Current Issue

• • 2024, Vol. 47 No. 3 Published date:30 June 2024 Last issue Next issue

  • Terminal-Side Computing Power Network: Architecture, and Key Technologies
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 1-9.
  • Abstract ( 153 )     　　
  • Computing power network has become one of the significant methods to address the imbalance between computing power supply and demand by integrating multidimensional resources such as computing power, networks, and data. Meanwhile, with the rapid progress of terminal device technology, the ubiquitous terminal-side computing power has not yet been fully exploited. Based on the analysis of the enhancement of future pan terminal capabilities and the transition from passive to active, the architecture and functional features of Terminal-Side Computing Power Network are proposed. Targeting the characteristics of terminal devices, the main key technologies of Terminal-Side Computing Power Network are elaborated in detail, including perception of terminal behavior and computing power, computing power switching based on mobility prediction, service orchestration technology for differentiated terminal services, task deconstruction and computing power scheduling, trusted transactions between active terminals, etc. Through the study of a cloud-edge-end collaborative computing power network simulation environment, the enormous computing power potential of Terminal-Side Computing Power Network has been verified. Finally, the challenges faced by Terminal-Side Computing Power Network were analyzed, providing reference for the future development directions.
  • Supplementary Material | Related Articles

  • Research on Semantic Coding Algorithm Based on Semantic Importance
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 10-16.
  • Abstract ( 95 )     　　
  • In order to meet the demand of efficient information transmission between intelligent communication entities, semantic communication research is faced with the basic problems of semantic information representation, encoding and transmission. In view of the communication resource shortage and task delay sensitive transmission scenario of everything intelligent link, a semantic coding algorithm based on semantic importance was proposed, and a semantic communication experiment platform was designed and built to realize semantic information transmission. Firstly, the intelligent task-oriented semantic communication system model was established based on the basic process of wireless communication. Secondly, a semantic importance extraction method was designed to obtain the importance weights of the feature maps to the task results by aggregating the channel features and adaptively calibrating the feature responses. Then, a semantic compression method was designed to filter the feature maps according to the importance weight in the actual transmission to achieve semantic data compression. Finally, the performance of the proposed algorithm was verified on the semantic communication platform by taking the task of surface defect classification of hot rolled strip steel in the industrial Internet scene as an example. Experimental result shows that the proposed algorithm has the advantages of small data transmission, strong anti-noise ability and short task processing delay.
  • Supplementary Material | Related Articles

  • In-Band Full Duplex Self-Interference Elimination Algorithm Based on Adaptive Filtering#br#
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 17-23.
  • Abstract ( 77 )     　　
  • In order to ensure the effective operation of in-band full-duplex communication system, aiming at the saturation problem of low noise amplifier and analog to digital conversion module in the system, combined with the nonlinear characteristics of power amplifier and the multipath effect of signal transmission, a two-stage self-interference elimination algorithm based on adaptive filtering algorithm is proposed. It is eliminated in analog domain and digital domain respectively. On this basis, a low complexity filter initial weight updating method is proposed to improve the recursive least squares algorithm to solve the problem that the traditional adaptive filter converges slowly when the channel changes. Simulation results show that the proposed algorithm can suppress the power of the self-interfering signal to the noise level, ensure the correct demodulation of useful signals, and effectively reduce the time required for the self-interfering signal elimination.
  • Supplementary Material | Related Articles

  • Study on the System Performance in Cognitive NOMA-V2V Networks with Outdated CSI#br#
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 24-29.
  • Abstract ( 66 )     　　
  • An outdated channel state information (CSI) modeling method based on geometry is proposed for vehi-cle-to-vehicle (V2V) channels. To address the bottleneck problems of high-speed time-varying characteristics and low resource utilization of V2V communication networks, a power allocation scheme of cognitive non-orthogonal multiple access (NOMA) V2V networks based on outdated CSI and imperfect successive inter-ference cancellation (SIC) is studied. The performance impact of outdated CSI on V2V communication networks based on a geometric three-dimensional (3D) two-cylinder channel model is studied, and the throughput region of cognitive NOMA V2V networks is analyzed. Under Quality of Service (QoS) constraints as to achievable rate and outage probability of primary user (PU) and secondary user (SU), a power allocation scheme is proposed. Simu-lation results show that the proposed scheme can effectively increase throughout of SU, improve outage perfor-mance of the system. The simulation results show that compared with the existing power allocation methods, the proposed scheme has higher throughput and better outage performance for V2V communication networks with outdated CSI.
  • Supplementary Material | Related Articles

  • Mutual Learning Prototype Network for Few-shot Text Classification
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 30-35.
  • Abstract ( 79 )     　　
  • Few-shot prototype networks are regarded as one of the effective methods to solve few-shot text classification problems. However, existing methods usually rely only on a single prototype for training and inference, which is susceptible to noise and other factors, resulting in insufficient generalization ability. To address this problem, a Mutual Learn-ing-Prototype Network(MLProtoNet) for small-sample text classification is proposed. On the basis of retaining the ex-isting algorithm to compute the prototype directly by text embedding features, thie paper introduces the BERT network, which inputs the text embedding features into BERT to generate a new prototype. Then, using the mutual learning algorithm, the two prototypes are mutually constrained and knowledge is exchanged to filter out the inaccurate semantic information. This process aims to enhance the feature extraction capability of the model and improve the classification accuracy by joint decision making of the two prototypes. Experimental results on two few-shot text classification da-tasets confirm the effectiveness of our proposed approach. Specifically, on the FewRel dataset, our method improves the accuracy by 2.97% in the 5-way 1-shot experiment compared to the current optimal method, and by 1.99% in the 5-way 5-shot experiment.
  • Supplementary Material | Related Articles

  • Intent-Based Network Policy Generation for Knowledge Definition
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 36-41.
  • Abstract ( 57 )     　　
  • To achieve on-demand services in all scenarios in the sixth-generation mobile communication system, the complexity and diversity of networks and services greatly affect the reliability and robustness of existing methods. Therefore, a framework that combines intent translation and network slicing is proposed for universal application in all scenarios. Firstly, given the diversity of services, a multi-step intent translation method that utilizes external knowledge and deep learning is proposed. Secondly, a multi-path resource scheduling algorithm based on mixed integer linear programming is proposed to improve the overall utilization of network resources and effectively respond to changes in network environments. Experimental results show that the proposed framework can achieve accurate division of business resources based on multi-tenant intent to ensure user experience and provide on-demand services in all scenarios.
  • Supplementary Material | Related Articles

  • DuC-GAN:A Novel Model for Enhancing GAN Training Stability
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 42-47.
  • Abstract ( 192 )     　　
  • This paper proposes a new framework, called "Dual-Cycle GAN(DUC-GAN)" to enhance the stability of training in generative adversarial networks (GAN). The framework addresses the issue of training instability in GAN by introducing an additional cycle between the generator and discriminator. This new cycle is composed of a frozen original discriminator and a new discriminator, both of which are trained together with the generator and switched based on the generator's performance. Testing on multiple datasets has shown that the proposed framework significantly improves the performance and training stability of GAN compared to existing methods, achieving faster convergence and better generation quality. The framework is also applicable to other GAN variants and is expected to become an important tool for future GAN research.
  • Supplementary Material | Related Articles

  • Reference-Modulated Code Shifted Differential Chaos Shift Keying Modulation Scheme
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 48-54.
  • Abstract ( 47 )     　　
  • A simple reference-modulated code shifted differential chaos shift keying modulation scheme is proposed, in which additional information bits are delivered by reference signals in code shifted differential chaos shift keying (CS-DCSK) communication system. In this proposed system, reference signals and information signals are modulated by independent code index modulations, which helps to greatly improve the bit transmission rate. According to the unique structure of CS-DCSK signals, a novel detection method is designed to reduce noises in both the reference and information signals, in which received signals are averaged based on segments before correlation. The bit error rate performance of the proposed system is analyzed theoretically and verified by simulation results in the additive Gaussian white noise and multipath Rayleigh fading channels. Results show that the proposed scheme can obtain not only higher bit transmission rate but also better bit error rate performance with reduced system complexity.
  • Supplementary Material | Related Articles

  • Collaborative caching mechanism of nodes based on cooperative game and deep learning
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 55-61.
  • Abstract ( 59 )     　　
  • With the development of wireless mobile communication, the contradiction between users' proliferating content demands and the limited wireless network resources is increasing. The use of Device-to-Device (D2D) communication to realize the sharing of cached contents between edge nodes can improve user experience and reduce the burden of traffic on the core network. This paper models the collaboration caching problem as a cooperative game considering the factors of interaction costs and individual rationality to optimize the system utility with limited cache space. According to whether the utility between nodes can be transferred, we discuss the cooperative game in two cases. Under the transferable utility (TU) game, the conditions for nodes to form a stable grand coalition are derived, and it is proved that the coalition has the nature of nuclear nonempty when the coalition cost of nodes satisfy certain conditions. For non-transferable utility (NTU) game, the rational nodes cannot ensure the formation of a stable grand coalition, and the number of formable coalitions increases dramatically with the number of users. Therefore, a deep reinforcement learning-based coalition formation algorithm is proposed to ensure the formation of stable coalitions within a limited time. Theoretical analysis and simulation results show that the proposed algorithm can converge to a Nash-stable optimal solution or asymptotically optimal solution, which outperforms other comparison algorithms.
  • Supplementary Material | Related Articles

  • Dynamic channelization structure of WOLA filter banks based on FRM
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 62-68.
  • Abstract ( 53 )     　　
  • Due to the problem of high complexity of prototype filter design in weighted overlap-add (WOLA) filter bank dynamic channelization structure, this paper introduces frequency response masking (FRM) into the dynamic channelization structure, and proposes a WOLA filter bank dynamic channelization structure based on FRM. In this structure, first of all, FRM technology is enhanced, and a new FRM is proposed to design a filter with complementary power features. Secondly, the filter designed by FRM is used as the prototype filter of WOLA filter bank dynamic structure. Lastly, the channels overlap to reduce signal distortion due to the blind zone. The simulation results show that this structure can realize dynamic signal reception and has a good reconstruction effect. Moreover, compared with the prototype filter designed directly to meet the reconstruction characteristics, the FRM design filter saves 79% of multiplier resources.
  • Supplementary Material | Related Articles

  • Fourier semi-supervised learning method for medical image segmentation
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 69-74.
  • Abstract ( 74 )     　　
  • The scarcity of labeled data is a challenging problem that affects the segmentation accuracy of medical images. Aiming to solve this problem, we propose a semi-supervised learning method based on fourier transform and consistent constraint, In the case of a small amount of annotated data, the output of unannotated data via Fourier transform interpolation and model segmentation is spa-tially consistent with the output of reverse operation, and the consistency regularization constraint for unannotated data is con-structed to improve the model performance of fully supervised learning. The experimental results on public datasets ACDC, Syn-apse and CTLN show that the proposed algorithm is superior to baseline methods and can be integrated with existing SOTA semi-supervised medical image segmentation methods to improve their segmentation performances.
  • Supplementary Material | Related Articles

  • Graph convolutional anomaly detection method for multivariate time series with fusion of explicit and implicit associations between sequences
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 75-82.
  • Abstract ( 72 )     　　
  • As the existing reconstruction-based anomaly detection methods cannot extract coupling relationships between components in cyber physical systems (CPS) well and tends to cause information omission, a graph convolutional method based on explicit and implicit associations fusion (GCEIAF) for multivariate time series anomaly detection is proposed. The improved cosine similarity function is used to extract explicit associations that can be measured by distance. An association extraction module based on multi-head self-attention mechanism is designed to capture implicit associations in a learnable way. The two kinds of associations are integrated to build a fusion graph, which is fed into a graph-convolutional autoencoder along with raw data to reconstruct time series combining temporal and spatial dependencies. The anomaly score of CPS can be calculated based on the reconstruction result, and then the anomalies are detected using the adaptive threshold selection algorithm. Experimental results on four public datasets indicate that GCEIAF outperforms the state-of-the-art and latest methods in terms of F1-Score. The experimental results also show that GCEIAF can interpret and analyze abnormal events by outputting the association weight matrix.
  • Supplementary Material | Related Articles

  • GFDM System Low-complexity Algorithm Based on Sparse Matrix
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 83-89.
  • Abstract ( 40 )     　　
  • A low-complexity modulation and demodulation algorithm based on sparse matrix is proposed to address the issue of high complexity in generalized frequency division multiplexing (GFDM) systems, which restricts its ability to meet the requirements of various application scenarios in future communication systems. Firstly, the system model of GFDM is analyzed, presenting two different modulation structures. Subsequently, the sparsity of the frequency domain coefficients of the prototype filter is leveraged to make the modulation and demodulation matrices sparse in frequency domain using different Fourier transforms. By rearranging the sparse matrices, various block diagonal matrices can be constructed. Finally, the block diagonal matrix is used to modulate and demodulate the data vector transformed to frequency domain, which can effectively reduce the computational complexity of GFDM system. Simulation results demonstrate that both low-complexity algorithms achieve good complexity performance under different numbers of subcarriers, while showcasing the advantages of GFDM's flexible time-frequency structure.
  • Supplementary Material | Related Articles

  • A Reconfigurable Intelligent Surface Assisted Secure Communication Secrecy Performance Optimization Scheme#br#
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 90-95.
  • Abstract ( 53 )     　　
  • In a reconfigurable intelligent surface (RIS) assisted secure communication system, to improve the secrecy perfor-mance of the system and overcome the hardware impairments caused by the system transceiver, a design method for optimizing the secrecy performance of RIS assisted secure communication is proposed. Considering a multiple-input single-output scenario with multiple eavesdroppers, the secure communication secrecy rate maximization problem is established by jointly optimizing the base station transmit beamforming and RIS phase shift beamforming under the maximum transmit power of the base station and RIS phase shift unit-mode constraints. For the nonconvexity and coupling of this optimization problem, an efficient algorithm based on alternating optimization and semidefinite re-laxation is proposed to transform the original nonconvex optimization problem into two convex optimization sub-problems and iteratively solve them to obtain the suboptimal solution of this optimization problem. Simulation results show that the proposed scheme in this paper can obtain higher secrecy rate compared with the conventional secure communication beamforming scheme.
  • Supplementary Material | Related Articles

  • Energy Consumption Optimization for Air-Ground Cooperation Based Covert Mobile Edge Computing System
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 96-102.
  • Abstract ( 50 )     　　
  • Aiming at the problem that it is difficult to obtain the accurate location information of the warden node Willie in practical applications, the robust energy consumption optimization algorithm for air-ground cooperation based covert mobile edge computing is presented. Specifically, under the condition that only the estimated location information of Willie is known, in order to robustly hide the computational offloading behavior between UAV and GBSs, the covertness constraints are analyzed from the perspective of the warden node Willie and the UAV, respectively. Then the weighted total energy consumption of the UAV is minimized by jointly optimizing the computation task allocation factor, the UAV-GBS association factor, the UAV power and the UAV trajectory. To tackle the formulated non-convex optimization problem, an efficient three-stage alternating iterative optimization algorithm with parameter based on block coordinate descent method, S-procedure and successive convex approximation method is proposed. Simulation results show that the proposed algorithm can save energy consumption and offload more computing tasks to the GBSs under given covert rate constraints, and has a desirable convergence.
  • Supplementary Material | Related Articles

  • A Neural Network Architecture for Learning Invariant Representations
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 103-110.
  • Abstract ( 66 )     　　
  • The presence of distortions in data refers to the fact that different input feature vectors may represent the same entity, which is one of the long-standing difficulties in machine learning. The study of the above problem has spurred the development of invariant machine learning methods with the abilities such as ignoring translation, rotation, illumination, and pose changes in images, which typically use pre -defined invariant features or invariant kernels, and require the designers to carefully analyze the types of distortions that may exist in the data. While it is straightforward to discover the possible types of distortions in image data, it is difficult in other domains. Our goal is to learn invariant representations from non-image data based only on information about whether any two samples are distorted variants of the same entity, without any information of what the distortions present in the data. In theory, given a sufficiently large number of samples, standard neural network architectures should be capable of learning invariance from data. In practice, we experimentally find that standard neural networks are struggling to learn to approximate even simple type of invariant representation. Therefore, we propose a new extended layer with richer output representations that is better suited for learning invariance from data.
  • Supplementary Material | Related Articles

  • Optimized Routing Algorithm for Latency Difference Between Work and Backup Paths in Power Optical Networks Based on Risk Balance#br#
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 111-116.
  • Abstract ( 50 )     　　
  • With the continuous development of smart grid, more and more new types of power communication services appear gradually. The traditional power communication optical transmission network based on Wavelength division multiplexing technology is difficult to meet the multi-granularity business transmission scenario. Elastic optical network can achieve network resource scheduling at the subwavelength level. It is the key technology of next generation power communication optical network. In order to solve the problem that the existing routing methods can not take into account both the protection requirements of power communication services and the elastic optical network routing and spectral allocation, a mathematical model for the work and backup routing of optical power communication network is proposed to describe the optimization goals and constraints of the work and backup routing conditions, such as the latency difference of the work and backup paths, the risk balance of the whole network, etc. Then, based on this mathematical model, an optimal routing algorithm based on risk balance for backup and work routes is presented. The algorithm is designed to find a pair of backup dual routes with low latency and balanced risk for power communication services. The results show that this method not only has a lower latency difference and blocking rate of work and backup routes (52.2% lower than the comparison algorithm), but also performs well in risk balance (up to 57.5% reduction).
  • Supplementary Material | Related Articles

  • Image Fusion Based on Improved Maximum Entropy Segmentation Algorithm and Rolling Guidance Filter
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 117-123.
  • Abstract ( 44 )     　　
  • With the change of application environment, the engineering deployment capability of most fusion algorithms for infrared and visible images is generally poor, and there are one or more problems such as insufficient target extraction, loss of details, algorithm complexity, low efficiency, and poor applicability. Aiming at the above problems, a fusion method of infrared and visible images is proposed based on improved maximum entropy algorithm (IMES) and rolling guided filter (RGF). First, the infrared target is extracted using IMES, and the visible image and infrared image are decomposed into basic layer and detail layer using the scale perception and edge preservation characteristics of RGF. Then, the base layer fusion image is obtained from the extracted infrared target and visible base layer image by the base layer fusion rules. Finally, the final fusion image is obtained from the base layer fusion image by the detail layer fusion rules. Experimental results show that the proposed algorithm has a clear target, clear texture details and rich detail information in the fused image. Moreover, the proposed algorithm is simple, efficient, and wide applicability. Compared with the other four algorithms, the proposed algorithm has advantages in both subjective and objective evaluations, and has certain engineering deployment capability.
  • Supplementary Material | Related Articles

  • Multi-target beam assignment algorithm for IRS-assisted millimeter wave MIMO systems
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 124-129.
  • Abstract ( 59 )     　　
  • A multi-objective optimized active and passive beam assignment algorithm based on energy efficiency (EE) and spectral efficiency (SE) trade-offs is proposed for the problem of low rank-induced MIMO multiplexing gain degradation in distributed intelligent reflecting surface (IRS)-assisted millimeter wave MIMO systems. First, a multi-objective optimization framework for EE and SE is developed using the linear weighting method. Secondly, the alternating optimization (AO) algorithm is used to jointly optimize active beamforming (ABF) and passive beamforming (PBF). Finally, the semi-positive definite relaxation algorithm (SDR) is used to solve the non-convex unit mode constraint problem of IRS. Simulation results show that the algorithm can flexibly adjust the trade-off coefficient according to the actual communication service to achieve the balance between EE and SE, and effectively improve the communication performance of millimeter wave MIMO system (mm Wave MIMO).
  • Supplementary Material | Related Articles

  • Real-time Optimization Algorithm of Ppower Allocation and Eenergy Scheduling for Base Station
  • Journal of Beijing University of Posts and Telecommunications. 2024, 47(3): 130-136.
  • Abstract ( 60 )     　　
  • In recent years, with the rapid development of mobile terminal equipment and cellular communication technology, the problem of huge increase in electricity consumption and high electricity cost becomes increasingly prominent. Aiming at the base station equipped with renewable energy sources and energy storage devices and connected to smart grid, this paper studies the real-time optimization of power allocation and energy scheduling for downlink communication in such base station, with the goal of reducing the power purchase cost of the base station. Considering the randomness of data arrival at the base station, the fluctuation of channel state, the intermittenity of renewable energy output and the time-variability of electricity price of smart grid, the power allocation and energy scheduling model of a base station in downlink communication is constructed under the constraint of energy storage causality and user maximum tolerance. Then a low-complexity real-time optimization algorithm is proposed based on the improved Lyapunov optimization theory. Through real-time allocation of transmission power and energy scheduling, the power purchase cost of the base station is minimized. At the same time, the data transmission service is guaranteed within the time delay that users can tolerate. Theoretical analysis shows that the proposed algorithm can make real-time decision only according to the current system state, and the optimization result is infinitely close to the optimal value. Finally, the simulation results show that the proposed algorithm can effectively reduce the cost of electricity purchase for network operators, and the cost of electricity purchase can be reduced by 37.1%, 29.8% and 15.7%, respectively, compared with the benchmark greedy algorithm.
  • Supplementary Material | Related Articles