RANSAC-based instantaneous real-time kinematic positioning with GNSS triple-frequency signals in urban areas

Abstract

The demand for high-precision positioning has risen substantially in modern urban settings. In that regard, Global Navigation Satellite Systems (GNSS) offer several advantages such as global coverage, real-time capability, high accuracy, ease of use, and cost-effectiveness. The accuracy of GNSS-based positioning, however, suffers in urban environments due to signal blockage, reflection, and diffraction, which makes it difficult to fix ambiguities correctly within a real-time kinematic (RTK). To address this issue, this paper applies random sample consensus (RANSAC) to develop a novel single-epoch triple-frequency RTK positioning method. In our proposed method, the ambiguities of the extra-wide-lane, wide-lane, and original frequencies are resolved sequentially. RANSAC then detects and excludes incorrectly fixed ambiguities. To validate the effectiveness of the proposed method, two static experiments (cases 1 and 2) and one dynamic experiment (case 3) were conducted in representative urban areas. The findings demonstrate that the proposed method outperforms all comparative methods in positional availability, with comparable positional accuracy in terms of root-mean-square errors (RMSEs). In cases 1, 2, and 3, the proposed method achieves 3D RMSEs of 2.74, 4.29, and 20.35 cm, and the positional availabilities of 100%, 75.0%, and 73.1%, using a 10-degree mask angle (and a carrier-to-noise ratio (C/N₀) threshold 35 dB-Hz). The corresponding RMSEs (positional availabilities) of comparative methods are from 1.51 to 4.04 cm (75.7 to 96.3%) in case 1, 4.19 to 7.78 cm (34.5 to 49.9%) in case 2, and 23.52 to 37.54 cm (15.4 to 33.9%) in case 3, respectively. Compared to these methods, the proposed method shows improvements of positional availabilities between 3.7 and 24.3 percentage points in case 1, between 25.1 and 40.5 percentage points in case 2, and between 39.2 and 57.7 percentage points in case 3.