In high precision positioning systems based on GNSS, ambiguity resolution is an important procedure. Correct ambiguity leads to positioning results which have high precision between millimeters and centimeters. However, when ambiguity is determined incorrectly, ensuring accuracy and precision of the positioning result is impossible. Ambiguity validation tests are required to obtain correct ambiguity when ambiguity resolution is performed based on ILS (integer least squares), which shows the best performance in point of theory and experiment when compared with other methods such as IR (inter rounding) and IB (integer bootstrapping). Comparison between candidates of the validation test is needed to judge ambiguity correctly, because ILS searches for candidates of integer ambiguity, unlike other methods which calculate only one integer ambiguity. Traditionally, ambiguity validation testing has been studied in statistical aspects. Therefore, the stochastic model of observations affects the ambiguity validation testing. This paper presents an experimental analysis of the relationship between ambiguity validation tests and the stochastic model of observations. R-ratio, F-ratio and W-ratio were adopted for analysis. Some stochastic models were also described to be exploited for analysis.

Keywords: ambiguity resolution, ambiguity validation test, stochastic model, variance-covariance matrix