Accuracy Improvement of Low-Cost GPS Receivers Using Decision Tree Algorithm

Abstract

Global Positioning System (GPS) is a breakthrough technology that allows users to obtain the location of a certain target anywhere in the world. A low-cost commercial GPS receiver can reach accuracy within 10 meters, which is not suitable for automated drone applications. This constraint raises the issue that low-cost GPS receivers cannot provide a precise location for a drone positioning system. To address this problem, the authors proposed the implementation of two low-cost single-frequency GPS receivers using a decision tree algorithm. Throughout the experiment, it was obtained that the lowest RMSE is 2.14 meters and 1.85 meters for the algorithm employed in case 1 and case 2 respectively with the average computational time of 100 milliseconds. Based on these results, it can be concluded that the lowest distance between the receivers has a significant role in improving the positioning accuracy. Additionally, employing the Dilution of Precision (DOP) parameter as the predictor enhanced the accuracy further. This result pointed out that the method proposed in this research could significantly improve the accuracy of low-cost GPS receivers for automated drone applications.