Ian F. Akyildiz

Zhifeng Hu, Chong Han, Wolfgang Gerstacker et al.

Terahertz (THz) space communications (Tera-SpaceCom) is envisioned as a promising technology to enable various space science and communication applications. Mainly, the realm of Tera-SpaceCom consists of THz sensing for space exploration, data centers in space providing cloud services for space exploration tasks, and a low earth orbit (LEO) mega-constellation relaying these tasks to ground stations (GSs) or data centers via THz links. Moreover, to reduce the computational burden on data centers as well as resource consumption and latency in the relaying process, the LEO mega-constellation provides satellite edge computing (SEC) services to directly compute space exploration tasks without relaying these tasks to data centers. The LEO satellites that receive space exploration tasks offload (i.e., distribute) partial tasks to their neighboring LEO satellites, to further reduce their computational burden. However, efficient joint communication resource allocation and computing task offloading for the Tera-SpaceCom SEC network is an NP-hard mixed-integer nonlinear programming problem (MINLP), due to the discrete nature of space exploration tasks and sub-arrays as well as the continuous nature of transmit power. To tackle this challenge, a graph neural network (GNN)-deep reinforcement learning (DRL)-based joint resource allocation and task offloading (GRANT) algorithm is proposed with the target of long-term resource efficiency (RE). Particularly, GNNs learn relationships among different satellites from their connectivity information. Furthermore, multi-agent and multi-task mechanisms cooperatively train task offloading and resource allocation. Compared with benchmark solutions, GRANT not only achieves the highest RE with relatively low latency, but realizes the fewest trainable parameters and the shortest running time.

Ian F. Akyildiz, Tuğçe Bilen

Sixth-generation (6G) wireless networks are expected to support autonomous, immersive, and mission-critical services that require not only extreme data rates and ultra-low latency but also adaptive reasoning, cross-domain coordination, and objective-driven control across distributed edge-cloud infrastructures. Current AI-enabled network management remains largely data-centric, relying on discriminative models that optimize intermediate quality-of-service metrics without explicitly reasoning about long-term service objectives. This article advocates a transition from bit-centric communication toward knowledge-centric coordination in 6G systems. Semantic communication prioritizes task-relevant information and contextual meaning over raw data delivery, while generative artificial intelligence enables predictive reasoning and adaptive policy synthesis aligned with dynamic service intents. Network optimization is therefore reframed around goal-oriented performance metrics capturing application-level outcomes rather than solely protocol-level indicators. To operationalize this vision, we introduce Kraken, a multi-agent architecture composed of a Knowledge Plane, a distributed Agent Plane, and a semantic-aware Infrastructure Plane. By integrating semantic communication, generative reasoning, and goal-oriented optimization over a shared knowledge substrate, Kraken enables scalable collective intelligence and outlines an evolutionary path from current 5G infrastructures toward knowledge-native 6G systems.

Ian F. Akyildiz, Tuğçe Bilen

Fluid Antenna Systems (FAS) introduce a new degree of freedom for wireless networks by enabling the physical antenna position to adapt dynamically to changing radio conditions. While existing studies primarily emphasize physical-layer gains, their broader implications for network operation remain largely unexplored. Once antennas become reconfigurable entities, antenna positioning naturally becomes part of the network control problem rather than a standalone optimization task. This article presents an AI-native perspective on fluid antenna networks for future 6G systems. Instead of treating antenna repositioning as an isolated operation, we consider a closed-loop control architecture in which antenna adaptation is jointly managed with conventional radio resource management (RRM) functions. Within this framework, real-time network observations are translated into coordinated antenna and resource configuration decisions that respond to user mobility, traffic demand, and evolving interference conditions. To address the complexity of multi-cell environments, we explore a multi-agent reinforcement learning (MARL) approach that enables distributed and adaptive control across base stations. Illustrative results show that intelligent antenna adaptation yields consistent performance gains, particularly at the cell edge, while also reducing inter-cell interference. These findings suggest that the true potential of fluid antenna systems lies not only in reconfigurable hardware, but in intelligent network control architectures that can effectively exploit this additional spatial degree of freedom.

Tuğçe Bilen, Ian F. Akyildiz

Sixth-generation (6G) wireless networks are expected to support immersive and mission-critical applications requiring ultra-reliable communication, sub-second responsiveness, and multi-Gbps data rates. Dense small-cell deployments are a key enabler of these capabilities; however, the large number of candidate cells available to mobile users makes efficient user-cell association increasingly complex. Conventional signal-strength-based or heuristic approaches often lead to load imbalance, increased latency, packet loss, and inefficient utilization of radio resources. To address these challenges, this paper proposes a Knowledge-Defined Networking (KDN) framework for intelligent user association in dense 6G small-cell environments. The proposed architecture integrates the knowledge, control, and data planes to enable adaptive, data-driven decision-making. Small-cell conditions are modeled using queueing-theoretic indicators that capture traffic load and waiting-time dynamics. Based on these indicators, a joint optimization objective reflecting latency and packet loss is formulated and solved via Lagrangian relaxation to obtain globally guided association policies. These optimization outcomes are then used to supervise a lightweight Learning Vector Quantization (LVQ) model, enabling fast and scalable inference at the network edge. Extensive NS-3 simulations under varying mobility, traffic load, packet size, and network density demonstrate that the proposed approach consistently outperforms conventional baselines. The framework reduces average latency by 30-45% in high-mobility and heavy-traffic scenarios and decreases packet loss by more than 35% under congestion. The results confirm that combining optimization-driven knowledge with lightweight learning enables scalable, QoS-aware user association for future dense 6G networks.

Ian F. Akyildiz, Tuğçe Bilen

Future sixth-generation (6G) networks must evolve beyond high-speed data delivery to support intelligent, context-aware services. Emerging applications such as autonomous transportation, immersive extended reality, and large-scale sensing require networks capable of interpreting context, anticipating system dynamics, and coordinating resources according to application objectives rather than relying solely on packet-level metrics. This article introduces Kraken, a knowledge-centric architectural vision for enabling collective intelligence in 6G networks. Kraken integrates three complementary capabilities: semantic communication, which prioritizes the transmission of task-relevant information; generative reasoning, which enables predictive modeling of network and application dynamics; and goal-oriented optimization, which aligns resource allocation with application-level outcomes. These capabilities are organized within a three-plane architecture consisting of an Infrastructure Plane, an Agent Plane, and a Knowledge Plane. Together, these planes enable distributed network entities to perceive context, reason about future states, and coordinate actions through shared semantic representations. The architecture leverages emerging technologies such as O-RAN, network digital twins, and scalable MLOps pipelines, providing a practical evolutionary path from current 5G systems toward knowledge-centric 6G infrastructures. Three representative scenarios illustrate how Kraken improves efficiency and responsiveness in autonomous mobility, immersive XR services, and infrastructure monitoring. The article also outlines key research challenges and discusses the transition from today's data-centric networks toward knowledge-centric collective intelligence in future 6G systems.

Tuğçe Bilen, Ian F. Akyildiz

Semantic communication is emerging as a key paradigm for 6G networks, where the goal is not to perfectly reconstruct bits but to preserve the meaning that matters for a given task. This shift can improve bandwidth efficiency, robustness, and application-level performance. However, most existing studies focus solely on encoder-decoder design and ignore network-wide decision-making. As data traverses multiple hops, semantic relevance may decrease, routing may overlook meaningful information, and semantic distortion can increase under dynamic network conditions. To address these challenges, this paper proposes a management-oriented semantic communication framework built upon Knowledge-Defined Networking (KDN). The framework comprises three core modules: a semantic-reasoning module that computes relevance scores by mapping semantic embeddings onto a knowledge graph that encodes task concepts and contextual relationships; a semantic-aware routing mechanism that forwards data along paths that preserve meaning; and a semantic-distortion controller that adaptively adjusts encoding and routing to preserve semantic fidelity. Our ns-3 results show clear benefits: semantic delivery success improves by 12%, semantic distortion decreases by 22%, re-routing events drop by 44%, and throughput efficiency rises by 14% compared to baseline methods (shortest-path, load-based, and distortion-only routing). These results indicate that meaning-aware and feedback-driven control is essential for reliable and scalable semantic communication in future 6G networks.