Grouping-sparsity enforcing LASSO-based outlier detection and correction for Doppler velocity determination in urban areas

Abstract

Velocity incorporates user dynamic characteristics, facilitating more precise predictions about the positioning. However, the positioning, velocity, and timing services derived from Global Navigation Satellite System (GNSS) undergo accuracy degradation in urban environments due to multipath/Non-Line of Sight (NLOS) effects. Fault detection and exclusion (FDE) methods can mitigate these effects. However, the existing methods, such as the multi-hypothesis separation solution (MHSS), exhibit high computational burdens and cannot perform accurate exclusion due to the excessive fault modes. In response, a fault detection and correction (FDC) method is developed to address outliers arising from multipath/NLOS in the Doppler measurements. To alleviate computational demands while simultaneously improving velocity estimation accuracy, multipath/NLOS sparsity assumptions and grouping constraints are introduced. Specifically, the grouping-sparsity enforcing Least Absolute Shrinkage and Selection Operator (GS-LASSO) is introduced to jointly detect and correct multipath/NLOS-induced outliers. A grouping strategy based on sky-map and carrier-to-noise ratio is introduced, which is coupled with a new cost function to improve sparsity estimation. To facilitate the implementation, a solver and parameter-tuning method incorporating false alarm rates are developed. The performance of GS-LASSO is compared with that of MHSS. The results show that GS-LASSO reduces greater velocity errors in the urban environment, while requiring limited computational load.