An Improved Approach to Network Ambiguity Validation by Applying Outlier Detection to the Baseline Measurement Errors

Abstract

As a prerequisite of network differential global positioning system applications, the network ambiguity must be determined. Ambiguity resolution and validation are important aspects of this process. However, validation theory is still under investigation. This paper presents an improved network ambiguity validation method that incorporates additional knowledge measured from the network. This process involves the detection of outliers of the baseline measurement errors. By breaking the spatial correlation, incorrectly fixed ambiguities cause the corresponding baseline measurement errors to appear as outliers, which may be discovered and identified with the proposed outlier detection algorithm and outlier identification algorithm, respectively. These detection and identification procedures are iteratively performed until all of the wrong baseline ambiguities are corrected. Because the validation procedure is unconnected to the initial integer ambiguity estimation process, any available ambiguity resolution method may be used to obtain the initial integers, without algorithm correction. When the network ambiguity combinations do not pass the validation algorithm, the method uses a direct estimation algorithm to obtain the correct ambiguity. By using a direct estimation algorithm rather than a search process, this new method consumes less computational time than conventional methods. This study compares the performance of this new method with those of the conventional F-ratio and W-ratio test validation algorithms by using Monte Carlo simulation techniques. Results from a field experiment conducted on data from the United States continuously operating reference stations (US-CORS) reveal that this validation algorithm accelerates the convergence process of ambiguity determination.