Edge6GLearnNet: Reinforcement-Driven Scheduling With Cross-Edge Attention in 6G Remote Learning MEC Environments

Abstract

The integration of high-performance networking, edge intelligence, and educational technology in distributed systems is pivotal for advancing digital learning innovations. Remote education platforms face increasing demands for low latency, personalization, and scalability across diverse and resource-constrained infrastructures. Conventional scheduling and orchestration methods often lack responsiveness, adaptability, and contextual awareness due to rigid optimization models and centralized architectures. To address these limitations, this study introduces a decentralized framework comprising a Spatiotemporal Federated Edge Learning Network (S-FELN) and a Context-Aware Multiscale Orchestration Protocol (CAMP). S-FELN combines federated learning with attention-based mechanisms to dynamically prioritize learner tasks and enable personalized inference under stringent real-time constraints. CAMP employs reinforcement learning to optimize task offloading, semantic fidelity, and synchronization by leveraging contextual factors such as learner engagement, device conditions, and edge congestion. Experimental evaluations in simulated environments reveal substantial improvements in latency reduction, workload distribution, and engagement-aware content delivery. This framework provides a scalable and intelligent edge-native solution, aligning with the objectives of immersive and distributed computing systems central to contemporary computer science research.