Study Of Wireless Sensor Network Localization Algorithm Based On RSSI

Wireless sensor network，which is a novel data acquisition technique，consists ofpower supply unit, sensing unit, information processing unit, communication unit, storageunit and other units of the sensor nodes by self-organized network. The location techniquebased on the RSSI is a kind of cheap location method, its RSSI value can be obtained directlyfrom the sensor nodes, Meanwhile, the hardware requirements is quite low, so it is a kind ofthe most commonly used methods for localization in wireless sensor networks.This paper mainly in the following three stages of improvement and innovation: dataacquisition and preprocessing stage, distance measurement, position calculation.In the dataacquisition and pre-processing stage, as the RSSI value is easily affected by the environment,we can firstly use a mean elimination method in processing data, then using the FCM modelto cluster the processed data, selecting the RSSI value which is the standard cluster centers tothe largest category, seeking this type of geometric mean of all RSSI value as the final result,finally verify the feasibility of the model.In the ranging phase, according to the statisticalmean value model, gaussian correction model, average elimination model in differentenvironment, we choose different environmental degradation of index and the problem ofgaussian correction model based on anchor node referenced that adjacent nodes and theknown node distance is a fixed value. FCM model based on anchor nodes was proposed. It ismore generality that the improved model is referenced of the measured RSSI values and theknown distance between anchor nodes and corrects of the measured RSSI values that is notonly eliminates the path dissipation index, and select anchor nodes nearest center distancefor reference instead of a fixed distance of adjacent anchor nodes.In the position calculationstage, taking into account the actual operation, there will be no solution to the situation. Weneed to analyst the two reasons when no solution appears and we will use the modifiedtrilateration for no solution for the situation because of data problems to add a determinationthreshold value to calculate the feature point overlapping area, in the program.within thethreshold range,we can calculate unknown node coordinates, or choose reference anchorsagain solving feature points and then through improved weighted centroid algorithm,we canget the final positioning results.Finally, by testing the wireless sensor node localization，the results show that the modelcan increase the accuracy to a certain extent and the model basically meets the applicationrequirements.