Mitigating the Correlations in INS-aided GPS Tracking Loop Measurements: A Kalman Filter Based Approach

The fusion of GPS and INS is undertaken at different levels to leverage the benefits of both technologies, and to provide a robust solution. Loose and Tight Coupling schemes have been in existence for more than a decade. Though these two configurations were used in many applications, the goal is always to improve the robustness of the navigation solution. With this objective, and the rapid developments in low-cost MEMS Inertial Measurement Unit (IMU) technology, the integration of GPS and INS has taken a new direction. The ‘UltraTightly Coupling’, or integration at the tracking loop level, is a new field of research. The basic concept behind the ultra-tight integration approach is that, the dynamics of the GPS receiver measured by an INS can be integrated with the GPS signals. This result in ‘dynamic-free’ GPS signals that are sent to the tracking loops facilitating a significant reduction in the carrier tracking loop bandwidth, hence providing accurate carrier and code phase measurements. Therefore accurate estimation of the Doppler frequency from the INS becomes paramount in leveraging the benefits of ultra-tightly coupled systems. However, there are many errors that inhibit this accurate estimation: approximations in modeling the INS errors, errors in the interpolator designs, synchronization issues, etc. The cumulative effects of these errors result in time correlations in the I (in-phase) and Q (quadrature) measurements, which subsequently bias the discriminator output that drives the NCO (Numerically Controlled Oscillator). If unaccounted for, these correlations result in sub-optimal navigation solutions. This paper proposes an effective tracking loop structure based on a Kalman filter that mitigates the effect of correlations due to inaccurate INS Doppler estimates. This proposed design also reduces the number of state vectors in the navigation filter, leading to reduced complexities. The equations for the system and discriminator error models are derived. Simulations are performed with different INS Doppler accuracies at a carrier loop bandwidth of 3Hz. The performance of the modified tracking loop is compared with those of the conventional ultra-tight tracking loop. The results show that a significant improvement in the tracking loop performance can be obtained by using the proposed design.