carrier shutdown orchestration

This study addresses the limitations of existing 5G NR energy-saving mechanisms, which are typically enabled in isolation and struggle to balance energy efficiency with network performance. To overcome this, the authors propose a holistic energy management framework that decouples hardware capabilities from higher-layer energy-saving features and introduces a logical “feature coordinator” to jointly orchestrate signaling, dynamic resource allocation, and advanced sleep modes. The framework establishes a compact taxonomy of energy-saving features, enabling, for the first time, coordinated optimization across multiple mechanisms. Evaluations on a 3GPP-aligned simulation platform using production-grade parameters demonstrate that the proposed approach significantly reduces gNodeB energy consumption compared to non-coordinated schemes, while incurring only negligible throughput degradation.

To address the joint optimization of energy consumption and performance in O-RAN, this paper proposes a traffic-aware dynamic cell sleep framework based on Proximal Policy Optimization (PPO), deployed within the O-RAN near-real-time RAN Intelligent Controller (RIC) architecture. It is the first work to adapt PPO to the RIC context, jointly modeling throughput degradation constraints, interference thresholds, and PRB load balancing for multi-objective adaptive decision-making. Evaluated on the TeraVM Viavi simulation platform and real-world RIC test data, the approach achieves a 32.7% improvement in network energy efficiency and an 18.4% gain in downlink throughput over baseline policies, while maintaining QoS guarantees. The core contributions lie in (i) the customization of reinforcement learning algorithms for O-RAN’s real-time control loop, and (ii) a novel joint optimization mechanism under multi-dimensional resource constraints.

This work addresses the lack of scalable mechanisms in current mobile networks to effectively translate sustainability goals into energy-efficient strategies. The authors propose a tool-augmented, lightweight large language model (LLM) agent embedded within the network control loop that interprets natural-language sustainability intents and converts them into telemetry-driven, energy-aware traffic scheduling commands to orchestrate User Plane Function (UPF) operations in an environmentally conscious manner. This approach represents the first integration of natural-language intent-driven control with edge networking, enabling non-zero migration under the resource constraints of Multi-access Edge Computing (MEC). Experimental results demonstrate a strong coupling between control latency and energy consumption, confirming that the lightweight LLM can accurately execute policies with low overhead while maintaining effective migration capabilities even under stressed MEC conditions.

To address the challenges of resource heterogeneity, high mobility, wide-area distribution, and operation beyond operator-controlled domains in 6G ultra-edge scenarios, this paper proposes an AI-driven end-to-end resource orchestration architecture. The architecture introduces a novel orchestration paradigm integrating AI-based forecasting, multi-source high-concurrency telemetry monitoring, and closed-loop adaptive execution—transcending traditional centralized control boundaries. It incorporates a lightweight real-time decision engine, infrastructure state prediction models, and adaptive actuators to ensure service resilience and enable proactive, prediction-driven scheduling. Under extreme conditions—such as user mobility up to 500 km/h and node offline durations on the order of minutes—the architecture reduces service interruption rate by 76% and accelerates resource scheduling response time by 4.2×, significantly enhancing both reliability and real-time performance of ultra-edge services.

This paper addresses direct, indirect, and implicit conflicts among xApps in O-RAN architectures arising from structural decoupling. It proposes the first xApp-level three-category conflict classification framework; develops a conflict graph-based KPI impact modeling method that integrates SLA/QoS thresholds for precise conflict detection; and designs a cooperative mitigation mechanism tailored to mobility robustness optimization (MRO) and energy-saving (ES) scenarios. Evaluated in a simulated environment co-deploying MRO and ES xApps, the approach significantly reduces KPI anomaly rates while enhancing both system stability and energy-efficiency synergy. Key contributions include: (i) the first formal, multidimensional definition of xApp-level conflicts; (ii) establishment of a conflict–KPI mapping model grounded in operational semantics; and (iii) realization of policy-driven, cross-xApp collaborative conflict mitigation—enabling adaptive, SLA-aware resource orchestration in disaggregated RAN environments.

This study addresses the challenge of high energy consumption and the difficulty of implementing energy-saving base station shutdowns in dense urban cellular networks. To overcome this, the authors propose leveraging High-Altitude Platform Stations (HAPS) to form a “Hypercell” that replaces the coverage of multiple terrestrial macrocells, thereby enabling coordinated shutdown of both coverage-layer and capacity-layer base stations for network-wide energy savings. Innovatively repositioning HAPS from its conventional role in non-terrestrial networks (NTN) as a mere coverage extender to an enabler of joint coverage-and-capacity shutdown, the work introduces two HAPS–Hypercell pairing architectures to support distributed carrier shutdown mechanisms. Evaluations based on 3GPP-compliant modeling and realistic channel simulations demonstrate substantial reductions in network power consumption while also revealing limitations of direct HAPS integration, offering critical insights for future green communication strategies.

This study addresses the challenge of base station energy consumption during low-traffic periods in 5G networks by proposing a machine learning–based energy-saving strategy that explicitly incorporates operator-defined quality-of-service (QoS) constraints. Leveraging real-world 5G data within a supervised learning framework, the method innovatively embeds QoS policies—such as throughput guarantees and maximum allowable service interruption—into the model optimization process through a policy-guided class reweighting mechanism during training. This enables a controllable trade-off between energy efficiency and service compliance. Experimental results demonstrate that the proposed approach significantly reduces base station energy consumption in operational networks while strictly adhering to the stipulated QoS requirements.

Current 5G/6G networks lack an end-to-end collaborative management mechanism spanning the full lifecycle of network slices. This work proposes METIS, a declarative orchestrator that introduces, for the first time, an application-oriented data model and a cascaded coordination architecture to enable automated orchestration across Day-0, Day-1, and Day-2 phases. METIS generates 3GPP-compliant configurations from application semantics, integrates cross-domain control between O-RAN and 3GPP, and uncovers a structural asymmetry wherein uplink SLAs critically depend on radio-side execution. Experimental results on a cloud-native testbed demonstrate that METIS supports slice creation in 22.4 seconds, updates in 5.1 seconds, and recovery in under 19 seconds, while guaranteeing 100% SLA compliance under concurrent overload and scaling to 63 instances with less than 0.03 CPU cores.

This work addresses the growing complexity of mobile networks by proposing an intelligent, dynamic energy-saving control strategy grounded in world models. The approach learns from historical data to predict how actions influence future network states and incorporates uncertainty estimation to enable robust decision-making. Notably, it allows for dynamic adjustment of optimization objectives without requiring retraining and autonomously generates energy-efficient configurations in closed-loop control that satisfy quality-of-service constraints. Experimental results demonstrate that the method consistently outperforms conventional approaches and reinforcement learning baselines—both in simulation and on real-world network data—achieving a superior trade-off between energy efficiency and service quality.

This study addresses the limitations of conventional optimization methods in the joint communication and control co-design for B6G networks, particularly regarding modular representation, requirements traceability, and design space analysis. To overcome these challenges, this work proposes a composition-driven methodology grounded in formal co-design theory. By introducing a compositional perspective, the proposed approach circumvents the bottlenecks inherent in joint optimization, thereby enabling the modular modeling of complex interacting subsystems and systematic exploration of the design space. The effectiveness of this methodology is validated through a wireless-assisted robotic control case study. Furthermore, this paper elucidates its complementary relationship with optimization-driven approaches. Ultimately, this research establishes a novel paradigm for cross-domain co-design within B6G scenarios, offering a rigorous framework to facilitate scalable and verifiable system integration.