Ilmu Komputer & AI editorial
Open AccessOA2026
AI/ML for D2D Communication in 5G/6G Networks and Beyond: Concept, Applications, Challenges, and Future Directions
A structured survey of artificial intelligence and machine learning approaches for device-to-device communication, spanning classical, advanced, and emerging paradigms toward 6G autonomy.
Iacovos I. Ioannou; C. Christophorou; V. Vassiliou· IEEE Open Journal of the Communications Society· 2026· DOI 10.1109/OJCOMS.2026.3694216
The core problem
Device-to-Device (D2D) communication enables direct links between proximate devices, bypassing base stations to enhance spectral efficiency, energy efficiency, system capacity, and ultra-low latency. These benefits are critical for emerging applications such as Vehicle-to-Everything (V2X), Industrial IoT, and extended reality. However, D2D communication faces persistent technical challenges: device discovery, mode selection, interference management, power control, security, radio resource allocation, cell densification and offloading, quality of service provisioning, millimeter-wave integration, and seamless handover. The dynamic and complex nature of these challenges has driven extensive research into Artificial Intelligence (AI) and Machine Learning (ML) solutions. This survey provides a structured, multi-dimensional review of AI/ML approaches for D2D communication, organizing the literature into three broad generations: classical techniques (e.g., Fuzzy Logic, Q-Learning, Neural Networks, GA/PSO/ACO), advanced paradigms (e.g., DRL, GNNs, FL, MARL), and emerging technologies (e.g., LLMs, Generative AI, Semantic Communication, RIS). Rather than cataloguing methods in isolation, th
Innovation
The survey reports indicative performance ranges for AI/ML techniques across D2D challenges, synthesized from heterogeneous studies. Key findings include:
- **Classical techniques** (e.g., Fuzzy Logic, Q-Learning) offer low complexity and fast convergence but struggle with high-dimensional state spaces and dynamic environments.
- **Advanced paradigms** (e.g., DRL, GNNs, FL, MARL) demonstrate improved adaptability and scalability, with DRL achieving significant gains in resource allocation and interference management. However, they incur higher computational costs and require extensive training.
- **Emerging technologies** (e.g., LLMs, Generative AI, Semantic Communication, RIS) show promise for autonomous, context-aware D2D control, but are in early stages of research maturity.
Cross-cutting analysis reveals trade-offs: while advanced and emerging techniques enhance performance and autonomy, they often increase complexity and energy consumption. Convergence and scalability trends indicate that MARL and FL are more suitable for distributed, dense deployments, whereas centralized DRL may face scalability bottlenecks. Autonomy readiness is highest for emerging techniques, but deploy
Device-to-Device (D2D) communication enables direct links between proximate devices, bypassing base stations to enhance spectral efficiency, energy efficiency, system capacity, and ultra-low latency. These benefits are critical for emerging applications such as Vehicle-to-Everything (V2X), Industrial IoT, and extended reality. However, D2D communication faces persistent technical challenges: device discovery, mode selection, interference management, power control, security, radio resource allocation, cell densification and offloading, quality of service provisioning, millimeter-wave integration, and seamless handover. The dynamic and complex nature of these challenges has driven extensive research into Artificial Intelligence (AI) and Machine Learning (ML) solutions. This survey provides a structured, multi-dimensional review of AI/ML approaches for D2D communication, organizing the literature into three broad generations: classical techniques (e.g., Fuzzy Logic, Q-Learning, Neural Networks, GA/PSO/ACO), advanced paradigms (e.g., DRL, GNNs, FL, MARL), and emerging technologies (e.g., LLMs, Generative AI, Semantic Communication, RIS). Rather than cataloguing methods in isolation, the survey synthesizes the literature through cross-cutting analytical perspectives including challenge coverage, performance and complexity trade-offs, convergence and scalability trends, autonomy readiness, and research maturity across more than 170 references, including recent works from 2023 to 2025 relevant to 6G-oriented D2D systems. Performance figures are reported as indicative ranges synthesized from heterogeneous studies and are not used for direct cross-paper ranking without normalization to common experimental conditions. The survey also discusses the DAI/BDIx framework as one possible path toward autonomous distributed D2D control and notes that initial ADROIT6G testbed validation has now been reported for BDIx-enabled control blocks in a beyond-5G stand-alone environment, while recent cluster-based cell-free D2D/6G results further broaden the supporting empirical base. Finally, twelve open issues are identified and a roadmap toward intelligent, distributed, and deployment-aware D2D systems in 6G networks is outlined.
The survey adopts a structured, multi-dimensional review methodology. The literature is organized into three broad generations of AI/ML techniques for D2D communication:
Why it matters
The survey provides a critical analysis of AI/ML for D2D communication, emphasizing that no single technique addresses all challenges. Classical methods remain relevant for low-complexity scenarios, while advanced paradigms offer a balance between performance and scalability. Emerging technologies promise full autonomy but require further maturation. The DAI/BDIx framework is presented as a viable architecture for autonomous distributed D2D control, integrating belief-desire-intention (BDI) agents with distributed AI. The ADROIT6G testbed validation of BDIx-enabled control blocks in a beyond-5G stand-alone environment marks a significant milestone, demonstrating practical feasibility. However, challenges persist: performance comparisons across studies are hindered by heterogeneous experimental conditions, necessitating normalization. The survey outlines a roadmap toward intelligent, distributed, and deployment-aware D2D systems in 6G, calling for standardized benchmarks, energy-efficient algorithms, and robust security mechanisms. Future directions include integrating semantic communication and RIS with AI/ML for context-aware, ultra-reliable D2D links, and leveraging LLMs for zero-shot decision-making in dynamic environments. The twelve open issues provide a research agenda for the community.
Who should read this
CS practitioners and researchers
Opening member content…