Multi-Layer Based Data Aggregation Algorithm for Convergence Platform of IoT and Cloud Computing


M. Mamun-Ibn-Abdullaha, M. Alib, M. Humayun Kabir

Sensor Networks (SN) are deployed in smart domain to sense the environment which is essential to provide the services according to the users need. Hundreds or sometimes thousands of sensors are involved in sensor networks for monitoring the target phenomenon. Large scale of sensory data have to be handle by the sensor network which create several problems such as waste of sensors energy, data redundancy. To overcome these deficiencies one most practice solution is data aggregation which can effectively decrease the massive amount of data generated in SNs by lessening occurrence in the sensing data. The aim of this method is to lessen the massive use of data generated by surrounding nodes, thus saving network energy and providing valuable information for the end user. The effectiveness of any data aggregation technique is largely dependent on topology of the network. Among the various network topologies clustering is preferred as it provides better controllability, scalability and network maintenance phenomenon. In this research, a data aggregation technique is proposed based on Periodic Sensor Network (PSN) which achieved aggregation of data at two layers: the sensor nodes layer and the cluster head layer. In sensor node layer set similarity function is used for checking the redundant data for each sensor node whereas Euclidean distance function is utilized in cluster head layer for discarding the redundancy of data between different sensor nodes. This aggregation technique is implemented in smart home where sensor network is deployed to capture environment related information (temperature, moisture, light, H2 level). Collected information is analyzed using ThinkSpeak cloud platform. For performance evaluation amount of aggregated data, number of pairs of redundant data, energy consumption, data latency, and data accuracy are analyzed and compared with the other state-of-art techniques. The result shows the important improvement of the performance of sensor networks.


Share this article

Get the App