Networked Differential GPS Methods

Historically, maritime organizations seeking accurate shipboard positioning have relied upon some form of differential GNSS, such as DGPS, WAAS, or EGNOS, to improve the accuracy and integrity of the GPS. Ground-based augmentation systems, such as DGPS, broadcast corrections to the GPS signal from geographically distributed terrestrial reference stations—often called beacons. Specifically, pseudorange corrections to the GPS L1 C/A signal are computed at each reference site, then broadcast in the nearby geographic area using a medium frequency (approximately 300 kHz) communications link. The user then adds these corrections onto their measured pseudoranges before implementing a position solution algorithm. Within the United States, the U.S. Coast Guard operates 86 DGPS reference beacons. Similar DGPS systems are operated in Europe and elsewhere around the globe. While current DGPS receiver algorithms typically use one set of pseudorange corrections from one DGPS reference site (often the one with the “strongest” signal), many user locations can successfully receive two or more different DGPS broadcasts. This suggests two obvious questions: “If available, how does one select the corrections to use from multiple sets of corrections?” and “Is it advantageous to combine corrections in some way?” A number of factors might influence the effectiveness of any particular station’s corrections. Some of these refer to the effectiveness of the communications link itself, including concerns about interference from other beacons (skywave interference from far-away beacons on similar frequencies, a notable problem in Europe) and self-interference (skywave fading). Other factors refer to the accuracies of pseudorange corrections: for example, ionospheric storm-enhanced plasma density (SED) events can cause the corrections to have large spatial variation, making them poor choices even for users close to a beacon. Earlier work in the area of DGPS beacon selection has identified several options, including choosing the beacon closest to the user or the beacon with the least skywave interference. There have also been suggestions on how to combine corrections when multiple beacons are available. The most common among these is a weighted sum of the corrections, where the weights are typically inversely proportional to the distance from the user to the individual beacon. This thesis re-examines the concept of multi-beacon DGPS by evaluating methods of combining beacon corrections based on spatial relativity. Recent research determines that DGPS accuracy performance is biased: the mean scatter of DGPS-corrected positions does not fall on the true receiver position. This finding was re-established this using networked DGPS methods both by processing GPS L1 C/A observables from dozens of CORS (Continuously Operating Reference Station) sites around the U.S.A. and via simulation using a Spirent GSS8000 GPS simulator. Specifically, we found that (a) the position solution computed using DGPS beacon corrections is typically biased in a direction away from the beacon and (b) the magnitude of the bias depends upon the distance from the beacon. This bias grows with a slope of approximately one-third of a meter per 100km of user-to-beacon distance. We also found that networking DGPS corrections decreases the errors of bias magnitude and scatter radius inherent in singlebeacon solutions. This thesis compares the performance of several multi-beacon algorithms assessed using both GPS simulator and real-world data. These algorithms include simple averaging, a weighted sum based on inverse-range to each beacon, a weighted sum based on inverse-range-squared to each beacon, and spatial linear interpolation correction. Spatial linear interpolation factors in distances and angles to the known locations of the DGPS transmitters. As part of this research effort, we developed a DGPS receiver using a softwaredefined radio platform. Ettus Research’s USRP was chosen as the SDR device to collect and digitize GPS and DGPS radio signals. For the real-world tests, we applied networked DGPS pseudorange corrections to post-processed CORS data. The results of these tests confirm the spatial behavior of the simulator trials with respect to bias magnitude and scatter radius. A complete description of this system is included in the thesis