A Novel method for time series Prediction in Temperature Readings from IOT devices using LSTM Neural Network

ali noshad¹ ahmadreza khonaksar²

1) Student of Computer Engineering,Iran University of Salman Farsi Kazerun Email:
2) Student of Computer Engineering, Iran Zand Institute of Higher Education Email:

Publication : 2nd International Congress On Engineering, Technology and Innovation(eticong.com/2nd)
Abstract :
Emerging technologies in recent years and major advances in Internet protocols and computing systems have made it easier to communicate between different devices than ever before. This means that as numbers grow and technologies become more mature, the volume of data published will increase. However, sensors in IoT and other applications are often associated with very large data sets, which can cause problems in storage, transmission and security, and privacy issues, Quick and reliable response in unreliable and restricted situations, and Network and storage constraints. In addition, interpreting sensor data in IoT applications, especially in embedded systems, is a challenging problem. The study introduces a new LSTM neural network architecture, as a deep learning approach to prediction and analysis of the time-series data of IOT devices. Temperature readings from IOT devices installed outside and inside of an admin room were used to train and test the model. The proposed model presented in this study, in the time series prediction of temperature reading, achieved a minimum of square error in all the predicted months in comparison with the ARIMA model. Based on the results obtained in this study can be seen that the proposed model (LSTM) it’s very accurate in comparison to the ARIMA model. The results indicate that the deep learning models considering sequence relation are promising in temperature series data prediction.
Keywords : Deep Learning LSTM Neural Network Internet of Things Data analytics Temperature series prediction