TURKISH JOURNAL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCES, cilt.28, sa.5, ss.2702-2717, 2020 (SCI-Expanded)
In this paper, we first design a sensor with low power and low cost components, namely an MSP430G2553 microcontroller (MCU), nRF24L01+ communication unit, and a light-dependent resistor (LDR). Then, connecting these custom-designed sensors with a central node, a fusion center (FC), we form a wireless sensor network (WSN) in which sensors measure light intensities and send their measurements to the FC for final inference. Since the default addressing structure of nRF24L01+ can simultaneously connect up to 6 devices, in order to have more connections, we define a new addressing scheme for the transmissions between sensors and the FC. Finally, as a test case, we consider the task of the WSN in field estimation. The FC learns the spatial dependence between sensor measurements called a variogram. Then, using the appropriate variogram model, we perform ordinary kriging (OK), where the light intensity at an unobserved location is estimated as a weighted sum of received LDR measurements. Our test results show that the estimated light measurement using OK at a specific location becomes quite close to the actual LDR measurement at that location under the suitable parametric variogram models.