The Global Positioning System (GPS) can be used for the attitude determination of a vehicle, by estimating baseline vectors between several GPS antennas in fixed positions, in the body frame. To achieve this, pseudorange and more precise carrier phase measurements (with millimeter precision) are used. However, these phase measurements have an unknown number of integer cycles that need to be determined, which are called phase ambiguities. From all the known methods, the LAMBDA method is recognized as the most efficient for the phase ambiguity determination. This method has been applied with success, mainly with dual-frequency measurements (in the L1 and L2 bands). On the other hand, when only single-frequency measurements (L1) are used, the LAMBDA method does not give a stable solution for the ambiguities, and so it does not allow a high level of precision in the estimation of the baseline vectors. In this case, the method known as Ambiguity Filter is employed, after the LAMBDA, to stabilize the obtained solution. The leverage of the Ambiguity Filter, relatively to alternative of using only the LAMBDA method, comes from using a prior knowledge of the baseline length in the ambiguity resolution problem, which the LAMBDA does not. Preliminary tests with the implementation of the Ambiguity Filter have shown that its performance is very sensitive to the precision with which the baseline length is known. This was the main motivation for the work here presented.
This paper presents a sensitivity analysis of the Ambiguity Filter to the precision of the prior knowledge of the baseline length, using only GPS L1 measurements. The conclusions are based on field tests, conducted at the university campus, with two GPS receivers in a static, fixed length, baseline configuration. The presented results show the performance of heading and pitch estimation, using the Ambiguity Filter, for different levels of uncertainty in the prior knowledge of the baseline length.