A multi-sensor circular particle filter applied to the fusion of the GPS-L2C channels

Angular data are involved in many applications. This paper investigates more precisely the phase GNSS signal processing, particularly when measurement fusion on several channels is possible. The classical approaches use local linearizations to filter and fuse measurements with optimal methods defined in the linear domain. These approaches do not fit when the angle reaches the bounds of its periodicity definition interval. Moreover, an ambiguity on the angle to estimate can appear in case of noisy measurements. This paper proposes tools defined in the circular domain in terms of Bayesian inference to solve these problems and improve the processing of angular data. The proposed tools are assessed on synthetic and real data in an original Phase Open Loop tracking architecture.