A Bayesian Approach for an Efficient Data Reduction in IoT

Abstract

Todays, Internet of Things (IoT) is starting to occupy a major place in our everyday lives. It has already achieved a huge success in several sectors and continues to bring us a range of new capabilities and services. However, despite the apparent success, one of issues which must be tackle is the big quantity of data produced and transmitted by the objects. Transmitting these big quantity of data not only increases the energy consumption of objects but can also cause network congestion. To meet this issue, a Bayesian Inference Approach (BIA) that can avoid the transmission of highly correlated data is proposed. An hierarchical architecture with smart devices and data centers is adopted. We evaluate our BIA approach using the data obtained from the M3 sensors deployed in the FIT IoT-LAB platform and three distinct scenarios. The obtained results prove the effectiveness of our BIA approach. The number of transmitted data and energy consumption are significantly reduced, and the information accuracy is maintained at a good level.