Please wait a minute...

Journal of Beijing University of Posts and Telecommunications

  • EI核心期刊

Current Issue

  • PAPERS

    • A Blockchain-Based D2D-Assisted Multi-Access Edge Computing Architecture for Resource Sharing
    • CHEN Wei, HUO Ru, WANG Shuo, HUANG Tao, LIU Yun-jie
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 1-9. DOI:10.13190/j.jbupt.2021-029
    • Abstract ( 1005 )     HTML( 552 )       
    • With the intelligent development of terminal devices and the growing applications of device-to-device(D2D)communication technology,terminal devices can share resource through a direct link. To improve the overall service capabilities of the network edge,a blockchain-based D2D-assisted multi-access edge computing architecture(BD-MEC)for resource sharing is proposed. In BD-MEC,considering the factors of the delay,energy consumption,and payment overhead,an offload decision-making based on game theory for multi-user scenario is proposed to meet the needs of different users. Moreover,in view of the malicious behaviors such as denial of service or payment,a smart contract-based resource sharing protocol and dispute handling method are proposed, which uses the blockchain to force the sharing parties to comply with the resource sharing protocol to achieve safe resource sharing. Simulations show that BD-MEC can effectively the reduce delay,energy consumption,and payment of computing tasks.
    • References | Supplementary Material | Related Articles
    • Network Traffic Prediction of Dropout Echo State Network
    • MU Xiao-hui, LI Li-xiang
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 10-13,20. DOI:10.13190/j.jbupt.2021-021
    • Abstract ( 775 )     HTML( 412 )       
    • An echo state network (ESN) based on Dropout method is proposed. The ESN based on Dropout method (Dropout ESN)is applied to the actual network traffic prediction task, in which the neurons in the reservoir are set to stop working with different probability. Dropout ESN is compared with the classical ESN to analyse the influence of the two networks on the prediction performance. In addition the normalized root mean square error of Dropout ESN and other models are compared and analyzed. Simulation results show that Dropout ESN has better prediction performance on network traffic than other ESN models.
    • References | Supplementary Material | Related Articles
    • A CPF-OFDM PDMA Downlink Transmission Scheme
    • ZHOU Li, MAO Zhen-dong, PENG Mu-gen, LIU Xi-qing
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 14-20. DOI:10.13190/j.jbupt.2020-273
    • Abstract ( 604 )     HTML( 350 )       
    • A typical pattern division multiple access (PDMA) utilizes the cyclic prefix based orthogonal frequency division multiplexing (CP-OFDM) as the multicarrier modulator,and hence its capacity is reduced owing to cyclic prefix insertion. To address the issue, cyclic prefix free OFDM(CPF-OFDM) wireless transmission technologies is studied. Then a PDMA scheme based on CPF-OFDM is proposed and the main structure for its transceiver is designed. In addition,to evaluate the system performance, the system capacity and bit error rate are evaluated with different key parameters, such as overloading rates, channel conditions and signal-to-noise ratios. Simulation results show that the proposed CPF-OFDM PDMA are harvest a significant gain in terms of capacity compared with a conventional PDMA based on CP-OFDM.
    • References | Supplementary Material | Related Articles
    • A Method for Targeted Sentiment Analysis
    • WANG Wen-zhu, XIAO Bo, CHEN Ke-hong
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 21-27. DOI:10.13190/j.jbupt.2020-276
    • Abstract ( 804 )     HTML( 717 )       
    • Targeted sentiment analysis(TSA)is a crucial task for fine-grained public opinion mining,which focuses on predicting the sentiment polarity towards a specific target in a given sentence. Most of existing works ignore the syntactic structure of the context sentence,and may pay attention to irrelevant context words when making sentiment judgments. To tackle the problem,a novel syntax aware model is proposed for TSA,which integrates the pre-trained bidirectional encoder representation from transformers models and a graph convolutional network over the dependency tree of the sentence to capture the context information and syntactic structure information of the sentence respectively. The proposed model uses the multi-head attention mechanism to aggregate the information to obtain the final target sentiment representation. The proposed model is also combined with the existing domain adaptive method to introduce domain knowledge and syntactic knowledge,which further improves the performance. The experimental results on several widely-used benchmark datasets demonstrate the effectiveness of the proposed model.
    • References | Supplementary Material | Related Articles
    • An Anomaly Detection Method Combining Mutual Information Estimation with Adversarial Autoencoder
    • HUO Wei-gang, WANG Xing, LIANG Rui
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 28-34. DOI:10.13190/j.jbupt.2021-009
    • Abstract ( 854 )     HTML( 655 )       
    • According to the information theory,the training objective of the unsupervised deep learning networks can be interpreted as maximizing the mutual information between the training samples and their representations. Adversarial autoencoder (AAE) learns the distribution of the training samples by the generative adversarial method. So the semi-supervised anomaly detection model based on the normal sample sets can be established using AAE. However,AAE cannot maximize the mutual information between the normal samples and their representations explicitly. A semi-supervised anomaly detection method based on mutual information estimation network and AAE (IAAE) is proposed. Firstly,the encoder and decoder of the AAE are trained to minimize the reconstruction error. Then,in the adversarial regularization stage,the aggregated posterior of the normal sample’s representations are matched to the arbitrary prior distribution,and the mutual information between normal samples and their representations is maximized. Finally,the mutual information between normal samples and their representations are estimated by fully connected neural network. The reconstruction error of the test sample and its mode divergence in the hidden space are used to calculate the abnormal score. The experimental results on public datasets show that the IAAE has better performance than the existing typical deep anomaly detection models in terms of F1 values.
    • References | Supplementary Material | Related Articles
    • Non-Data-Aided Estimation and Compensation of Amplitude, Frequency and Phase for Spectrum Combination
    • CHEN Jian-mei, QIU Hong-bing, ZHENG Lin, YANG Chao
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 35-40. DOI:10.13190/j.jbupt.2021-030
    • Abstract ( 532 )     HTML( 354 )       
    • In cognitive radios,broadband signals need to be applied with spectrum division and combination filter banks to efficiently utilize spectrum holes. As the amplitude,frequency and phase of different sub-spectra are interfered by the wireless channel,when the sub-spectra are combined,the amplitude,frequency and phase errors distort the recombination signal,leading to the degradation of detection perfor-mance. A non-data-aided estimation and compensation algorithm is proposed. By utilizing the amplitude-frequency-phase consistency between adjacent sub-spectra in the transition band and performing the alignment process in the frequency domain,the amplitude,frequency and phase are modified before sub-spectra combination to recover the original signal. Simulation results show that the proposed algorithm can effectively solve the amplitude-frequency-phase distortion caused by sub-spectra segmentation. The bit error rate performance of the proposed algorithm is close to that of the pilot estimation algorithm with the same system parameters. The proposed algorithm can not only obtain high-precision amplitude-frequency-phase estimation and compensation,but also utilize spectrum efficiently.
    • References | Supplementary Material | Related Articles
    • Network Slicing-Oriented Joint Allocation Algorithm of 3C Resources in MEC Systems
    • ZHENG Yuan-peng, ZHANG Tian-kui, ZHU Guang-yu, SHEN Hong
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 41-47. DOI:10.13190/j.jbupt.2021-005
    • Abstract ( 997 )     HTML( 574 )       
    • Mobile edge computing (MEC) has become one of the emerging research topics of future mobile network. In the fog radio access network based on network slicing,a network slicing-oriented communication,computation and caching (3C) joint resource allocation algorithm is proposed for MEC systems. A network slicing-oriented resource allocation model with multi-MEC cooperation is given. In this model,the limitations of wireless and backhaul bandwidth and the influence of MEC computation and caching resource allocation on service delay of network slicing are considered. The utility value of resources obtained by the user is defined by the service delay. A system utility maximization problem is proposed,which takes user association in different slices,computation and cache resources allocation as optimization variables. An iterative algorithm based on successive convex approximation is adopted to acquire approximate optimal solution. The performance of the proposed algorithm is verified by simulation. The results show that the proposed algorithm can optimize the total utility value of the system and improve the resource utilization efficiency of the network slicing-oriented MEC system.
    • References | Supplementary Material | Related Articles
    • A Split Sliding Window-Based Continuous Chinese Sign Language Recognition System
    • WANG Xin-yan, WANG Qing-shan, MA Xiao-di, LIU Peng, DAI Hai-peng
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 48-54. DOI:10.13190/j.jbupt.2021-001
    • Abstract ( 788 )     HTML( 597 )       
    • A large proportion of the world's disabled population is accounted for the individuals with hearing impairment which can communicate with people through the sign language. However, sign language is not mastered by the public, and there are still big obstacles between the individuals with hearing impairment and the normal people. A continuous Chinese sign language recognition system based on split sli-ding window (SSW) to realize automatic sign language recognition is proposed. The SSW system divides the sign language signal selected through the sliding window, and deletes one group of data to get new data in the original order, which is inputted to the sign language recognition neural network for training to obtain the gesture prediction value of a single sign language word. Finally, the majority voting strategy based on threshold is used to judge the identified prediction values. The SSW system is trained on 30 sign language sentences collected by 20 volunteers. The results show that the average accuracy of the SSW system reachs 83.9% on the test dataset, which is 16.7% higher than the long short-term memory model.
    • References | Supplementary Material | Related Articles
    • Robust Signal to Interference Plus Noise Ratio Balancing for Simultaneous Wireless Information and Power Transfer in Multicell Multicast Coordinated Networks
    • CHEN Dong-hua, ZHU Xuan
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 55-60. DOI:10.13190/j.jbupt.2021-018
    • Abstract ( 617 )     HTML( 304 )       
    • This paper proposes a robust simultaneous wireless information and power transfer optimization scheme for multicell multicast coordinated networks with imperfect channel state information. The proposed scheme aims to maximize the minimal signal to interference plus noise radio under the constraints of energy harvesting of each user and power consumption of each cell,by which the worst quality of service of users is thus optimized on condition of the energy harvesting requirement and the system power budget. To solve the problem,the original problem is reformulated to a quasi-convex problem by using semi-definite relaxation and S-Lemma, which, thereby, can be solved by bisection iterative algorithm. Simulations under different transmission parameters not only verify the correctness of the system designs, but also show good convergence properties of the iterative algorithm.
    • References | Supplementary Material | Related Articles
    • Modeling and Analysis of A2G 3D Wideband Nonstationary Channel
    • SUN Jing-jing, ZHANG Zhi-zhong, DENG Bing-guang
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 61-66. DOI:10.13190/j.jbupt.2021-013
    • Abstract ( 714 )     HTML( 961 )       
    • A stochastic model based on three-dimensional geometry for wideband non-stationary multi-input multi-output air to ground (A2G) channel is proposed. The model uses a single concentric cylindrical to simulate the distribution of the ground end scatterers, and further subdivides the non-line-of-sight component into the single-bounce scattering component and the ground reflection component. Time-varying angle and time-varying speed are introduced to simulate the non-stationarity of A2G channel. The space-time-frequency correlation function and space Doppler power spectral density function are derived. In addition,statistical characteristics such as envelope level crossing rate and average fading duration are also analyzed. The results show that the flight direction,altitude,elevation angle and other parameters of unmanned aerial vehicle can significantly affect the channel statistics and non-stationarity. The simulated values and the measured data match well, which indicates the correctness and applicability of the proposed channel model.
    • References | Supplementary Material | Related Articles
    • Frequency Centroid Zero Watermarking Algorithm Combine with Coding Interleaving
    • ZHANG Tian-qi, YE Shao-peng, BAI Hao-jun, LIU Jian-xing
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 67-73. DOI:10.13190/j.jbupt.2021-008
    • Abstract ( 608 )     HTML( 328 )       
    • The transparency of traditional digital watermarking algorithms decreases with the increase of embedding strength, and their robustness and stability are poor. To solve these problems a frequency centroid zero watermarking algorithm based on polarization code and two-dimensional interleaving algorithm is proposed. In the proposed algorithm,the watermark information is encoded by polarization code in turn,and the interleaving matrix is constructed by two-dimensional interleaving. Then, the low-frequency coefficients of the carrier image are obtained by non-subsampled contour transform. Next, the one-dimensional matrix is obtained by block non-negative matrix decomposition. According to the characteristics of frequency distribution,the local frequency centroid is calculated to construct the feature matrix. Finally, the exclusive or operation of the interlaced matrix and the feature matrix containing watermark information are carried out to generate zero watermark. Experimental results show that the zero watermarking algorithm can effectively resist conventional signal processing attacks,geometric attacks and combinatorial attacks,and the normalized coefficient of watermark extraction is more than 0.88,which is robust and stable for practical applications.
    • References | Supplementary Material | Related Articles
    • Power Control Algorithm under Nonlinear Battery Model in Communication Systems with Energy Harvesting
    • CHEN Hai-lin, LEI Wei-jia
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 74-80. DOI:10.13190/j.jbupt.2020-275
    • Abstract ( 478 )     HTML( 219 )       
    • Based on Lyapunov optimization framework,an online power control scheme is proposed to maximize the long-term average transmission rate for wireless communication systems with energy harvesting devices at the transmitter. The power control algorithm takes into account the energy loss in the charging and discharging processes of the rechargeable battery,and uses a nonlinear mathematical model to describe the charging and discharging efficiency. The constraint condition of battery power is transformed into the stable requirement for the energy virtual queue,and the negative value of the transmission rate that needs to be maximized is taken as the penalty term. Based on the current channel state and battery energy state,the average transmission rate is maximized by minimizing the drift-plus-penalty under the constraint of the harvested energy. Simulation results show that the performance of the proposed algorithm is slightly lower than that of the off-line water-filling algorithm,but is much better than those of the greedy algorithm and the half-power algorithm. In addition, the proposed algorithm also outperforms the existing algorithm that adopts Lyapunov method without considering the chargeing and discharging efficiency.
    • References | Supplementary Material | Related Articles
    • Multi-Modal Transportation Recommendation Based on Graph Embedding and CaGBDT
    • SUN Quan-ming, CHANG Lei, MA Cheng, QU Zhi-jian
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 81-87,106. DOI:10.13190/j.jbupt.2021-010
    • Abstract ( 777 )     HTML( 423 )       
    • To solve the problems in transportation recommendation service,such as single recommended methods and ignoring the user travel preferences,a cascade gradient boosting decision tree (CaGBDT) model is proposed based on the multi-grained cascade forest structure. CaGBDT uses the cascade structure to increase the depth, and then realizes the deep-level representation learning of features. Meanwhile to solve the imbalance of sample class,an index optimization layer based on Powell algorithm is proposed. By searching a threshold for each class,the weight of the prediction results of the model is modified to maximize the evaluation indexes. In addition,a user travel global relationship graph can be constructed by the CaGBDT model via referring the user's travel record, and the spatial context relationship of the user's travel is extracted automatically by the graph embedding method to improve the efficiency of feature extraction.
    • References | Supplementary Material | Related Articles

    REPORTS

    • A Lightweight Graph Convolutional Network Recommendation Model Incorporating Text Information
    • ZHANG Dong, CHEN Hong-long
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 88-93. DOI:10.13190/j.jbupt.2021-014
    • Abstract ( 669 )     HTML( 371 )       
    • In the recommendation model based on graph convolution network,the graph convolution only aggregates information from the input nodes with identifier information, which will decrease the recommendation precision and, thus, lead to a bottleneck problem. To solve this problem,a lightweight graph convolution network recommendation model based on text information fusion is proposed by considering enriching node features with auxiliary information. The model extracts text comment features from convolution neural network and adds them to the node embedding of graph. To simplify the structure of graph convolution network,the proposed lightweight graph convolution framework is used to transmit the fused feature information linearly on the user-movie item graph to learn the embedding of the user and movie item. The weighted sum of all sub-levels of the graph convolution is used as the final feature output for predicting the rating. Experimental results on three real datasets show that the proposed method can alleviate the bottleneck problem of information aggregation and improve the accuracy of recommendation. The model can also alleviate the cold start problem.
    • References | Supplementary Material | Related Articles
    • Internal-External Convolutional Networks for Network Intrusion Detection
    • WANG Yi-fei, MO Shuang, WU Wen-rui, FAN Shao-hua, XIAO Ding
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 94-100. DOI:10.13190/j.jbupt.2021-007
    • Abstract ( 799 )     HTML( 441 )       
    • Network intrusion detection is an important research topic in the field of network security which is used to distinguish normal and abnormal network behaviors by analyzing traffic characteristics to realize intrusion traffic detection. To solve the problems of the complex feature extraction process,and insufficient information extraction in existing intrusion detection models,an intrusion detection model based on internal and external convolutional networks is proposed. Firstly,an one-dimensional convolutional neural network is used to extract the internal features of the traffic data. Then, an undirected homogeneous graph is obtained by calculating the similarity of the internal features. In addition the communication behavior of the traffic on the external network side is modeled as a directed heterogeneous graph,and graph convolutional network is used to learn embedding containing multiple interactive behaviors of network traffic from two graghs. Finally, the learned flow embedding is input into the classifier for final classification. Experimental results show that compared with existing methods,the detection accuracy and false alarm rate of the proposed model are better than those of the compared models.
    • References | Supplementary Material | Related Articles
    • A Depth Estimation Method for Multi View and High Precision Images
    • LI Jian, CHEN Yu-hang
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 101-106. DOI:10.13190/j.jbupt.2020-255
    • Abstract ( 554 )     HTML( 457 )       
    • High-precision images are challenging to reconstruct effectively in dense multi view reconstruction. To solve the problem, a learning-based depth estimation method is proposed. In the method,the dilated convolution neural network is used to extract image features,and the long short-term memory network is applied to construct and optimize the cost volume. Besides,the supervised and unsupervised training methods are adopted. Experimental results on two real scene multi view image datasets show that the proposed method not only outperforms state-of-the-arts methods,but also is several times less in GPU memory application compared with traditional methods and other learning-based methods. Therefore, the proposed method can reconstuct high-precision images, while improving the accuracy and integrity of model depth prediction.
    • References | Supplementary Material | Related Articles
    • Structural Analysis of Hidden Danger Description Text Based on ERNIE-CRF-ESL
    • AI Xin-bo, GUO Yan-jun, XIE Yun-hao, CHEN Cheng
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 107-113. DOI:10.13190/j.jbupt.2021-003
    • Abstract ( 630 )     HTML( 414 )       
    • The safety hazard description text is recorded by natural language description,which has the problem of subjective arbitrariness. The existing sequence annotation-related models cannot extract key knowledge information from the safety hazard description. Based on the characteristics of the safety hazard description text,a sequence annotation method is designed for the safety hazard description text,and the enhanced representation from knowledge integration (ERNIE) model is proposed for word vector feature extraction. Based on the conditional random fields (CRF) module and the information extraction (ESL) module,a structured parsing method of safety hazard description text is constructed. An experiment is carried out on a description text of a hidden safety hazard in a mega-city. The experimental results show that the proposed model achieves a 65.1% precision rate in the text structured parsing task. The proposed algorithm can obtain more knowledge information from the unstructured data of urban safety hazards,and then standardize the safety hazards investigation and recording work.
    • References | Supplementary Material | Related Articles
    • Mobile Crowdsensing Scheme with Strong Privacy-Preserving
    • SHI Rui, FENG Hua-min, YANG Yang, YUAN Feng, LIU Biao
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 114-120. DOI:10.13190/j.jbupt.2021-004
    • Abstract ( 814 )     HTML( 460 )       
    • To realize the identity privacy, credential revocation, credit incentive features and mediate the contradiction between the identity tracking of malicious users and the privacy protection of honest users in a mobile crowdsensing system, a mobile crowdsensing scheme with strong privacy-preserving is proposed. Based on threshold cryptography, the new scheme distributes the identity tracking capability of anonymous users to multiple entities, which guarantees that multiple trackers can cooperatively reveal the real identity of users. Pointcheval-Sanders signature and Camenisch-Lysyanskaya accumulator based on Rivest-Shamir-Adleman assumption are combined to realize efficient and secure revocation of credentials. The privacy-preserving credit management mechanism is constructed by adopting the Pointcheval-Sanders signature. The security and experimental analysis of the scheme is carried out. The experimental results show that the scheme not only meets the security requirements, but also has feasibility in practical deployment.
    • References | Supplementary Material | Related Articles
    • Two-Stage Network Inpainting Algorithm Based on BDCN and U-net Edge Generation
    • LI Hai-yan, WANG Wei-hua, GUO Lei, LI Hai-jiang, LI Hong-song
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 121-126. DOI:10.13190/j.jbupt.2020-268
    • Abstract ( 888 )     HTML( 338 )       
    • To repair the large irregular missing areas of an image, and obtain reasonable structure and fine-detailed textures,a two-stage network image inpainting algorithm based on bi-directional cascade network (BDCN) for perceptual edge detection and U-net incomplete edge generation is proposed. Firstly,the algorithm edges are extracted using the BDCN network. In the first stage,down-sampling is used to extract the features of the missing image edges based on the edge generation network of the U-net network architecture. The information inputted by the up-sampling layer and the information of each down-sampling layer is then combined to restore the image edge texture details. In the second stage,the hole convolution is applied for down-sampling and up-sampling,which adopts the residual network to reconstruct the missing image with rich details. The proposed algorithm is compared with the existing classic algorithm on the public datasets. Experimental results demonstrate that the proposed algorithm can obtain reasonable results and fine texture details,and its performance is superior to those of the contrast algorithms.
    • References | Supplementary Material | Related Articles
    • Click-Through Rate Prediction of Multi-Head Self-Attention in Hyperbolic Space
    • HAN Yue-lin, WANG Xiao-yu
    • Journal of Beijing University of Posts and Telecommunications. 2021, 44(5): 127-132. DOI:10.13190/j.jbupt.2021-017
    • Abstract ( 601 )     HTML( 363 )       
    • In recommendation systems,understanding the complex functional interactions behind user behaviors is crucial to predict the clicking probability of users on advertisements or commodities. Efforts have been made to find low-dimensional representations and meaningful combinations of sparse and high-dimensional original features. Among them,the deep & cross network can explicitly cross features at each layer. However it treats all crossing features "equally" and does not consider the influence of different features on the results,which may eliminate some useful information. Therefore,a prediction model of click-through rate of multi-head self-attention neural network in hyperbolic space is proposed. In hyperbolic space,the model uses Lorentzian distance instead of inner product, to measure the similarity and correlation between features, which can avoid dimension disaster. Experimental results show that the model is superior to the deep & cross network on predicting click-through rate data sets in terms of accuracy.
    • References | Supplementary Material | Related Articles