Astraea: Deploy AI Services at the Edge in Elegant Ways

Astraea: Deploy AI Services at the Edge in Elegant Ways

연구 분야: Networking

학회: 2020 IEEE International Conference on Edge Computing (EDGE)

Edge computing extends the boundary of the cloud to the network edge, providing low latency and high bandwidth computing paradigm. Computation is trending to be offloaded to the edge to reduce service response time and energy consumption. In this paper, we propose Astraea, a novel AI service deployment platform that could automatically convert a raw AI model to RESTful web services at edge nodes. Experimental results show that a common AI model is able to be deployed at edge nodes in less than two minutes by Astraea, and gains 2Sx to up to 110x performance improvement compared to the traditional scheme.