Real-time Approach for Decision Making in IoT-based Applications

Nowadays, the IoT applications benefit widely many sectors including healthcare, environment, military, surveillance, etc. While the potential benefits of IoT are real and significant, two major challenges remain in front of fully realizing this potential: limited sensor energy and decision making in real-time applications. To overcome these problems, data reduction techniques over data routed to the sink should be used in such a way that they do not discard useful information. In this paper, we propose a new energy efficient and real-time based algorithm to improve the decision making in IoT. At first data reduction is applied at the sensor nodes to reduce their raw data based on a predefined scoring system. Then, a second data reduction phase is applied at intermediate nodes, called grid leaders. It uses K-means as a clustering algorithm in order to eliminate data redundancy collected by the neighbor nodes. Finally, decision is taken at the sink level based on a scoring risk system and a risk-decision table. The evaluation of our technique is made based on a simulation from data collected on sensors at Intel Berkeley research lab. The obtained results show the relevance of our technique, in terms of data reduction and energy consumption.