Details of Research Outputs

With the development of 5G and IoT, all kinds of real life data are collected and recorded by a large number of sensors. It is of great significance to mine and analyze the hidden information in the data to provide predictions for the future. However, due to interference or instability of the collection equipment, the collected sensory data is often incomplete, and this incompleteness, hinders the in-depth analysis of data in the cloud. Therefore, processing around missing values is particularly important. Relying on cloud machine learning methods is not enough to deal with the problem of missing data in the Artificial Intelligence of Things (AIoT) environment, and edge computing provides a promising solution. In this paper, Gated Recurrent Units Filling (GRUF) is applied to the edge nodes. A mobile edge node can not only find the historical information of the current missing data node, but also grasp the data of the nodes adjacent to the missing data node. This ensures that the missing data is restored to the maximum extent at the source. The experimental results show that the missing value filling based on edge computing (MVFEC) not only outperforms other filling methods in quality, but also greatly reduces the bandwidth and energy consumption of AIoT.