Energy efficient data prediction model for the sensor cloud environment

• Published in: 2017 International Conference on IoT and Application (ICIOT) pp. 1 - 3
• Main Authors: Das, Kalyan, Das, Satyabrata, Mishra, Ananya, Mohapatra, Aurobindo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2017
• Subjects: Algorithm design and analysis; Cloud computing; Data models; Data prediction; Logistics; Meteorology; Prediction algorithms; Predictive models; Sensor Cloud; Wireless Sensor Network
• DOI: 10.1109/ICIOTA.2017.8073619

The sensors are used for many applications in the recent time. The sensors generally connect with each other wirelessly to form a Wireless Sensor Network (WSN). Cloud computing is an emerging technology where the end users pay and access the services without worried about the infrastructure. Sensor cloud combines sensor network with the cloud computing in which the end users can access to the sensor network through the cloud computing. Sensor cloud should be energy efficient as the battery life of the sensor is finite and huge amount of energy is consumed in the cloud computing environment to provide services to the end users. The users request to access the sensor through the cloud system redirects every time to the sensor network, which causes more transmission in the sensor network as a result more energy is consumed. In this paper, we have compared mainly the accuracy and time consumed by various prediction schemes using some activation functions. From our analysis, we found that the Rprop-algorithm using logistic activation function is suitable as it provides nearly 97.2 percentage accuracy within an admissible delay of 13 seconds. Our proposed sensor cloud model integrates Rprop-prediction scheme using the logistic activation function in cloud system which predicts future sensor data, such that users request are replied at cloud level which saves energy as number of transmissions are reduced in the sensor network.