VTC 2025 Spring Conference’s Shorts

Optimized Multi-Scale Semantic Parameter Selection and Transmission for Vehicular Edge Com...

With the advancement of intelligent driving technology, vehicular networks generate vast amounts of decentralized data that need to be processed. As a distributed paradigm, Federated Learning (FL) enables data integration and processing across various vehicles. However, traditional FL methods face significant challenges in vehicular networks, including high communication overhead and the difficulty of meeting strict latency and reliability requirements. To address these challenges, we propose a Multi-scale Semantic Selection-based FL (MSSFL) scheme, which integrates multi-scale semantic parameter selection and transmission optimization to reduce the system’s communication cost. The proposed scheme selects parameters with high semantic importance and allocates bandwidth proportionally based on their quantity to enhance communication efficiency. We further formulate an optimization problem to minimize both parameters’ transmission cost and upload delay. To solve this problem, we develop an alternating iterative solution using the block coordinate descent (BCD) method, which alternately optimizes the semantic parameter selection and bandwidth allocation strategy. Experimental results validate the effectiveness of the proposed framework in enhancing both communication efficiency and model accuracy.Optimized Multi-Scale Semantic Parameter Selection and Transmission for Vehicular Edge Computing NetworksXiao Wang, Yunlong Lu, Hao Wu, Beijing Jiaotong University; Yueyue Dai, Huazhong University of Science and Technology; Yaru Fu, Hong Kong Metropolitan University

Fair Resource Allocation with Dynamic Multi-Layer Service Area Management in Local Network...

Cybernetic avatars (CAs) are set to revolutionize society by enhancing human physical and cognitive capabilities through advanced ICT and robotics, as part of Japan’s Moonshot Goal 1 project. CAs enable individuals to overcome physical and spatial limitations, facilitating participation in remote and collaborative activities. This effort is part of building a reliability-ensuring platform using Local 5G networks that guarantees seamless CA teleoperation. However, when numerous CAs operate concurrently, limited wireless communication capacity can significantly degrade service quality and fairness in resource allocation. To address this challenge, we propose a dynamic multi-layer service area adjustment method that enhances fairness by dynamically adjusting coverage area boundaries in real time. This method optimizes resource block (RB) allocation efficiency, significantly improving fairness as measured by Jain’s Fairness Index under varying network loads. Comparative performance evaluations using system-level simulations against the traditional fixed coverage service area method demonstrate that the dynamic approach effectively balances system throughput and resource utilization, offering a robust and adaptive solution for emerging Local 5G network deployments.Fair Resource Allocation with Dynamic Multi-Layer Service Area Management in Local Networks Supporting Cybernetic AvatarsArif Dataesatu, Atsushi Wakayama, National Institute of Information and Communications Technology; Kazuo Ibuka, Homare Murakami, NICT; Takeshi Matsumura, National Institute of Information and Communications Technology

Digital-Twin-Enabled Channel Access and Power Control for Smart Grids in Communication Net...

In this paper, we propose a novel channel access and power control scheme for smart grids in communication networks. The scheme is named digital twin-based memory recall optimization (DMRO), which aims to extract meaningful patterns from noisy network traffic measurements and support more sophisticated decision-making processes for optimizing channel access and power control for smart grids. Specifically, we design a pattern extraction method that minimizes the Frobnius norm between the collected measurements and the expected k-rank approximation of the measurements in order to extract useful information. Then, considering the interference and signal-to-interference-plus-noise ratio (SINR) constraints in the wireless environment, we develop a digital twin-based distributed channel access and power control scheme to improve the latency taming and energy utilization efficiency of the phasor measurements units (PMU). We consider both the real-time traffic prediction and the paired optimization scheme on the digital twin side, and utilize memory recall to enhance local model robustness to optimize from a more diverse set of situations by replaying underrepresented experiences. Simulation results demonstrate that the proposed DMRO scheme can achieve high traffic prediction accuracy and improve the latency taming and energy utilization efficiency even increasing industrial channel interference or the number of PMUs.Digital-Twin-Enabled Channel Access and Power Control for Smart Grids in Communication NetworksQihao Li, Jilin University; Qiang (John) Ye, University of Calgary; Fengye Hu, College of Communication Engineering, Jilin University

Coded Distributed Computing over Multi-server Clustered Network for Federated Learning

In this paper, we focus on the application of coded distributed computing (CDC) in a multi-server clustered network (MSCN), which is designed to accelerate the gradient update process in federated learning (FL) by considering both communication and computational heterogeneity. As the number of participating devices increases and resource heterogeneity becomes more pronounced, reducing total execution latency (TEL) has become a critical challenge. To address this issue, we focus on optimizing the matching between heterogeneous devices and server task loads to improve resource utilization while enhancing the robustness and fault tolerance of the FL system. To minimize TEL, we propose a greedy algorithm and an iter-genetic algorithm for device assignment, named GADA and IGADA, based on the task allocation for the single server (TASS) algorithm, respectively. Based on simulation results and theoretical analysis, we confirm that our proposed algorithms substantially reduce the TEL in various scenarios compared to existing CDC methods, with complexity markedly lower than that of the exhaustive scheme.Coded Distributed Computing over Multi-server Clustered Network for Federated LearningWenjing Mou, Harbin Institute of Technology, Shenzhen; Shushi Gu, Harbin Institute of Technology (Shenzhen); zhikai zhang, Pengcheng Laboratory; Guixiang Lei, Harbin Institute of Technology (Shenzhen); Zhang Qinyu, Harbin Institute of Tech.; Wei Xiang, La Trobe University

Bundle Transmission Control in Multi-Source Delay/Disruption Tolerant Near-Earth Satellite...

The use cases of satellite communications are rapidly expanding towards 6G technology. Long propagation distances and dynamic changes in the network topology create an unstable environment with greater delays and disruptions than those of terrestrial networks. Delay/disruption-tolerant networks (DTN) have been developed to address these challenges. Since the near-earth satellite network is a scheduled network, contact plans can be created beforhand. Contact graph routing (CGR) is a contact-plan-based method used to determine the shortest path to deliver a bundle, a data unit of a DTN, to its destination. However, the conventional contact graph-based bundle transmission control methods do not consider the transfer order or bundle generation from multiple sources, such as observation satellites, ground, and marine users. Therefore, the bundle delivery ratio will decrease in future satellite networks, in which traffic of various sizes and acceptable delays are expected to flow in. In this study, we propose a bundle transmission control method for a multi-source DTN that considers the bundle variety. The simulation evaluation results demonstrated that the proposed method achieved a higher bundle delivery ratio than the conventional methods.Bundle Transmission Control in Multi-Source Delay/Disruption Tolerant Near-Earth Satellite Networks for Improved Delivery RatioKazuma Mashiko, Hiroaki Hashida, Yuichi Kawamoto, Nei Kato, Tohoku University; Yohei Hasegawa, Masayuki Ariyoshi, NEC Corporation

Prediction-based Task Allocation and Processor Control for Distributed Green Data Centers ...

The scale of data centers has increased significantly in recent years, with their energy demands adversely impacting the environment. Consequently, distributed green data centers, equipped powered renewable energy, have gained considerable attention. However, conventional optical fibers installations for such data centers incur high costs and face locations constraints. Therefore, this study explores the use of optical satellite communications for distributed green data centers. Efficient task allocation is crucial to minimize service delays. However, task transmission time depends on satellite–ground link performance, while processing time is affected by the energy at data centers. Additionally, the non-linear relationship between the server performance and power consumption highlights the inefficiency, task allocation without future-aware considerations. To address this, we propose a method for task allocation and processing performance control that minimizes service delays through predictions of task generation, link performance, and power generation. In the proposed method, we formulate an optimization problem based on prediction data and find its solution by exploration. Simulations demonstrate that the proposed method significantly reduces tasks transmission and processing times.Prediction-based Task Allocation and Processor Control for Distributed Green Data Centers with Optical Satellite LinksHiroto Oshima, Hiroaki Hashida, Yuichi Kawamoto, Nei Kato, Tohoku University; Kazushi Sugyo, Yohei Hasegawa, Masayuki Ariyoshi, NEC Corporation

IRS-Aided Microwave Power Transfer Using Curve Fitting-Based Phase Optimization

This study aims to enhance the transmission efficiency of microwave power transfer, which has garnered significant attention as a method of wireless power supply for IoT devices and drones, by using Intelligent Reflecting Surface (IRS). IRS is a plate-shaped device that can control the reflection characteristics of radio waves, and it is expected to improve the transmission efficiency when placed between a wireless transmitter and a wireless receiver. However, existing methods of reflection phase optimization for IRS are predominantly based on sequential feedback or a small number of discrete reflection phases, both of which have problems, including feedback overhead and beam shape limitations. In this paper, we propose a novel reflection phase optimization method based on curve fitting. This approach is capable of continuous reflection phase control and optimization of the reflection phase of all reflective elements simultaneously. Through numerical simulation and actual experiments using an implemented IRS prototype, it was observed that the proposed method enhanced the received power by up to 4 dB compared to a conventional sequential optimization method. However, it was also observed that the range limitation of the reflection phase in the implemented IRS hindered the effectiveness of the proposed method.IRS-Aided Microwave Power Transfer Using Curve Fitting-Based Phase OptimizationAkise Kumashiro, Kazuhiro Kizaki, The University of Osaka; Takuya Fujihashi, Osaka University; Shinya Sugiura, The University of Tokyo; Hiroki Wakatsuchi, Nagoya Institute of Technology; Takashi Watanabe, Osaka University; Shunsuke Saruwatari, University of Osaka

Age of Information Analysis for NOMA-Aided Grant-Free Transmissions

This paper aims to study the impact of non-orthogonal multiple access (NOMA) assisted grant-free transmissions on reducing age of information (AoI) in status updating systems. Particularly, an uplink communication scenario is considered, where multiple users upload their status updates to a destination competitively. To mitigate collisions among users, K reception signal-to-noise ratio (SNR) levels are pre-configured, which can be chosen randomly by the users. By applying NOMA, successive interference cancellation (SIC) is carried out at the receiver to decode the signals from different users sequentially. Different from most existing works which adopt generate-at-will (GAW) model for modeling the generating process of status updates’ arrivals, this paper considers a more general model to characterize the randomness of the status updates’ arrivals. Closed-form expressions for AoI achieved by the proposed NOMA assisted grant-free scheme is obtained. Numerical results are provided to validate the accuracy of the analytical results and also demonstrate the superior performance of the proposed scheme in reducing AoI compared to conventional OMA based methods.Age of Information Analysis for NOMA-Aided Grant-Free TransmissionsYanglin Ye, Yanshi Sun, Momiao Zhou, Hefei University of Technology; Zhiguo Ding, Khalifa University; Zhengqiong Liu, Hefei University of Technology

A GNN-based Discrete Choice Experiment for Heterogeneous Computational Task Offloading

A multitude of 6G smart devices will integrate applications that require contextual data processing, such as healthcare remote sensing and interactive applications. These are computationally intensive tasks that require quick execution within stringent time constraints. Given the heterogeneity of these applications as well as the communication and computation uncertainty due to the shared nature of the edge environment, computation competition and access interference pose serious challenges in the offloading of computational tasks. In this respect, the offloading decision-making of edge devices is influenced and shaped by prospect-theoretic characteristics. To address this challenge, a graph neural network (GNN)-based discrete choice experiment (DCE) is proposed to model the choice of the appropriate offloading server based on a feature — attribute matching. An encoded vector that is validated using interference, channel gain, and transmission rates, is encoded into the embedding space, and offloading solution for each edge device is obtained. The performance evaluation results show that the proposed GNN-based DCE lowers the expected overheads, which maximizes the amount of offloaded computational tasks.A GNN-based Discrete Choice Experiment for Heterogeneous Computational Task OffloadingMduduzi Comfort Hlophe, Sunil Maharaj, University of Pretoria

A closed-loop 2×4 downlink MIMO Framework for 5G New Radio using OpenAirInterface

We present the first-of-a-kind closed-loop 2 × 4 MIMO implementation for the downlink of 5G Open RAN using OpenAirInterface (OAI), which is capable of transmitting up to two transmission layers. Our implementation is a fully functional 5G New Radio (5G NR) system, including the 5G Core Network (5G CN), 5G Radio Access Network (5G RAN), as well as 5G NR User Equipment (UEs). This serves as a foundational framework for further advancements in the context of emerging Open RAN (O-RAN) development. A key feature of our implementation is the enhanced Channel State Information (CSI) reporting procedure at the UE, which includes Rank Indicator (RI), Precoding Matrix Indicator (PMI), and Channel Quality Indicator (CQI). It is adjusted for the extended configuration to maximize data rates. To demonstrate the performance of our implementation, we measure the downlink data rates using iperf3 in two scenarios: (i) fixed channels to assess two-layer data transmission and (ii) Rice1 channels for general transmission analysis. The obtained simulation results demonstrate that, compared to the existing 2×2 MIMO configuration in the OAI, our implementation improves the data rates in almost all scenarios, especially at the high Signal-to-Noise-Ratios (SNRs).A closed-loop 2×4 downlink MIMO Framework for 5G New Radio using OpenAirInterfaceDuc Tung Bui, Le-Nam Tran, University College Dublin