Deep Learning Based Fog Multi-Service Platform for Edge Intelligence

This thesis deals with how to design and implement the deep learning-based fog multi-service delivery platform, targeting to provide a wide range of intelligent edge services while achieving the desirable quality of service at the same time. Different chapters of this thesis can be considered as incremental advances of a fog service platform, eventually leading to the complete form of edge intelligence under fog context in terms of architecture support, service provisioning, edge energy consumption control/reduction, in-fog learning design, fog communication and fog collaboration for model training, as well as intelligent service inference. While the research in edge intelligence field starts to draw more and more attention from both academia and industry in recent years, this thesis makes a sound contribution to the multi-disciplinary research, including Internet of Things, Artificial Intelligence, Fog/Edge Computing, and Service Computing.