IOT device code translators using LSTM networks

Abstract

Currently, the Internet of Things (IOT) platform used by large buildings to manage the indoor climate uses different controllers and sensors from multiple manufacturers. Communication between these devices requires a human in the loop to translate each devices data to to be compatible with a common integration engine and storage historian. The subject matter expert needs to decipher the non-standard naming convention used for each device and manually translate thousands of codes each time a new device has to be integrated. To aid the human translator, we propose a technique to implement a smart translator using Deep Neural Networks (DNN) by automatically assigning any registers with recognized data patterns to standardized labels.