Multi-resolution data communication in wireless sensor networks

Frieder Ganz, Payam M. Barnaghi, François Carrez

Abstract: There is an increasing trend in using data collected by sensor devices to enable better understanding of the physical world for humans and support the creation of pervasive environments for a wide range of applications in different domains such as smart cities, and intelligent transportation. However, the deluge of data created and communicated and the low-processing capabilities of the used sensor devices lead to bottlenecks in the processing and interpreting of the data. We introduce a data reduction approach that submits high-granular data in times of high activity in the sensor readings and low-granular data in times of low activity. We consider and discuss different methods to measure activity in the data and modify the symbolic aggregate approximation algorithm that uses a fixed window length to adapt the length according to the data activity for ultimately less data communication between sensor node and sink/gateway. We evaluate our approach over real-world data sets and show that reduction of data size while maintaining the features of the data can be achieved.