AI Assisted Cloud Resources Allocation for Edge Computing

The growing adoption of edge-cloud computing has underscored the need for intelligent resource allocation frameworks capable of managing dynamic and heterogeneous workloads. This paper proposes a novel hybrid AI-driven framework that integrates convolutional neural networks (CNNs) and transformers to address the complexities of spatial and temporal data in edge-cloud environments. The framework leverages spatial-temporal attention mechanisms to capture critical workload patterns, cross-attention fusion to dynamically integrate spatial and temporal features, and adaptive filters to optimize resource allocation. Extensive experiments on a realworld edge computing dataset demonstrate that the proposed framework outperforms state-of-the-art methods in terms of workload prediction accuracy, energy efficiency, and latency reduction. Specifically, the model achieves a 20.7% reduction in latency and an 18% improvement in energy efficiency compared to the best-performing baseline. Furthermore, scalability analysis highlights the robustness of the framework under increasing workloads, making it suitable for real-world deployment. This study contributes to the advancement of intelligent and adaptive resource management in edge-cloud computing systems, addressing key challenges such as heterogeneity, scalability, and real-time performance optimization.