Current Issue

20 August 2021, Volume 37 Issue 8

Previous Issue    Next Issue

Comprehensive Review

AIoT: a taxonomy, review and future directions

Jiyi WU, Wenjuan LI, Jian CAO, Shiyou QIAN, Qifei ZHANG, Rajkumar BUYYA

2021, 37(8):  1-17.  doi:10.11959/j.issn.1000-0801.2021204

Asbtract ( 3826 )   HTML ( 727)   PDF (934KB) ( 2504 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

AIoT (artificial intelligence of things) is an integrated product of artificial intelligence (AI) and internet of things (IoT).Currently, it has been widely used in smart cities, smart homes, smart manufacturing, and driverless.However, the research of AIoT is still in its infancy, facing with many problems and challenges.In order to clarify the concept and provide possible solutions, a comprehensive survey was carried out on AIoT.Firstly, a clear definition of AIoT was provided, along with the brief introduction on its background and application scenarios.And then a novel cloud-edge-end hybrid AIoT architecture for intelligent information processing was constructed.Based on the research framework of AIoT, the research status and solutions were discussed, including AI integrated IoT data acquisition, complex event processing and coordination, cloud-edge-end integration, AI-enhanced IoT security and privacy, and AI-based applications, etc.Finally, it identified the open challenges and offers future research directions.

Research and Development

Neural network and Markov based combination prediction algorithm of video popularity

Xuesen MA, Shuyou CHEN, Xiangdong XU, Zhaokun CHU

2021, 37(8):  18-26.  doi:10.11959/j.issn.1000-0801.2021116

Asbtract ( 242 )   HTML ( 36)   PDF (775KB) ( 399 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Caching popular video into user-side in advance improves the user experience and reduces operator costs, which is a common practice in the industry.How to effectively predict the popularity of videos has become a hot issue in the industry.On account of the shortcomings of traditional prediction algorithms such as poor nonlinear mapping ability, low prediction accuracy and weak adaptability, a video popularity prediction algorithm based on a neural network and Markov combined model (Mar-BiLSTM) was proposed.Information dependencies were preserved by constructing bidirectional memory network model (bi-directional long short-term memory, BiLSTM), the prediction accuracy of the model was further improved by using Markov properties while avoiding the increase of the complexity of the model caused by the introduction of external variables.Experimental results show that compared with traditional time series and classic neural network algorithms, the proposed algorithm improves predicting accuracy, effectiveness and reduces the amount of calculation.

Beamforming algorithm for cognitive satellite and terrestrial network based on UAV relay

Yanli LI, Zhi LIN, Zining WANG, Ba XU, Ming CHENG, Jian OUYANG

2021, 37(8):  27-37.  doi:10.11959/j.issn.1000-0801.2021202

Asbtract ( 201 )   HTML ( 22)   PDF (986KB) ( 278 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Two beamforming (BF) schemes to achieve spectrum sharing by suppressing inter-system interferences in cognitive satellite and terrestrial networks were proposed, where the satellite and UAV cooperative network was used as the secondary network, while the terrestrial network was used as the primary network.Specifically, considering that only the statistical channel state information was available, a constrained optimization problem was formulated to maximize the signal-to- interference-plus-noise ratio of the secondary user under the constraints of the maximum transmit power at the UAV and the interference power of the primary user.Then, an iteration-based BF scheme was proposed to solve the constrained optimization problem.To reduce the computational complexity of the iterative algorithm, a zero-forcing based BF scheme was further proposed.Finally, computer simulations verify the correctness and effectiveness of the proposed BF schemes.

Improved CPD based DOA estimation of nested array

Sibei CHENG, Xiao LUO, Bochou JIANG, Yuting WANG, Huanyu WU

2021, 37(8):  38-45.  doi:10.11959/j.issn.1000-0801.2021119

Asbtract ( 162 )   HTML ( 8)   PDF (788KB) ( 249 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In order to avoid searching the peak value in space domain, when estimating nested array’s the direction of direction of arrival(DOA), the canonical polyadic decomposition(CPD) was applied into the nested array, namely using the one time singular value decomposition(SVD), bilinear mapping and tensor decomposition to obtain the steering vector matrix and arrival angle.However, the existing CPD algorithm only can be applied in noiseless environment, the algorithm was improved by utilizing SVD two times, and was made to be applied in both noiseless and noisy environments.The simulation results demonstrate that in the same signal to noise ratio(SNR) and snapshot, the DOA estimation algorithm of nested array based on the improved CPD has better performances and less running time than the MUSIC and space smoothing algorithms.

An improved YOLOv4 algorithm for pedestrian detection in complex visual scenes

Shuai KANG, Jianwu ZHANG, Zunjie ZHU, Guofeng TONG

2021, 37(8):  46-56.  doi:10.11959/j.issn.1000-0801.2021198

Asbtract ( 427 )   HTML ( 68)   PDF (2477KB) ( 443 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

At present, the difficulty of pedestrian detection has been dramatically increased because of some problems, such as the dark or exposed illumination, bad weather, serious occlusion, large difference size of pedestrians and blurred images in complex visual scenes.Therefore, an improved YOLOv4 algorithm was proposed, which improved the detection performance of pedestrian detection in complex visual scenes, aiming at the problems of low accuracy and highly missed detection rate.Firstly, the self-annotation data set pedetrian were constructed.Secondly, the hybrid dilated convolution (HDC) was added into the backbone network to improve the ability of pedestrian feature extraction.Finally, in order to obtain more detailed feature, the spatial jagged dilated convolution (SJDC) structure was proposed to replace the spatial pyramid pooling structure.The experimental results show that the average precision (AP) of the proposed algorithm can achieve 90.08%.The proposed algorithm can substantially improve AP by 7.2%, and the log-average miss rate (LAMR) reduce by 13.69% compared with the original YOLOv4 algorithm.

TAGAN: an academic paper adversarial recommendation algorithm incorporating fine-grained semantic features

Jinyang SUN, Baisong LIU, Hao REN, Jiangbo QIAN

2021, 37(8):  57-65.  doi:10.11959/j.issn.1000-0801.2021197

Asbtract ( 183 )   HTML ( 13)   PDF (796KB) ( 269 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Academic paper recommendation aims to provide users with personalized paper resources.Collaborative filtering methods face the problems of highly sparse data and lack of negative samples.Considering the above challenges, an academic paper recommendation algorithm TAGAN（title and abstract GAN）which incorporated fine-grained semantic features was presented.Firstly, based on titles and abstracts provide abundant semantic features, convolutional neural networks (CNN) was used to extract the global features of the titles, a two-layer long and short-term memory network (LSTM) was built to model abstract words separately.At the same time, the attention mechanism was proposed to associate the title and the abstract semantically.Then, the semantic features of the paper were integrated into the recommendation framework based on generative adversarial network (GAN).The generative model will fit the user’s interest preferences and can effectively replace the negative sampling process.Finally,through the experimental comparison on the public dataset, TAGAN is better than the baseline models in all indicators, which verifies the effectiveness of TAGAN.

Region inpainting algorithm of mouth-muffles for facial recognition

Yue LI, Yaguan QIAN, Xiaohui GUAN, Wei LI, Bin WANG, Zhaoquan GU

2021, 37(8):  66-76.  doi:10.11959/j.issn.1000-0801.2021193

Asbtract ( 679 )   HTML ( 75)   PDF (1296KB) ( 617 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Facial recognition under occlusion is a well-known difficult problem in real scenes.Especially after the outbreak of COVID-19, in airports, stations and other places that need to verify the identity of visitors, the mouth-muffle occlusion greatly reduces the facial features that can be important for identification, and the accuracy of face recognition algorithm decreases.Face de-occlusion was studied, and a novel framework was proposed to restore face, which used the edge generation network to generate edge maps.On this basis, the occluded region was restored by the region completion network while preserving identity information.In order to improve the performance, the spatial weighted adversarial loss and identity-preserving loss were introduced to train the above two networks.Then two face datasets with mouth-muffles were constructed by face landmarks.The experimental results show that the accuracy of facial recognition algorithm ArcFace on the face dataset restored by the proposed model was 98.39%, which was 4.13% higher than that of directly using ArcFace.

Hybrid precoding method for millimeter-wave massive MIMO systems based on IAFS algorithm

Haoyi CHEN, Guangqiu LI, Hui LI

2021, 37(8):  77-84.  doi:10.11959/j.issn.1000-0801.2021192

Asbtract ( 121 )   HTML ( 10)   PDF (760KB) ( 128 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The millimeter-wave massive multiple-input multiple-output (MIMO) systems can overcome the adverse effects of the free-space signal path loss through the partial connection hybrid precoding method, which has the advantages of low hardware complexity and high energy efficiency.When the number of input data streams is equal to the number of radio frequency (RF) links, the hybrid precoding method based on partially connected structure and serial interference cancellation can be used.When the number of input data streams is not equal to the number of RF links, a hybrid precoding method based on improved artificial fish swarm (IAFS) algorithm was proposed.The core idea is that based on the spectral efficiency optimization criteria and the characteristics of partial connected structure, the spectral efficiency optimization problem of analog recoding matrix variables was transformed into the spectral efficiency optimization problem based on vector variables, and then the IAFS algorithm was used to solve the spectrum efficiency optimization problem.The simulation results show that the proposed method has good spectral efficiency and energy efficiency under the condition of low signal-to-noise ratio, and is expected to be applied in the real scene.

Outdoor location scheme with fingerprinting based on machine learning of mobile cellular network

Zhichao ZHOU, Yi FENG, Xiaohan XIA, Yuyao FENG, Chao CAI, Jiahui QIU, Lihui YANG, Yunxiao WU

2021, 37(8):  85-95.  doi:10.11959/j.issn.1000-0801.2021201

Asbtract ( 235 )   HTML ( 27)   PDF (1306KB) ( 315 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The positioning scheme based on mobile cellular network technology is one of the important technical approaches to provide network optimization, emergency rescue, police patrol and location services.The traditional positioning scheme based on cell base station location information has low positioning accuracy and large positioning error, so it cannot meet the requirements of some positioning applications.The scheme based on fingerprint location can greatly improve the location accuracy, save computational cost and enhance the usability based on the coarse location scheme of the cell and become the hotspot of the research.Rasterization and non-rasterization of outdoor fingerprint location scheme based on machine learning were studied and analyzed to meet the business requirements of outdoor fingerprint location.By means of parameter weighting, data fitting and other methods, large-scale fingerprint data were cleaned to improve the effectiveness of data sources.Through the realization of sub-modules such as demarcating research area, rasterizing, constructing fingerprint database, training model, correcting model, non-rasterizing, rough positioning coupling, matching parameter and training parameter, the operation efficiency and positioning accuracy of the algorithm were analyzed and optimized, and the key indexes affecting the algorithm performance were determined.Then, the performance of two fingerprint-based localization schemewas analyzed based on the simulation results.Finally, the typical scenarios of the fingerprint location scheme based on machine learning in practical application were presented.

5G industry wireless private network technology and business requirement model

Xuezhi ZHANG, Jingbo ZHAO, Huijie ZHAO

2021, 37(8):  96-104.  doi:10.11959/j.issn.1000-0801.2021194

Asbtract ( 280 )   HTML ( 35)   PDF (702KB) ( 642 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

5G industry wireless private network has become one of the most important services for operators enabling the vertical industry.The gradual maturity of virtualization, multi-access edge computing, network slicing and other technologies have provided a key technical guarantee for the development of 5G industry wireless private network.The development status, key technology, requirement model, challenges and development recommendations of 5G industry wireless private network were introduced.The concept and classification of 5G industry wireless private network were introduced, the key technologies were generalized, and the design ideas and the main reference index were put forward aiming at business requirement.Finally, the challenges faced by operators when developing 5G wireless private network services were summarized and development suggestions were given.

Topic: Cloud-Network Convergence

Research of cloud-network operating system based on ONAP open-source architecture

Dong WANG, Qiong SUN, Honglei XU

2021, 37(8):  105-110.  doi:10.11959/j.issn.1000-0801.2021208

Asbtract ( 305 )   HTML ( 31)   PDF (1672KB) ( 490 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Cloud-network convergence is an output of the deep convergence of communication technology and information technology.Open network automation platform (ONAP) is an open-source automatic network orchestrator managed by Linux Foundation Networking.It is able to support the operation and orchestration of multiple network applications based on unified open architecture.A systemic solution was proposed to deeply research design-time run-time, closed-loop automation and E2E slicing use case based on ONAP open-source architecture to support the operation and orchestration requirements of cloud-network operating system in the future.This solution focuses on the agile and iterated development of efficient and applicative cloud-network operating system for the requirements of cloud-network convergence operation based on open-source unified architecture.

Cloud-network convergence solution based on SRv6

Wei WANG, Peng WANG, Xiaoyu ZHAO, Honglei XU

2021, 37(8):  111-121.  doi:10.11959/j.issn.1000-0801.2021205

Asbtract ( 413 )   HTML ( 69)   PDF (1238KB) ( 910 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The development background and requirements of cloud-network convergence were introduced, and the principle and advantages of SRv6 technology were explained.At the same time, cloud-network convergence solutions based on SRv6 was proposed, and the technical architecture was studied, including SD-WAN, cross domain networking, cloud service and cloud interconnection.By comparing the traditional solution with the solution based on SRv6, the advantages of the latter in solving the existing cloud network business pain points were explained, and the significance of cloud-network convergence was described.

PCEP application in cloud-network convergence and end to end traffic assurance scheme

Yue WANG, Aijun WANG, Honglei XU

2021, 37(8):  122-127.  doi:10.11959/j.issn.1000-0801.2021203

Asbtract ( 282 )   HTML ( 31)   PDF (688KB) ( 323 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

With the booming development of new technologies such as cloud computing, big data and the continuous upgrading of communication infrastructure, the major domestic ISPs have proposed the development strategy of building a centralized control and flexible cloud network infrastructure under the trend of intelligent-traction network transformation.As a path calculation protocol widely used in the existing SDN architecture, PCEP protocol was often used to solve the problem of cross-layer and cross-domain path computing under complex network environment.The application of PCEP protocol in cloud-network convergence scenario was discussed and an end-to-end traffic guarantee scheme was introduced based on PCEP protocol which could ensure the connection-oriented network communication and make end-to-end service guarantee in native IP environment.It helps to improve the operators' ability of real-time perception, agile response and intelligent analysis of the network and meets the operation and maintenance management requirements such as global scheduling of communication resources, comprehensive opening of capabilities and flexible architecture.

5G core network cloud-network integration operation ＆ maintenance

Xue ZHANG

2021, 37(8):  128-135.  doi:10.11959/j.issn.1000-0801.2021200

Asbtract ( 585 )   HTML ( 49)   PDF (663KB) ( 843 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Most operators in the world have turned to 5G independent networking architecture and have begun commercial deployment of 5G independent networking architecture.The 5G core network adopts NFV technology, which forms a variety of virtual resources for general computing, storage, and network hardware to achieve dynamic and flexible deployment of network functions and resources based on needs.With large-scale commercial deployment of 5G core network with network function virtualization technology, the operation ＆amp; maintenance of 5G core network is also facing new challenges.The difficulties of operation ＆amp; maintenance brought was analyzed by the virtualization deployment scheme of 5G core network, the architecture and scheme of 5G core network operation ＆amp; maintenance system construction were proposed, and suggestions on the organizational structure of 5G core network cloud-network integration operation ＆amp; maintenance were put forward, so as to provide reference for 5G core network cloud-network integration operation ＆amp; maintenance.

Research and implementation of the IBN for cloud-network convergence service of operators

Xin ZHANG

2021, 37(8):  136-141.  doi:10.11959/j.issn.1000-0801.2021207

Asbtract ( 269 )   HTML ( 30)   PDF (735KB) ( 355 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Faced with the development of cloud computing technology and the increasing business needs, cloud-network convergence has become the inevitable choice of network architecture reform.The era of cloud-network convergence brought great challenges to the traditional network operation and maintenance, and the network needs to evolve to a more intelligent and convenient mode.The intent-based networking(IBN), which is widely studied by the scholars, has become the best choice to solve this problem.The development process, architecture and key steps of IBN were introduced.Combined with the needs of operator’s services, the business scenario of cloud-network convergence was taken as an example to analyze the implementation process of IBN for cloud-network convergence.It is expected to provide a reference for the future practice of IBN and cloud-network convergence services.

Engineering and Application

Application of computer vision in intelligent security

Zhihong CHEN, Mingxiao WANG

2021, 37(8):  142-147.  doi:10.11959/j.issn.1000-0801.2021195

Asbtract ( 521 )   HTML ( 41)   PDF (754KB) ( 449 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The development of computer vision and deep learning technology in the field of artificial intelligence were summarized, the relationship between them was expounded, and its future development trend was proposed combining with 5G technology.The actual case of face capture in a community was analyzed, and the application of computer vision technology in the intelligent security industry of urban governance was focused on, including process design, business logic and algorithm principle, hoping to provide reasonable suggestions and new ideas for experts and scholars in related fields.

5G inter-band cooperation technology

Lifang CAO, Tianming JIANG, Wei DENG, Zhuo CHEN

2021, 37(8):  148-154.  doi:10.11959/j.issn.1000-0801.2021139

Asbtract ( 262 )   HTML ( 34)   PDF (3547KB) ( 978 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

With the rapid growth of large traffic demand services such as AR/VR, it is imperative to increase the peak rate of single user.The basic principles of multi-band joint transmission technology in different protocol versions were mainly studied, such as carrier aggregation, supplementary uplink and dual connectivity, which could improve the performance of single user peak rate.Then the deployment requirements of different technologies to the network were analyzed, especially for the non-co-BBU or non-co-RRU scenarios to do qualitative performance analysis, subsequent quantitative analysis or test verification will further verify the accuracy of qualitative analysis.After that, the upstream and downstream peak rate and edge rate corresponding to different technologies were calculated, including performance comparison of different protocol versions.At last, the industry status quo of different technologies was introduced.