A Nonlinear Filtering Approach for Robust Multi-GNSS RTK Positioning in Presence of Multipath and Ionospheric Delays

In this paper, we develop a new approach of precise positioning using three carrier phase multi-Global Navigation Satellite System (GNSS) measurements in presence of multipath and ionospheric delays. We propose a new nonlinear filter to estimate the user position as well as all the unknown parameters including the integer ambiguities and the ionospheric errors. First, we use a kernel representation of the conditional density and apply a local linearization which yields a Kalman-like correction enhancing the particle filter correction. This new particle Kalman filter approach, is designed to be efficient for the non-Gaussian state and nonlinear measurements model, reduces the number of needed particles, and reduces the risk of divergence. The proposed procedure for multifrequency ambiguity resolution is based on four steps: 1) at each epoch, we compute the float solution adaptively to the dynamic environment by minimizing the noise level and estimating the ionospheric errors using the proposed robust Bayesian particle Kalman filter (RobPKF); 2) we introduce a new carrier phase multipath indicator and use it to derive a related constraint to reject integers candidates that are affected by multipath errors; 3) we apply the LAMBDA method to search the integer ambiguities; and finally 4) validate the fixed solution using a statistical test. We show in this work that the efficient integration of multifrequency/multisystem carriers provides more redundancy in the measurements and better observability for multipath and ionospheric errors estimation for long-baseline RTK positioning. A major advantage of this method is that it is independent of frequencies choice and therefore can be applied for any multi-GNSS measurements (e.g., Global Positioning System (GPS), Galileo, and their combination). Real-time and postprocessing test results show the effectiveness of the developed overall real-time kinematic (RTK) software.     

Robust radar detection of CA, GO and SO CFAR in Pearson measurements based on a non linear compression procedure for clutter reduction

This paper deals with the constant false alarm rate (CFAR) radar detection of targets embedded in Pearson distributed clutter. We develop new CFAR detection algorithms-notably cell averaging (CA), greatest of selection (GO) and smallest of selection SO-CFAR operating in Pearson measurements based on a non-linear compression method for spiky clutter reduction. The technique is similar to that used in non uniform quantization where a different law is used. It consists of compressing the output square law detector noisy signal with respect to a non-linear law in order to reduce the effect of impulsive noise level. Thus, it can be used as a pre-processing step to improve the performance of automatic target detection especially in lower generalised signal-to-noise ratio (GSNR). The performance characteristics of the proposed CFAR detectors are presented for different values of the compression parameter. We demonstrate, via simulation results, that the pre-processed compression procedure is computationally efficient and can significantly enhance detection performance.