Current Issue

20 September 2024, Volume 40 Issue 9

Previous Issue    Next Issue

Research and Development

Architecture modeling of satellite Internet simulation platform based on MBSE

Hongguang LI, Yaoqi LIU, Yiqing ZHOU, Jinglin SHI

2024, 40(9):  1-12.  doi:10.11959/j.issn.1000-0801.2024201

Asbtract ( 171 )   HTML ( 29)   PDF (3608KB) ( 179 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The traditional text based system engineering method has some problems in the development of high complexity satellite Internet simulation platform, such as poor system design coordination and insufficient early simulation verification. The model-based systems engineering (MBSE) method was proposed to model the architecture of satellite Internet simulation platform. Firstly, a DV-MBSE and an integration verification architecture with external software were proposed. Secondly, based on MBSE methodology, the requirements analysis, functional decomposition and interaction structure modeling of the top architecture of the satellite Internet simulation platform were carried out. Finally, the validity of the model was verified by running logic verification and external model integration verification, thus supporting the design of satellite Internet.

MM-LoRa-Mod: A non-coherent LoRa modulation scheme for underwater acoustic communications

Xiao YU, Weikai XU, Haixin SUN, Qi LIU

2024, 40(9):  13-27.  doi:10.11959/j.issn.1000-0801.2024194

Asbtract ( 76 )   HTML ( 10)   PDF (10402KB) ( 29 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

LoRa (long-range) is an emerging low-power wide-area network technology. Facing challenges such as frequency selective fading multipath channels, existing LoRa schemes are subjected to issues like low data rates and high system complexity. To address these issues, a multi-mode LoRa (MM-LoRa-Mod) based modulation scheme was proposed. The frequency shifts of the multiple multiplexed chirp signals for information transmission were utilized in this scheme, thereby significantly enhancing data rates. Additionally, for underwater acoustic communication scenarios, a low-complexity non-coherent detection algorithm for MM-LoRa-Mod was introduced. Through the evaluation of MM-LoRa-Mod’s bit error rate performance in multipath channels, Watermark underwater acoustic channels, and shallow water acoustic channels, excellent anti-multipath fading and anti-Doppler frequency shift capabilities were demonstrated, indicating the high reliability of MM-LoRa-Mod in underwater acoustic communication scenarios.

Design and verification of the onboard 6G core network architecture and network functions

Shangguang WANG, Chao WANG, Xiao MA, Ruolin XING, Ao ZHOU

2024, 40(9):  28-43.  doi:10.11959/j.issn.1000-0801.2024202

Asbtract ( 156 )   HTML ( 28)   PDF (3763KB) ( 242 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

In order to comprehensively improve the in-orbit service capability of satellite networks, it is necessary to integrate 6G core networks and satellite networks. In view of the urgent need of 6G core networks deployment， the system architecture of the onboard 6G core network was designed, including distributed architecture, offline autonomy, intelligent network elements and so on. The network elements of the onboard 6G core network were optimized, including access and mobility management, session management, distributed service registration and discovery, etc. The effectiveness of the proposed 6G core network architecture was verified by the on-orbit deployment and simulation tests. The simulation results show that the average communication delay between network functions of the centralized onboard 5G core network is 109.3 ms, while the value of the distributed onboard 6G core network is 60.3 ms, which is a 44.8% reduction in comparison. In addition, the average delays of 5G centralized service registration and service discovery are 40.53 ms and 40.04 ms respectively, while the average delays of 6G distributed service registration and service discovery are 35.18 ms and 34.91 ms respectively, with average reductions of 13.2% and 12.8% respectively.

Joint channel and impulsive noise estimation method based on approximate message passing for NOMA systems

Youming LI, Chongya MA, Yonghong WU, Qiang GUO

2024, 40(9):  44-53.  doi:10.11959/j.issn.1000-0801.2024205

Asbtract ( 51 )   HTML ( 4)   PDF (4916KB) ( 73 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

To address the channel estimation problem for non-orthogonal multiple access (NOMA) systems under non-Gaussian impulsive noise, a joint channel and impulsive noise estimation method based on approximate message passing was proposed, by exploiting the joint sparsity of the channel and impulsive noise. Firstly, based on sparse Bayesian learning theory, a compressed sensing equation was constructed by using all subcarriers, and then a joint estimation optimization problem for the channel, impulsive noise, and data symbols was proposed. To address this hyperparameter nonlinear non-convex problem, an expectation maximization (EM) implementation algorithm based on Gaussian generalized approximation message passing and sparse Bayesian learning (SBL) theory was designed. Simulation results show that compared to the SBL method based on EM, the proposed algorithm exhibited a slight degradation in terms of mean square error (MSE) for channel and impulsive noise estimation, bit error rate (BER). However, the complexity of the proposed algorithm was reduced by one order of magnitude.

DA-ResNeXt50 method for radio frequency fingerprint identification based on time-frequency and bispectral feature fusion

Mengdi CHEN, Wei ZHANG, Lei SHEN, Fuqiang LEI, Jiafei ZHANG

2024, 40(9):  54-65.  doi:10.11959/j.issn.1000-0801.2024208

Asbtract ( 108 )   HTML ( 13)   PDF (4233KB) ( 190 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

To address the problems that a single feature in radio frequency fingerprint recognition could not fully represent the integrity of the signal and that the differences between features of different classes were small, which limited the recognition accuracy, a DA-ResNeXt50 (ResNeXt50 with dense connection and ACBlock) method for radio frequency fingerprint identification based on time-frequency and bi-spectral feature fusion was proposed. Firstly, short-time Fourier transform (STFT) and bi-spectrum transform were performed separately on the signals collected from different devices, the resulting images were bi-narized and then concatenated. By taking advantage of the advantages of both transformations in the time-frequency domain and high-order statistical characteristics respectively, the radio frequency fingerprint features of different devices can be extracted and characterized more comprehensively. Then, the DA-ResNeXt50 network model was proposed. Borrowing from the idea of dense connection, each layer of the four-layer residual unit was directly connected to all previous layers, promoting feature reuse and transmission, which enabled it to better capture subtle differences between classes. Finally, the asymmetric convolution block (ACBlock) was used to replace the 3×3 convolution in the last residual unit of the model. This effectively increased the receptive field of the network and enhanced the skeleton part of the convolutional kernel, thereby improving the performance of radio frequency fingerprint recognition. The experimental results show that compared with that of using a single feature extraction method, the proposed feature fusion approach significantly improves performance. Compared with various classical models, the improved model has higher recognition accuracy.

Swin Transformer lightweight: an efficient strategy that combines weight sharing, distillation and pruning

Bo HAN, Shun ZHOU, Jianhua FAN, Xianglin WEI, Yongyang HU, Yanping ZHU

2024, 40(9):  66-74.  doi:10.11959/j.issn.1000-0801.2024209

Asbtract ( 222 )   HTML ( 17)   PDF (2702KB) ( 118 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Swin Transformer, as a layered visual transformer with shifted windows, has attracted extensive attention in the field of computer vision due to its exceptional modeling capabilities. However, its high computational complexity limits its applicability on devices with constrained computational resources. To address this issue, a pruning compression method was proposed, integrating weight sharing and distillation. Initially, weight sharing was implemented across layers, and transformation layers were added to introduce weight transformation, thereby enhancing diversity. Subsequently, a parameter dependency mapping graph for the transformation blocks was constructed and analyzed, and a grouping matrix F was built to record the dependency relationships among all parameters and identify parameters for simultaneous pruning. Finally, distillation was then employed to restore the model’s performance. Experiments conducted on the ImageNet-Tiny-200 public dataset demonstrate that, with a reduction of 32% in model computational complexity, the proposed method only results in approximately a 3% performance degradation at minimum. It provides a solution for deploying high-performance artificial intelligence models in environments with limited computational resources.

A multi-feature fusion exercise recommendation model based on knowledge tracing machines

Bin ZHUGE, Ying WANG, Mengfan XIAO, Lei YAN, Bingyan WANG, Ligang DONG, Xian JIANG

2024, 40(9):  75-87.  doi:10.11959/j.issn.1000-0801.2024210

Asbtract ( 119 )   HTML ( 16)   PDF (2756KB) ( 79 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The subject of personalized exercise recommendation holds significant relevance within the domain of personalized services in smart education. Nevertheless, traditional algorithms have often lacked a deep understanding of student characteristics and failed to adequately explore the relationship between knowledge mastery and question-answering behaviors, leading to low recommendation accuracy. To address these issues, combining the knowledge tracing machine and the user-based collaborative filtering algorithm, as a KTM-based multi-feature fusion exercise recommendation model, SKT-MFER was proposed. Firstly, as a knowledge tracking model, KTM-LC, incorporating student learning behaviors and learning abilities, was constructed to accurately assess the student’s knowledge mastery level. Subsequently, two filters were implemented to ensure the exercise recommendation’s accuracy: the first was an initial screening utilizing the knowledge point mastery matrix to eliminate students who were similar to the target student, and the second was a filtering process considering the combined similarity of cognitive state similarity and exercise difficulty similarity. Through extensive experiments, it proves that the proposed method yields better results than some existing baseline models.

Efficient certificateless aggregate signcryption scheme with conditional privacy protection for V2G networks

Jie LIU, Xinyue FAN, Jiahui HE

2024, 40(9):  88-98.  doi:10.11959/j.issn.1000-0801.2024211

Asbtract ( 68 )   HTML ( 7)   PDF (1489KB) ( 69 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

To address the issue of message confidentiality in certificateless signature schemes and the shortcomings of most signcryption schemes, such as incomplete security functions and low efficiency, a certificateless aggregate signcryption scheme based on elliptic curve encryption technology was proposed for vehicle-to-grid (V2G) networks. Part of the vehicle’s public and private keys were generated by the vehicle itself, avoiding the problem of key escrow. Verification efficiency was improved by the local aggregator through aggregate unsigncryption. Conditional privacy protection was provided by a pseudonym mechanism, ensuring the anonymity of legitimate vehicles and the traceability and revocation of malicious vehicles. Binary polynomials were used to achieve autonomous updating of vehicle pseudonyms. In the random oracle model, the scheme was proven to meet indistinguishability under adaptive chosen ciphertext attacks (IND-CCA2) and existential unforgeability under adaptive chosen message attacks (EUF-CMA). The security of the scheme was verified by the Scyther formal analysis tool. Performance analysis shows that, compared to recent signcryption schemes, the proposed scheme achieves an average reduction of approximately 12.9% in communication overhead and 84.4% in aggregate unsigncryption computation costs, while also meeting higher security requirements.

A microscopic image stitching algorithm with improved energy function

Jiang’an WANG, Wenqian LIANG, Le HUANG, Linzhen QIN

2024, 40(9):  99-108.  doi:10.11959/j.issn.1000-0801.2024212

Asbtract ( 72 )   HTML ( 10)   PDF (4426KB) ( 63 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Addressing the challenge of traditional optimal suture algorithms in effectively eradicating splicing seams after fusing microscopic images, an image fusion optimization algorithm based on the optimal suture was proposed. Firstly, ORB was used to extract features from the to-be-spliced images and locate the overlapping region. Secondly, an improved energy function was designed and combined with the dynamic planning technique to effectively search for the optimal suture line in the overlapping region. Based on this line, the improved fusion method was further used to fuse the two images. Finally, the Poisson fusion method was utilized to eliminate the splicing seams existing between the overlapped and non-overlapped regions to obtain the final image. The experimental results show that this method improves the similarity of the pixels on both sides of the optimal suture line by about 5%, reduces the energy loss, eliminates the splicing seam, and improves the image quality.

A fast VVC intra-coding algorithm based on graph neural network and statistical analysis

Tiansong LI, Haokun LIU, Shaoguo CUI, Shucen LIU, Yan CHEN, Hongkui WANG

2024, 40(9):  109-122.  doi:10.11959/j.issn.1000-0801.2024213

Asbtract ( 74 )   HTML ( 8)   PDF (3877KB) ( 125 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

VVC as the latest generation of video coding standards, further improves video compression quality by introducing a variety of efficient coding tools. However, the VVC standard introduces the QTMT division structure and expands the intra prediction modes from 35 to 67, resulting in a sharp increase in coding complexity. Firstly, a fast algorithm for intra-frame coding unit (CU) division based on graph neural network was proposed, in order to reduce the complexity of intra-frame coding of VVC. An efficient graph neural network model was used to directly predict the optimal partition mode of CU, thus skipping redundant CU partition traversal. Secondly, a fast algorithm for intra-frame mode selection based on spatial correlation and texture features was proposed. The average direction variance and Sobel gradient operator were used to determine the texture direction, some angle prediction modes were skipped, and the correlation between prediction modes to streamline the rate-distortion mode list were combined. Experimental results show that this algorithm can save 64.04% of encoding time at the cost of increasing BDBR by 2.29%.

Edge-cloud collaboration empowered immersive Web3D

Yanan XU, Guidan ZHAO, Daquan FENG, Wenzhe SHI, Ping LU

2024, 40(9):  123-135.  doi:10.11959/j.issn.1000-0801.2024193

Asbtract ( 103 )   HTML ( 7)   PDF (1678KB) ( 255 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Leveraging the natural cross-platform accessibility, Web3D has shown wide application prospects in digital content fields with a browser as a carrier, including virtual reality, 3D modeling and visualization, to name a few. Nonetheless, constructing a diversified and shared immersive Web3D service platform is still in its infancy. How to ensure seamless communication, computing, and storage services has become a core challenge in dealing with the dynamic changes in network environments, limitations in browser-based rendering capabilities, and the impending surge in interactive demands. The progression of cutting-edge technologies such as mobile Internet and 5G networks, coupled with the emergence of edge computing, has propelled Web3D service modes from solitary terminal or cloud-based rendering toward a collaborative edge-cloud paradigm. Such a paradigm effectively integrates the resources from the remote cloud, edge nodes, and end users, thereby fulfilling the requirements of rapid response, high-speed data transmission, and real-time interaction. Focusing on the application of edge-cloud collaboration frameworks in immersive Web3D services, how this computing mode tackle the intricate issues encountered in establishing Web3D platforms was discussed. Firstly, a Web3D scenario-oriented edge-cloud collaboration framework and the key challenges faced in constructing immersive Web3D service platforms were discussed. Secondly, the optimization strategies for edge-cloud collaboration-driven Web3D were discussed from the perspectives of resource scheduling, transmission and loading, pre-caching schemes, and model lightweighting, respectively. Finally, open research questions were explored.

Model split-based data privacy protection method for federated learning

Ka CHEN

2024, 40(9):  136-145.  doi:10.11959/j.issn.1000-0801.2024206

Asbtract ( 110 )   HTML ( 12)   PDF (3902KB) ( 72 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Split learning (SL) enables data privacy preservation by allowing clients to collaboratively train a deep learning model with the server without sharing raw data. However, the SL still has limitations such as potential data privacy leakage. Therefore, binarized split learning-based data privacy protection (BLDP) algorithm was proposed. In BLDP, the local layers of client were binarized to reduce privacy leakage from SL smashed data. In addition, the leakage-restriction training strategy was proposed to further reduce data leaks. The strategy combines leak loss of local private data and model accuracy loss that enhances privacy while maintaining model accuracy. To evaluate the proposed BLDP algorithm, experiments were conducted on four commonly benchmarked datasets and the leakage loss and model accuracy were analyzed. The results show that the proposed BLDP algorithm can achieve a balance between classification accuracy and data privacy loss.

Engineering and Application

Cloud federation three-layer architecture: game-theoretic QoS modeling

Kun MA, Chuangyue HU, Yuzhi ZHANG, Fangchao MA, Xiaodong WANG, Zhonghua DANG, Wenhong XIAO, Shuangxi CHEN

2024, 40(9):  146-161.  doi:10.11959/j.issn.1000-0801.2024200

Asbtract ( 66 )   HTML ( 4)   PDF (4457KB) ( 96 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Facing intense competition from large public cloud providers both domestically and internationally, the survival difficulties of small and medium cloud providers have increased. To address this, establishing a cloud federation based on cooperation has become a viable strategy for these providers. However, there is a complex interplay between pursuing individual maximum benefits and ensuring the overall quality of service (QoS) of the federation. A QoS based cloud federation model was proposed to address the above issues, covering a three-layer architecture of cloud computing. At the application to virtual layer, an innovative task allocation strategy based on the differential evolution (DE) algorithm was proposed, specifically designed to handle multi-QoS task distribution challenges. At the virtual to physical layer, a model for virtual machine migration that coexists with cooperation and competition was proposed, suitable for balancing energy consumption and QoS in a cloud federation game-theoretic computing environment. The experimental results indicate that the proposed solution improved the service quality of cloud computing environments and revealed the relative advantages of running cooperation and competition modes in cloud federation.

Architecture design, key technologies, and application research of GPU heterogeneous resource pool platform based on virtualization

Wancai ZHANG, Nan ZHANG, Wenqing YANG, Tao WANG, Wenqiang ZHANG

2024, 40(9):  162-175.  doi:10.11959/j.issn.1000-0801.2024216

Asbtract ( 193 )   HTML ( 44)   PDF (4230KB) ( 223 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

The current challenges facing the field of artificial intelligence include high prices and market supply disruptions. The traditional single-card, single-use model results in low resource utilization and efficiency. Furthermore, existing technological research methods make it difficult to support the efficient management and scheduling of diverse heterogeneous GPU resources. Based on this, a virtualization-based GPU heterogeneous resource pool platform was proposed. Firstly, the overall architecture, logical architecture, and functional architecture of the platform were planned and designed. Secondly, key technologies were studied, and a virtualization heterogeneous GPU resource pool framework and a scheduling model based on time slicing + load balancing were proposed. Finally, based on the methods described, various innovative application models were proposed, including multiservice single-card stacking, cross-pull, cross-machine integration, hybrid deployment, and time division multiplexing. The research method proposed provides enterprise-level AI applications with GPU computing resources that are compatible with multiple GPU manufacturers, support remote access, flexible partitioning and aggregation, and flexible scheduling. Following the completion of calculations and an in-depth analysis, it has been demonstrated that a reduction of up to 60% in the number of GPU cards can be achieved while simultaneously enhancing operational efficiency by a factor of four.

Research on public network and private network diversion scheme for 4G/5G users

Hongyuan MA, Zheng FENG, Wei ZHOU, Xuguang SONG, Xiaole YANG

2024, 40(9):  176-184.  doi:10.11959/j.issn.1000-0801.2024187

Asbtract ( 96 )   HTML ( 14)   PDF (2452KB) ( 187 )   Knowledge map   

Figures and Tables | References | Related Articles | Metrics

Facing the demand of 4G/5G users accessing public network services and private network services simultaneously, the limitations of the existing technical schemes in the industry were reviewed, and a network-side non-inductive diversion scheme was proposed. It only needed to establish a public network IP service channel by the UE. When the network detected that users used private network services, a private network IP service channel for users on the network side was established by SMF/PGW-C to divert the private network services of users. There was only one IP channel in the mobile phone terminal to transmit public network services and private network services, corresponding to one public network IP channel and one private network IP channel on the network side, and SMF/PGW-C completed the mapping and splicing of the user-side IP channel and the corresponding network-side public network IP channel and private network IP channel, as well as the replacement of the user’s IP address, thus realizing the diversion of the user’s public network services and private network services. This scheme not only meets the demand of dual-domain private network services, but also allows for the separate management and charging of users’public network service IP channels and private network service IP channels by the network side, and has no additional requirements for terminals.