EdgeLearn-NDN: An Edge-Cloud Cooperative Continuous Learning Framework for NDN

초록

The rapid evolution of smart environments such as hospitals and surveillance systems requires low-latency, adaptive intelligence across distributed edge devices. Named Data Networking (NDN) provides a data-centric architecture that enables efficient content retrieval and semantic interoperability. This poster presents EdgeLearn-NDN, an edge-cloud collaborative framework for continuous learning in NDN-based systems. A global model is trained in the cloud using diverse semantic data and deployed to edge cameras for real-time inference and event naming. Edge nodes transmit lightweight semantic feedback to the cloud, where collective updates refine the master model. The updated model is then redistributed via NDN, forming a continuous learning loop. The proposed design enhances semantic consistency, detection accuracy, and adaptability while minimizing bandwidth consumption in dynamic smart environments. © 2025 Copyright held by the owner/author(s).

키워드