Research And Implementation Of Localization Algorithm For Mobile Nodes In WSN

To obtain the coordinates of sensor nodes is a basic problem needing to be solved in wireless sensor networks applications. For most applications, sensing data are meaningless without knowing the location of nodes. The location of mobile sensor nodes is variable momently, which results in difficulties in the implementation of localization and ensuring localization accuracy. Therefore, the research of effective localization algorithm in wireless sensor networks to ensure the localization accuracy of mobile nodes has important theoretical significance and application value.Currently, most localization schemes in wireless sensor networks are about the localization of stationary nodes, that is to say the nodes are at rest. In this case, RSSI-based localization schemes fundamentally reduce the errors caused by instability of RSSI by collecting more RSSI samples and filtering them. However, in many application environments, the localization system can’t make sure that the unknowed nodes are in static state or movement state, such as sensor nodes fixed in the miners working in the underground and others fixed in the wild animals for the research of their living habits, etc. In this case, the mobility of sensor nodes makes it difficult to collect multiple effective RSSI samples in the same place. So it is no longer effective to reduce the localization errors by repeating the sampling and localization accuracy can hardly be guaranteed. Therefore, this paper improved the traditional localization schemes from three hands:in the distance, the paper proposed improved self-adaptive log-normal shadow model to improve the adaptability of distance model and thus improve the distance accuracy; in localization algorithm, the paper proposed fuzzy centroid localization algorithm which could be adjusted with precision, was easy to implement, and had high precision; in localization errors correction, the paper adopted Kalman filter, using Markov process theory to establish equations of states to predict the coordinates of the unknown nodes at the next time according to the states of them at some point and combining with the measured coordinates to estimate the optimal value of coordinates to achieve the objective of reducing localization errors.The paper has conducted experiments for the theories put forward from three hands. Experimental results show that compared to traditional localization schemes,the work in the paper has effectively improved the RSSI-based localization accuracy in wireless sensor networks with less cost of computational complexity.