A neural network based optimization for wireless sensor node position estimation in industrial environments

The sensor node position estimation is essential in wireless sensor networks. Among many localization schemes, the position estimations based on Received Signal Strength Indicator (RSSI) are mostly used in various systems and applications. However, RSSI data are highly affected from multipath propagation caused by the reflections from walls or objects. These reasons conduct the improper phenomena to radio signals. The significant variation of RSSI influences to the position estimation error especially in industrial environments. In this paper, we present a sensor node position estimation method in industrial environments and its optimization to reduce the error from multipath propagation by using neural networks. An experiment was performed in an electrical machine laboratory to evaluate the designed system in the real environment. The experimental results show that the average position error was reduced to 0.5 m.