Performance Evaluation of the 3D MAP Based Precise Positioning and its Application
Nobuaki Kubo and Rei Furukawa, Tokyo University of Matine Science and Technology, Japan
Alternate Number 2
Because of the widespread use of car navigation systems and smart phones, Global Navigation Satellite System (GNSS) positioning has been widely used. In recent years, the use of applications that require stable high-precision positioning, such as those for information construction, is also increasing. In this study, both a 3D building model and a ray tracing method are used to estimate multipath profiles, including the amplitude, delay, and phase of the surrounding building environment, and to generate GNSS signals by using the ray tracing Software Defined Radio method to consider the profiles. We describe the evaluation results by comparing the proposed framework for the desk study of the positioning environment of an urban area with the actual measured results. We also show the performance of the accuracy of prediction for GNSS using only our simulation tool. It will be definitely important for future ITS applications in terms of availability and reliability assessment.
With the advent of multi-GNSS, the number of satellites increase greatly, and the positioning accuracy of navigation services is therefore increased. On the other hand, positional accuracy is degraded by waves reflected from buildings and other such multipath waves when driving in urban areas. In order to bypass the variation in positioning accuracy resulting from multipath error and the time of positioning, it is necessary to carry out simulation over an extended period and evaluate that its positioning accuracy is valid according to actual measurement.
In previous work on GNSS positioning in urban areas, the authors reported the simulation results using 3D maps with the ray tracing method in terms of the availability and accuracy compared with the actual measurement. It is somehow important to know the future performance of GNSS using the simulation. In the actual test, POSLV manufactured by Applanix was used to produce the precise reference positions. These positions are quite important to compare the test results between the actual navigation solution and simulation based solution. Also, simulation requires each precise positions to analyze the radio propagations between our antenna and 3D map based obstacles.
In our new test, we actually did similar test but we tried to analyze more in details between simulation results and actual test results. For example, we can easily detect the direct signals using the simulation results in urban areas and we selected satellites which has direct waves in each epoch. By using these satellites information, we checked the actual navigation performance like as RTK. It shows that the performance of RTK matched well about 80-85% between the actual measurements based RTK and simulation based RTK. Is means that we can predict where we can get the precise positions using only “3D map”. We don’t need the actual measurements. To make the simulation results closer to the actual test results in the above (obtained 80-85% match), we devised a few modifications in the software development. We are going to introduce them in the paper as well as presentation. In addition to the prediction performance, we also introduce the RTK improvement nearby obstacles for survey use in the static condition. By using 3D map information, we selected the satellites which has the direct signals and we compared the RTK solutions between normal RTK without 3D map information and proposed RTK using 3D map information. Finally, RTK fix rate in the challenged areas was improved a lot in the different 4 places. The most significant improvement was from 53.8 % to 99.1 % and the least one was from 69.4 % to 96.0 %.