Application-driven critical values for GNSS ambiguity acceptance testing

Integer ambiguity estimation and validation are crucial steps when solving the carrierphase based GNSS model. For the validation, different ambiguity acceptance tests have been proposed. For those tests often fixed critical values are used, with the important disadvantage that the performance of the tests varies a lot depending on measurement set-up and circumstances. Therefore it is better to use model-driven critical values such that it is guaranteed that the failure rate will not exceed a user-defined threshold. This contribution will study the model-dependency of the critical values for two well known acceptance tests, the ratio test and difference test, and then specifically for a given application. This means that mainly the satellite-receiver geometry and number of epochs will be variable. It will be shown that critical values do exhibit a strong dependence on these factors, and it will not be possible to simply use a fixed (i.e., constant) application-driven critical value.