of the power grid. Data sharing, lifecycle management, security, and user information privacy are issues that need to be addressed urgently. This paper analyzes the characteristics of the multi-combination of power grid data across services and introduces smart contract, cross-chain, and security encryption-related technologies. Based on the effective combination of smart contract and CP-ABE, data sharing schemes, including data sharing mechanism and data access control model are designed. Given this scheme, the blockchain system’s overall architecture is proposed, including the main chain, side chain, data sharing, and cross-chain information interaction. Finally, the underlying blockchain service platform is built using the Hyperledger open-source framework. We deploy the platform to verify the feasibility of the scheme according to the requirements of the data center, trust center, and blockchain-distributed nodes.
Software-defined networking (SDN) and network virtualization (NV) are key technologies for future networks, which allow telecommunication service providers (TSPs) to share network resources with users in a flexible manner. Since TSPs have limited virtualized network resources, it is critical to develop effective virtual network embedding (VNE) algorithms for an SDN network to improve resource utilization. However, most existing VNE algorithms ignore the security issues of SDN networks, which may be subject to malicious attacks due to their openness feature. Therefore, it is necessary to develop secure VNE (SVNE) for SDN networks. In this paper, we researched the relationship between resource value and node security-level, and we found that there are differences in the resource value of different nodes. Based on this analysis, we define the evaluation indicators considering differences in resource value for the SVNE problem. Then, we present a mixed-integer linear program (MILP) model to minimize the cost of SVNE. As the formulated optimization problem cannot be solved conveniently, we design two node-ranking approaches to rank physical and virtual nodes, respectively, and we propose two novel SVNE algorithms based on the node ranking approaches. Finally, simulation results reveal that our proposed algorithm is superior to other typical algorithms.