A New Method for Coverage Prediction for the Wide Area Augmentation System (WAAS)

To provide a detailed analysis framework for WAAS accuracy prediction, a new method has been developed which combines GPS geometry simulation with leastsquares covariance analysis. This approach considers all standard ranging errors for each WAAS reference station receiver and user, including satellite clock, SA-induced latency, ephemeris, multipath, receiver noise, troposphere and ionosphere errors. Computer simulation allows us to project mean, 2σ, and 3σ position error for a grid of user positions across a wide geographic area. The simulation updates orbit positions of the GPS satellites augmented by four geosynchronous spacecraft. For each resulting geometry, satellite geometry for all WRS's is determined. The simulation then loops through a grid of user positions. The satellite geometry for each user is computed, and the user's clock/ephemeris error is projected through covariance matrices that link the error at each WRS to the user (including non-WAAS errors). Ionosphere errors are handled by a separate covariance algorithm running in parallel. This algorithm uses a MITRE-like grid of ionosphere corrections, but instead of using interpolation, a "truth" model of vertical ionosphere covariances is developed and projected (using a weighted least-squares solution) from the WRS pierce points to the master station grid. The user then projects the error covariances back to his own pierce points (including non-iono. WRS errors). The resulting clock/ephemeris and ionosphere covariances for each user are added together. The user then computes a weighted least-squares position and finds his expected 1σ position error from the resulting covariance. The simulation stores histograms of vertical position error at each grid point and compares them to the 4.1-m 2σ ILS navigation requirement for airborne Category I precision landing. Availability is tabulated for each user position, as is the maximum availability outage period. Results presented for the Stanford 3-WRS experimental WAAS network show that the Stanford WAAS meets this accuracy requirement over a large area. The results also validate the observed vertical errors from Stanford’s WAAS flight trials. Finally, this paper presents accuracy coverage predictions for proposed FAA NSTB WAAS testbed networks that cover the entire Continental U.S.