The accuracy of fingerprint databases is very important to position fingerprint locations. However, due to the measurement of the received signal strength cost and artificial constraints, the measured RSSI is limited, resulting in insufficient accuracy of fingerprint databases. Therefore, we adopt the spatial interpolation algorithm to obtain a more accurate location fingerprint database and innovatively propose an improved universal Kriging interpolation method, introducing variable parameters, overcoming the shortcomings of large measurement errors in the variation function, and generating an accurate location fingerprint database. The simulation results show that compared with the universal Kriging interpolation method and the inverse distance weighting method, the standard error of prediction is increased by 76.33% and 96.50%, respectively.