Towards Edge Computing Based Distributed Data Analytics Framework in Smart Grids

摘要

Edge computing, as an emerging paradigm empower the network edge devices with intelligence, has become a prominent and promising future for Internet of things. Meanwhile, machine learning method, especially deep learning method has experience tremendous success recently in many application scenario. Recently, deep learning method applied in IoT scenario is also explored in many literatures. However, how to combine edge computing and deep learning method to advance the data analytics in smart grids has not been fully studied. To this end, in this paper, we propose ECNN (Edge-deployed Convolution Neural Network) in edge computing assisted smart grids to greatly enhance the ability in data aggregation and analytics. We also discuss how to train such network in edge computing distributively. Experiments shows the advantage of our paradigm.