Research On INS-aided Carrier Phase Tracking Loop Based On Kalman Filter

In recent years,much progress has been made for GNSS in military and civilian use.However,GNSS still faces great challenges in high dynamic scenarios(e.g.high precision guided weapon)and complex signal environment in urban areas.Signal tracking loops in conventional GNSS receivers have fixed loop coefficients and face a trade-off between dynamic characteristics and noise characteristics.Besides,tracking channels in conventional scalar receivers are independent of each other and information sharing among different channels does not exist.Therefore,conventional receivers can not maintain reliable and continuous tracking and navigation performance in high dynamic and complex signal environments.In order to solve this problem,the KF based carrier tracking loop with natural self-tuning technique is adopted in the level of signal baseband processing,and the INS aided carrier tracking loop technology is adopted in the level of upper class architecture.The main contributions of this paper are as follows.Firstly,similarities and differences between 3^rd order conventional PLL and KF-based PLL are analyzed through mathematical derivation and simulation tests from three aspects:time of transient response,carrier doppler extraction,carrier phase errors caused by different NCO tuning methods.In the aspect of transient response,based on the discriminator output and the proposed transient time search method,the time of transient response of 3^rd order conventional PLL and KF-based PLL is tested.The test results demonstrate that KF-based PLL has much shorter transient time(smaller than 0.4s)than conventional PLL does due to its larger transient bandwidth.In the aspect of carrier doppler shift extraction,through mathematical derivation and simulation tests,it is demonstrated that under the stimulation of frequency rate ramp signal,the carrier doppler shift extracted from the velocity accumulator of 3^rd order PLL has a steady state error but good noise characteristic,while the carrier doppler shift extracted from the loop filter output has no steady state error but bad noise characteristic.Based on the structural equivalence of KF-based and conventional PLLs,the carrier doppler frequencies extracted from the same place of the two kinds of tracking loops share the same dynamic characteristics and noise characteristics.In the aspect of the carrier phase error caused by different NCO tuning methods,the influence of carrier phase tuning and carrier frequency rate tuning on the carrier phase error over one coherent integration interval is quantitatively analyzed.Besides,a method that the acceleration accumulator estimate is used to tune the frequency rate of NCO is proposed in the paper.It is much easier to implement compared to the previous methods.The analyses result demonstrates that the phase error may exceed 360 degrees,i.e.,a cycle slip may happen,if the signal tracking loop utilizes a long integration interval or works under high dynamics.Therefore,frequency rate feedback to NCO is crucial for accurate carrier phase measurements in applications where high signal dynamics or long coherent integration intervals may influence the tracking performance of receivers.Secondly,the software platform of GNSS/INS scalar-based deep integration architectures with KF-based signal tracking loops is implemented.The receiver platform can work in two modes:scalar-based receiver+integrated navigation filter and scalar based deep integration architecture.The parameters of KF based signal tracking loop and integrated navigation filter are set according to their physical meanings.On the one hand,the measurement noise variance R can be changed with signal conditions.On the other hand,a parameter design method for KF tracking loop in the deep integration architecture according to the inertial sensor errors and receiver oscillator errors is proposed.According to the analyses,for INS aided tracking loops,white noise and bias instability of accelerometer output account for the main reason for the INS aiding doppler error while the ephemeris calculation error and the instrument error of gyroscope can be neglected during one second of GNSS update under low dynamic scenarios.Since each parameter is set according to its physical meaning,the empirical and tentative trials in the KF parameter optimization process are avoided.Moreover,the proposed parameter design method also provides a reference for the conventional tracking loop choosing the optimal bandwidth in the deep integration architecture since these two types of tracking loops share the equivalent structures.Thirdly,based on the GNSS/INS hardware simulator,the KF-based tracking loop?conventional tracking loop and scalar based deep integration architecture are tested and evaluated under common vehicular dynamic scenarios where partial(two)or all satellites(ten visible satellites)signals are attenuated.The test results demonstrate that the self-tuning KF-based PLL has 3-7d B gain in carrier frequency tracking sensitivity,2-4d B gain in carrier phase tracking sensitivity,and 5d B gain in navigation solutions over conventional PLL under 20ms coherent integration time and 13.4Hz initial bandwidth.The improvement of carrier tracking sensitivity brought by INS aiding depends on the number of satellites with weak signals.In the scenarios where partial satellites are attenuated,the sensitivity of carrier frequency tracking and navigation solutions of INS aided PLL is better than 13d B-Hz;INS aided PLL has at least 5-8d B gain in carrier frequency and 1-2d B gain in carrier phase over KF-based PLL without INS aiding.In the scenarios where all satellites are attenuated,INS aided PLL has only 3-4d B gain in carrier frequency tracking and navigation solutions over KF-based PLL without INS aiding,and1d B gain in carrier phase for only a few satellites.In conclusion,a clear and reasonable scheme of INS aided signal tracking loop based on Kalman Filter is proposed;similarities and differences between 3^rd order conventional PLL and KF-based PLL are comprehensively and quantitatively analyzed;GNSS/INS deep integration software platform is implemented;a parameter design method for KF tracking loop in the deep integration architecture according to the inertial sensor errors and receiver oscillator errors is proposed;the improvement of carrier tracking under weak signals brought by KF-based tracking and INS aided tracking is tested and verified through GNSS/INS hardware simulation data.The research in this paper indicates that the problem of carrier tracking and navigation performance decline of conventional PLL under weak signals can be effectively relieved by KF-based tracking and INS aided tracking.