Research On The Tracking Technology Of GNSS/INS Deep Integration Based On Hardware Prototype

Global navigation satellite system (GNSS) and inertial navigation system (INS) are highly complementary. Therefore the research and development of GNSS/INS integrated navigation have been a hot spot in the past two decades. According to the GNSS information used for data fusion, GNSS/INS integrated navigation could be divided into loosely-coupled system that based on GNSS navigation results, tightly-coupled system based on GNSS observations, and deeply-coupled system based on GNSS baseband signals. INS has been the main navigation method before the popularity of GNSS. The loosely coupled system solved the divergence of INS by GNSS aiding of position and velocity, which is a relatively mature technology. Besides GNSS aiding INS, the tightly coupled system could also make INS aiding GNSS in the navigation data processing level and improve the stability and reliability of the GNSS positioning. The deeply coupled system, which was originally promoted at year2000, could further take advantage of the dynamic characteristics of INS to improve the acquisition and tracking performance of satellite signals in the GNSS receivers.According to the architecture of GNSS receiver’s tracking loop, the deeply coupled systems are divided to scalar and vector mode. The scalar deeply coupled system is based on the conventional GNSS receiver tracking loops, with independent tracking channels. The INS aiding information is inserted into each channel of tracking loop to help the signal tracking separately. Therefore, the scalar deeply couple is relatively simple, reliable, and easy to implement. In the vector deeply coupled system, the INS data and the GNSS signal tracking data are integrated in a centralized navigation filter to make a centralized data fusion mechanism. Then the estimates from the filter are feedback to control every tracking channel, which forms a unified signal tracking strategy. Therefore, the vector deeply couple can make best use of the available information and get better performance. However, because of the system complexity and the heavy computation load, it is a challenge to develop a stable and reliable real-time vector deeply coupled system that can head for engineering and production.The deeply-coupled system development requests a relatively high technical threshold, which needs to be proficient in the GNSS baseband signal processing, the inertial navigation, and the GNSS/INS navigation algorithm, and the capability for system integration. Therefore, most studies on the deeply-couple technology remain in the simulation stage. Previous works have limitations of1) No systematic and complete theoretical model for GNSS signal tracking error analysis of the deep integration.2) Lack of hardware platform for real-time deep integration research.3) The GNSS baseband performances of deeply-coupled system haven’t been well analyzed, evaluated and optimized. These limitations block the progress and application of the GNSS/INS deeply-coupled technology.This thesis studies the GNSS/INS deeply-coupled technology based on the scalar deep integration for GPS L1receiver. It establishes a complete set of error models and corresponding quantitative analysis methods for the INS aided carrier tracking loop; implements real-time deep integration based on an integrated hardware prototype platform; and makes quantitative evaluation to the dynamic tracking accuracy and continuity of the INS aided tracking loop. The main contributions of this research include:Firstly, the thesis establishes the error propagation models of the INS aided tracking loop, and analyzes the tracking error of the loop quantitatively. It derives the transfer functions between the error sources (including thermal noise, oscillator phase noise, inertial measurement unit (IMU) error, and the delay of Doppler aiding information) and the tracking loop error of the deep integration, which provides a powerful tool to analyze the effects of the error sources on the INS aided tracking loops. The analysis shows that:with INS aiding, the tracking loops filter only need to track the residual errors of the aiding information; the major error source from the INS aiding information that affects the static error of the tracking loop, is the velocity estimation error from the GNSS/INS solution, while the bias error and the noise of the IMU are minor; the IMU scale factor error and the aiding information delay are the major factors affecting the dynamic error of the tracking loop. The transient and steady state response characteristics of the tracking loops with and without INS aiding are analyzed as well. The results show that, with INS aiding, the steady state tracking error can be reduced through compressing loop bandwidth and extending integration time. Furthermore, the transient period should be shorter and the dynamic error should be smaller.Secondly, a hardware/software integrated GNSS/INS scalar deep-coupled prototype is successfully developed. The INS aided GNSS carrier tracking loop and INS controlled open-loop tracking are realized. A real-time deep integration scheme is proposed. The key modules (including the baseband tracking loops, INS data sampling and computation, and Doppler aiding information generation) are designed; and real-time optimizations are made in terms of the system operation and aiding information delay. Furthermore, the errors of open-loop tracking are modeled and analyzed; an open-loop control strategy is designed to improve the continuity of the signal tracking.Thirdly, the performance of the designed deeply coupled prototype is fully evaluated based on a GPS/IMU hardware simulator and real field tests. The proposed INS-aided tracking loop error models and the relevant analytical results are verified at the same time. 1) Effect of every error source in the INS aided tracking loops is evaluated quantitatively. The test results are consistent with the conclusions based on the model analysis, and therefore verify the correctness of the error models and analysis methods.2) The transient and steady state characteristics of the tracking loops with and without the INS aiding are tested. The result shows that the INS aiding could improve the steady state tracking performance by extending the integration time to20ms and by compressing the bandwidth to3Hz under normal dynamic conditions. Their transient period and error have obvious advantages over the conventional tracking loops.3) The INS-aided open-loop tracking performances with satellite signals blocked partially and completely are both tested. The results agree with our theoretical analysis:When satellites are partially blocked, low-grade INS can maintain continuous open-loop tracking of the carrier frequency and code phase within the tolerant error range, while mid-grade INS can maintain the open-loop tracking of the carrier phase. However, if all the satellites are completely blocked, INS can only keep open-loop controlling for several seconds.4) The performances of the final GNSS/INS integrated navigation solutions with and without INS aided tracking loops are compared in the dynamic tests. The results demonstrated that the INS aiding leads to a considerable decrease of the velocity errors for both GNSS PVT and the GNSS/INS integrated solution. But there is no improvement of positioning accuracy, especially in open sky environment. When the satellite signals are severely blocked and attenuated, the continuity of the receiver’s position and velocity is significantly improved with the INS aided tracking loop.In summary, this thesis does a thorough study on the GNSS/INS deeply-coupled system with the scalar tracking loop. It proposes a set of complete and practical error models for the INS aided tracking loop error analysis. A fully integrated real-time deep integration hardware prototype is designed and implemented as test platform. Comprehensive simulation tests and real tests are made to verify the proposed error model analysis and the expected performance advantages of the deeply-couple. The proposed error models, design methods, and hardware prototype developed in this thesis can be further applied to the key performance study of the GNSS/INS deeply coupled system, such as the sensitivity and anti-interference under dynamic conditions.