A Multi-Perspective Intelligent Resource Orchestration Framework for Energy-Aware Cloud-Edge Computing

Abstract

Energy-aware resource orchestration has become a critical requirement in modern cloud-edge computing environments, where heterogeneous resources, dynamic workloads, network variability, and increasing energy consumption create major challenges for sustainable service delivery. Conventional orchestration approaches are commonly compute-centric and reactive, often prioritizing CPU availability, memory capacity, or basic load balancing while treating network conditions, data locality, energy behavior, and Quality of Service (QoS) constraints as secondary factors. This leads to inefficient task placement, excessive data movement, increased latency, higher operational energy consumption, and reduced adaptability in distributed cloud-edge infrastructures. To address these limitations, this paper proposes a Multi-Perspective Intelligent Resource Orchestration Framework for energy-aware and performance-optimized cloud-edge computing. The proposed framework introduces a central Multi-Perspective Orchestration Intelligence Engine that jointly models compute, network, and data-locality perspectives for context-aware decision-making. The engine integrates workload prediction, real-time telemetry analysis, Energy-QoS optimization, and adaptive decision policies to support task placement, scheduling, auto-scaling, migration, and workload consolidation. Unlike conventional schedulers, the proposed approach formulates orchestration as a multi-objective optimization problem that simultaneously minimizes energy consumption, latency, data transfer cost, and QoS violations while improving resource utilization. The proposed framework is designed for heterogeneous cloud-edge continuum environments consisting of cloud data centers, fog or micro data centers, edge gateways, IoT devices, and application workloads. Experimental analysis demonstrates that the proposed model achieves improved orchestration performance compared with default Kubernetes scheduling, energy-aware scheduling, and ML-based scheduling approaches. The proposed model records 31.20% energy saving, 37.52% latency reduction, 97.60% SLA satisfaction, and 88.90% resource utilization, indicating its ability to balance sustainability and performance. Ablation analysis further confirms that the integration of compute awareness, network awareness, data locality, workload prediction, and Energy-QoS optimization progressively improves overall orchestration efficiency. These results establish the proposed framework as a scalable, adaptive, and energy-efficient solution for next-generation cloud-edge resource management.