Analysis Of Node Localization Technology In Wireless Sensor Network

In the context of wireless sensor networks, a node’s location must be known for its data to be meaningful in many cases. Therefore, one of the most basic functions of wireless sensor network is to determine the location of the incident or to get position of message. It plays a critical role in the effectiveness of the sensor network applications. Received signal strength (RSS)-based localization has been widely used because of low complexity and easy deployment. However, when there are random obstructions in the network, the location accuracy would be highly affected. To solve the problem, this paper proposed a new RSS-based node localization system.Firstly, a system model is proposed to accommodate the environment. Each link of data experience three types of fading based on experimental results. There are no obstructions in the network, the link is totally obstructed and the obstructions appear in the network without obstructing the link. RSS values are modeled as a mixture of Gaussians (MoG) and its parameters are updated by background learning. The core of the model is that the value which belongs to different Gaussian models can be distinguished. In this way, then the values belong to the same set could be put together. Background learning algorithm is proposed to update the parameters of the model. The environmental interferences caused by random obstructions ate removed using this model.Secondly, a modified localization algorithm is carried out to localize the position of nodes. The modified RSS data are translated into distance information using path loss model and LS estimation is used to localize nodes. To make a tradeoff between localization accuracy and energy loss, node selection, path loss estimation and unknown nodes’initialization are implemented in the localization algorithm. Through these modifications, the algorithm can adapt to the environment more effectively.On the basis of theoretical studies, experimental platform of wireless sensor networks is built using CC2530chip and embedded hardware development platform. Develop MATLAB receiver program to complete data collection and analysis. Simulate the proposed system with MATLAB and collect experimental data under different environments with the self-built platform. Experimental results show that the proposed sensor node localization system can distinguish interfered RSS from received RSS. The node localization accuracy can be improved using the proposed system. In the context of our experiment, the mean error square can reach0.1-0.3m2.