As currently defined, IEEE Network covers the following areas: 1. network protocols and architectures, 2. Protocol design and validation, 3. Communication software and its development and test, 4. Network control and signalling, 5. network management, 6. Practical network implementations including local area networks, (LANs), metropolitan area networks (MANs), and wide area networks, (WANs), 7. Switching and processing in integrated (voice/data) networks and network components, 8. Micro-to-host communication.

Special Issue on Edge-Enabled Collaboration Between Large-scale and Lightweight Models in Sensor-Cloud Networks
截稿日期: 2025-10-01

With the swift evolution of Internet-of-Things (IoT) technology, the sensor-cloud technique has emerged as a pivotal bridge to connect the physical and digital realms, powerfully driving the progress of intelligent networks. However, traditional data processing paradigms often grapple with constraints like high latency and substantial energy consumption, particularly when faced with an overwhelming volume of data. Herein, edge intelligence, a synergistic integration of edge computing and artificial intelligence techniques, has arisen as a potent remedy. This innovation brings about transformative changes through its low-latency characteristics and high energy efficiency while bolstering data security by prioritizing privacy safeguards. 

Large Language Models (LLMs), such as ChatGPT and Sora, have attracted great attention and achieved huge success in academia and industry recently. Furthermore, the synergy between large-scale and lightweight models exhibits tremendous promise for tackling intricate issues and enhancing decision-making precision in sensor-cloud networks. This synergy enables effective data processing and analysis even on resource-constrained edge devices. More precisely, large-scale models excel at processing and analyzing complex data yet demand more substantial computational power and storage. In contrast, lightweight models offer reduced computational complexity and compact storage requirements, fitting the scenarios demanding high real-time responsiveness and minimal resource availability. A current focal point of research revolves around how to effectively orchestrate collaboration between large-scale and lightweight models within sensor-cloud networks by harnessing the unique strengths of the twomodel types.

This Special Issue (SI) delves deeply into cutting-edge technologies, application challenges, and developmental trajectories concerning the edge-enabled collaboration between large-scale and lightweight models in sensor-cloud networks, providing readers with a comprehensive, penetrating, and visionary discourse. All topics relevant to advancing edge-enabled collaboration between large-scale and lightweight models are of interest. Consequently, the topics of interest include but are not limited to the following:

    Novel IoT architectures that enable edge intelligence in sensor-cloud networks
    Edge intelligence technologies and their applications, such as object detection, augmented/virtual reality, and artificial intelligence-generated content.
    Strategies for optimizing and deploying large-scale and lightweight models in sensor-cloud networks.
    Dynamic resource scheduling for large-scale and lightweight model collaboration in sensor-cloud networks.
    Load balancing strategies for large-scale and lightweight model collaboration in sensor-cloud networks.
    Collaborative training and inference of large-scale and lightweight models in sensor-cloud networks.
    Model compression and acceleration for large-scale and lightweight model collaboration.
    Distributed learning and collaborative inference with different-scale models in sensor-cloud networks.
    Distributed sensing and collaborative decision-making in sensor-cloud networks.
    Innovation and practice of collaboration between large-scale and lightweight models in sensor-cloud networks.
    Tailored solutions for smart cities utilizing collaboration between large-scale and lightweight models.

Submission Guidelines

Manuscripts should conform to the standard format as indicated in the “Information for Authors” section of the Paper Submission Guidelines.

All manuscripts to be considered for publication must be submitted by the deadline through the magazine’s IEEE Author Portal submission site. Select the appropriate issue date and topic from the “Please Select an Article Type” drop-down menu.

Important Dates

Submissions Deadline: 1 October 2025
Initial Decision Notification: 1 December 2025
Revision Submission Deadline: 1 February 2025
Final Decision Notification: 1 March 2026
Final Manuscript Due: 15 March 2026
Publication Date: May 2026

Guest Editors 

Tian Wang (Lead Guest Editor)
Beijing Normal University, China

Yao Liu
University of South Florida, USA

Geyong Min
University of Exeter, UK

Nektarios Georgalas
British Telecommunications, UK

Yan Zhang
University of Oslo, Norway

Special Issue on Agentic AI for Next Generation Wireless Networks
截稿日期: 2025-12-31

The 6G era will introduce an unprecedented level of connectivity, requiring ultra-efficient, self-organizing networks that balance high-performance communication, low energy consumption, and sustainability. As next-generation wireless networks become more prevalent, managing interactions among devices, infrastructure, and services in real-time becomes significantly more complex. These networks must process massive streams of data from intelligent edge devices, IoT sensors, distributed computing nodes, and satellite networks, all while ensuring energy efficiency and carbon footprint reduction.

Traditional AI-driven network optimization techniques often rely on predefined rules or centralized control, making them less adaptable to real-world dynamic changes. Agentic AI, in contrast, enables a shift toward fully autonomous, self-learning, and self-optimizing networks by actively interacting with the environment, learning from data, and making real-time, energy-efficient decisions.

By integrating distributed intelligence, multi-agent learning, and energy-aware control algorithms, Agentic AI can transform next-generation wireless networks, ensuring: Dynamic resource allocation based on energy availability, network congestion, and real-time demand, Sustainable and intelligent spectrum management for energy-efficient communication, Proactive self-optimization of networks with minimal human intervention, Efficient orchestration of renewable energy sources (e.g., solar, wind) for powering communication infrastructure, Adaptive, self-healing network capabilities that mitigate failures and enhance reliability. 

Agentic AI also plays a critical role in sustainable network operation by coordinating edge computing, wireless access, and core network functionalities, ensuring optimal energy distribution, latency minimization, and real-time processing. By actively adjusting transmission power, optimizing routing paths, and managing computing loads dynamically, agentic AI significantly enhances the lifespan of energy-constrained devices while ensuring reliable, low-latency, and secure network performance.

We also want to emphasize on the difference between Agentic AI and multi-agent reinforcement learning. Agentic AI emphasizes on autonomy of AI agents, enabling them to make decisions, take decisions, and learn independently to handle complex decision-making and problem-solving tasks. Agentic AI is more about individual agents operating with a high degree of independence and adaptability. On the other hand, multi-agent reinforcement learning (MARL) is a technique that is used to make decisions in a shared environment, which involves multiple autonomous agents. MARL is more concerned with how these multiple agents interact with each other and their environment, aiming to maximize cumulative rewards through the interactions. Agentic AI might or may not use MARL depending on the task at hand. Furthermore, there are many alternatives to MARL, such as decentralized learning, domain knowledge interaction, and mean field game theory, that can be used instead of MARL. There is another term, which many researchers are using, which is Model Context Protocol (MCP). As MARL focuses on the interaction between Agents, MCP enhances the capabilities of Agentic AI by enabling better interoperability, autonomy, and scalability through standardized communication protocols. 

This Special Issue (SI) aims to bring together cutting-edge research and real-world deployments exploring the integration of Agentic AI next-generation wireless networks. This means that Agentic AI techniques can use MARL as well as MCP if the solution is dealing with multiple agents and their interactions for solving hierarchical or unified complex problems. Submissions that focus on large-scale testbeds, real-world validation, and standardization efforts are particularly encouraged to drive sustainable, efficient, and autonomous next-generation networks.

This SI welcomes contributions in areas including, but not limited to:

    Agentic AI for Distributed and Federated Learning in next-generation wireless networks.
    Agentic AI for Multi-Agent Reinforcement Learning and Energy-Efficient Intelligence.
    Agentic AI for Sustainable Resource Allocation and Network Slicing in 6G.
    Agentic AI for Cooperative Sensing and Energy-Aware Network Optimization.
    Agentic AI for Security, Privacy, and Trust in Energy-Conscious Networks.
    Agentic AI for Self-Healing, Self-Optimizing, and Adaptive Network Control.
    Agentic AI for Cross-Layer Optimization in Self-Organizing 6G Networks.
    Agentic AI for Large-Scale Real-World Deployment and Testing.
    Agentic AI for Ethical and Socio-Technical Sustainability in Next-Gen Networks.

Submission Guidelines

Manuscripts should conform to the standard format as indicated in the “Information for Authors” section of the Paper Submission Guidelines.

All manuscripts to be considered for publication must be submitted by the deadline through the magazine’s IEEE Author Portal submission site. Select the appropriate issue date and topic from the “Please Select an Article Type” drop-down menu.

Important Dates

Manuscript Submission Deadline: 31 December 2025
Initial Decision: 15 March 2026
Revised Manuscript Due: 30 April 2026
Final Decision: 30 May 2026
Final Manuscript Due: 15 June 2026
Publication Date: July 2026 

Guest Editors 

Kapal Dev
Munster Technological University, Ireland

Merouane Debbah
Khalifa University, UAE

Guangyi Liu
China Mobile, China

Flavia Delicato
Fluminense Federal University, Brazil

Cedric Westphal
Futurewei Technologies, USA 

Sunder Ali Khowaja
Dublin City University, Ireland

Soumaya Cherkaoui
École Polytechnique de Montréal, Canada