A detection method of mismatched measures in GNSS coordinate time series: Fuzzy logic and IQR (Interquartile Range) based approach

Abstract

This study presents a new approach for outlier detection aiming to enhance the accuracy and
reliability of GNSS data. Unlike traditional approaches, this approach combines fuzzy logic and
the Interquartile Range (IQR) method to improve outlier detection. The fuzzy logic-based
method is employed to flexibly model data characteristics. Individual outlier scores are
calculated for each data point using fuzzy logic, and these scores are then utilized to identify
outliers in the dataset. By determining a threshold value based on the spread of the data
around its center using the IQR method, data points scoring above this threshold can be
considered outliers. The combination of these two methods ensures a more reliable and
accurate outlier detection. When applying the proposed approach to a test signal containing
obvious outlier values, it is observed that the processed time series exhibits a better normal
distribution and improves performance metrics, indicating enhanced signal quality.
Experimental results demonstrate the effectiveness of the proposed approach in effectively
detecting outliers in GNSS coordinate time series. Overall, the proposed approach offers a
promising solution for outlier detection in GNSS data by integrating fuzzy logic and the IQR
method. It provides improved accuracy and reliability, leading to enhanced data analysis and
interpretation in GNSS applications.