Prediction of UWB Positioning Coordinates with or Without Interference Based on SVM

Abstract

Through UWB positioning technology, accurate positioning in the house can be achieved. However, the obstruction of indoor houses leads to the decrease of positioning accuracy, and UWB can’t judge whether there is signal interference when collecting data. Therefore, judging whether there is signal interference is the key and difficult point of UWB accurate positioning. In order to judge which data are collected under the interference of signals and which data are collected without interference of signals, The fast and effective programming tool by Python to deal with pretreatment problems of data set, and by way of the effective algorithm based on support vector machine, establishes a binary classification prediction model to judge whether the collected data have interference or not. Ultimately, 486 samples data sets were collected for training by the algorithm, and 162 samples data sets were used to test whether there was any interference to the signal. Six prediction methods, such as k-nearest neighbor algorithm, light gradient boosting machine, decision tree, naive bayes, logistic regression, and neural network, were used for comparison with support vector machine algorithm, and the results were visualized by Python and other tools. Finally, the F1-score, recall, accuracy, and AUC of the prediction outcomes of support vector machine model were the highest, and the model had a high classification. Through the prediction of the data set with or without interference, we can further use the data set to select an appropriate algorithm to predict the coordinate of target points.