Node localization is one of the key technologies in applications of wireless sensor networks（WSNs）

and precision of ranging determines precision of localization to a great extent.In order to improve precision of localization

improvement of ranging precision is studied.According to RSSI and corresponding distance value by experiment

first the experimental data is filtered

predictive model of improved Elman neural network is established

then using characteristics of high stability and high precision of the improved Elman neural network to train the sample.Eventually predictive distance value is obtained

and the precision of distance is within 1m.