Long-baseline Triple-frequency GPS Integer Ambiguity Resolution Using Different Observation Models

Abstract

For GPS long-baseline relative positioning, the effects of ionospheric and tropospheric delays on GPS observations cannot be fully eliminated by double-difference because the spatial correlation decreases as the inter-station distance increases. Currently, the most commonly used two observation models for estimating dual-frequency long-baseline vectors are: 1) using the ionosphere-free linear combination of dual-frequency observables and 2) directly using the original L1 and L2 carrier-phase observations. These two models can be referred to as ionosphere-free and ionosphere-float models respectively. In this paper, similar models but for triple-frequency case was developed and ambiguity resolution algorithms using the triple-frequency ionosphere-free models were derived. Furthermore, the reliability of the ambiguity resolution based on ratio values and bootstrapped success rates for both models was assessed using simulated triple-frequency GPS observations. The numerical results showed that the coordinates and success rate estimated from the two models are almost identical for the dual-frequency case. However, in the case of triple-frequency, the success rate of the ionosphere-float model is twice that of the ionosphere-free model. This indicates that the ionosphere-float model is more suitable for triple-frequency long-baseline RTK, which requires fast integer ambiguity resolution.