Robust locating using LPS LOSNUS under NLOS conditions

In practical applications, the accuracy of a LPS is often affected by non-line-of-sight (NLOS) errors which disable proper locating results. In this article, a robust localization algorithm capable of minimizing the effects of NLOS errors during position estimation is presented. The robust localization algorithm uses hyperbolic multilateration in combination with proximity based grouping and final determination of the position by a geometric dilution of precision (GDOP) weighted mean. GDOP weighting ensure that high quality solutions are assessed over low quality solutions while still being able to benefit from averaging. Proximity of two positions is defined by their respective GDOPs combined with the TOA uncertainty which ensures a tight boundary to avoid outliers. This grouping method yields remarkable results in a real world environment. The article includes a problem description, a description of the system and algorithms, an evaluation of the GDOP based on our system configuration, a performance comparison with other NLOS mitigation methods and a real world measurement showing the excellent suitability of our system as an LPS.