Simulation-Based Evaluation of WAAS Performance: Risk and Integrity Factors

The Wide Area Augmentation System (WAAS) uses an array of monitor stations that combine to compute GPS user corrections over the Continental U.S. Preliminary experiments have concluded that ionospheric spatial decorrelation is the most significant of the WAAS error sources. Recent data on ionospheric errors has been combined into a probability model that describes the likelihood of rare-event ionospheric decorrelations over a range of conditions. Using this model and simpler models for troposphere and other errors, computer simulations of WAAS performance for randomly located users, using the MITRE ionospheric grid algorithm, have been conducted. These simulation studies focused on performance sensitivity to various algorithm parameters and the potential of user-based RAIM algorithms to meet the WAAS availability and integrity requirements. It was found that both standard residual statistics and a new method of projecting spatial decorrelations from the gridpoint estimates show promise in improving overall WAAS performance. Remaining ionospheric uncertainty prevents us from meeting all the requirements, but more detailed experiments will allow us to improve our models and offer better performance. 1.0 Introduction The Wide Area Augmentation System (or WAAS) has in the past few years grown from a theoretical study into an FAA proposal to build a network of about 20 monitor stations (or WRS's) coordinated by a master station (WMS) which computes DGPS corrections and transmits them to users from the ground or through geosynchronous satellites, which also serve as GPS ranging sources [1]. The components of this system could be in place by 1997. WAAS development is being supported by experiments at MITRE and Stanford University which use smaller test networks of 3 or 4 monitors to gather accuracy data [2,4,13]. These preliminary tests have provided data that allows us to better model wide-area DGPS performance for the full-scale system. Since current data suggests that ionospheric spatial decorrelation is the most serious threat to WAAS accuracy and integrity, we have constructed a probability model based on the latest experimental data to model our uncertainty regarding the magnitude of "worst-case" ionospheric errors as well as the error magnitudes to be expected under more normal conditions. This model forms the basis for a series of WAAS performance simulations in which accuracy and integrity for users randomly located in the Continental U.S. is measured. Our goal at this stage of WAAS program development is to use computer simulation to project WAAS performance given what is now known. Due to the limited error data and uncertain error models we have now, it is very difficult to meet the performance requirements issued by the RTCA for Category I aircraft precision landing [1,9], but our results under these limitations suggest that it will be feasible to meet the requirements once data from WAAS full-system tests is available in a couple of years. 2.0 Breakdown of WAAS Error Sources WAAS employs GPS corrections computed by a network of reference stations to remove most of the satellite-based errors that exist without differential corrections. Small errors due to spacecraft clock and ephemeris remain, however, and depending on the latency (time to reception) and age (time to last usage) of DGPS corrections, Selective Availability (SA) will contribute a ranging error estimated by the simple kinematic relation: