Global navigation satellite systems (GNSS) and inertial navigation systems (INS) are naturally complementing positioning and navigation technologies, because each one compensates for the other’s deficiencies. Therefore, as demand for reliable positioning and navigation in a variety of GNSS-challenged environments grows—most notably, due to the development of autonomous vehicles—so does demand for GNSS-INS integration and manufacturers’ need to simulate it.
GNSS and INS Help Each Other
GNSS receivers have astounding absolute positional accuracy, but are subject to signal loss or degradation, due to occultation (urban canyons, thick canopy, indoor environments, etc.), multi-path, unintentional interference, intentional jamming, and spoofing. Conversely, INS are low noise and accurate in the short term, provide an excellent dynamic response at high data rates, produce data that is solely dependent on the motion of the platform and local gravity, and determine proper orientation (roll, pitch, and heading), but suffer from drift — the steady decrease in the accuracy of the position estimate due to the accumulation of errors.
By periodically re-initializing an INS, a GNSS receiver constrains its drift; conversely, the INS fills the gap when GNSS signals fail and helps to remove measurement outliers. Thus, integrated GNSS-INS positioning, navigation, and timing (PNT) systems can benefit from the best of both worlds. From guided missiles to autonomous vehicles to pedestrian dead reckoning, many PNT systems now maintain accuracy this way.
Sensor Fusion Algorithms
In recent years, many kinds of GNSS-INS sensor fusion algorithms have been proposed, developed, and adopted, for different grades of IMUs. The three most common approaches are loose, tight, and ultra-tight integration. The basic difference between the loose and the tight approaches is the type of data shared by the GNSS receiver and the INS sensors. In a loosely coupled system, the GNSS receiver is aided with position and velocity, enabling it to keep tracking during tight turns, jamming, and other adverse conditions. The positions and velocities estimated by the GNSS receiver are blended with the INS navigation solution through a navigation Kalman filter. In a tightly coupled system, on the other hand, the correlators are pre-set according to the inertial motion that the platform is undergoing. GNSS raw measurements (pseudoranges and Doppler shift) are processed through a Kalman filter jointly with the INS measurements, helping the receiver track the signal rather than just helping it at the end. The ultra-tight integration approach involves the GNSS receivers’ digital tracking loops, which are typically not accessible using commercial off-the shelf (COTS) products.
The main advantage of the tight integration is the ability to update the hybrid navigation solution also in scenarios with poor GNSS signal quality or limited view of the satellites by predicting pseudoranges and Doppler trends. This approach is particularly beneficial in urban environments, but it is still being researched and developed and requires INS containing high-grade inertial measurement units (IMUs). By contrast, most micro electromechanical systems (MEMS) IMUs are only loosely coupled with GNSS receivers. This is the most common GNSS-INS integration approach, though this may change soon with further improvements in the performance of MEMS IMUs.
A big challenge in simulating GNSS-INS integration, whether the two systems are loosely or tightly coupled, is the synchronization between the IMU stimulus and the GNSS stimulus that are going to be seen by the unit under test. For example, if you are simulating an aircraft is flying straight and level at 200KT, then making a 90 degree turn to the right, you need to make sure that your GPS stimulus and your IMU stimulus all happen at the same time, so that the solutions that the navigator is calculating from those two different sensors do not diverge. However, the standard satellite message rate is 50bps, the satellites are orbiting at speeds greater than 17,000MPH, a typical IMU has a 256Hz update rate, the system processor runs on a 100MHz clock, and the signal generators are on another frequency. You must ensure that all those timings are synchronized, so that the GNSS and INS data outputs do not diverge either in time or stimulus from each other.
Previously, navigation systems that used only GNSS receivers could be tested in a laboratory with a regular simulator. When dealing with embedded GNSS-INS (EGI) systems, the navigators believed the INS portion of the navigator, permitting only stationary testing with a GNSS simulator. Testing an EGI in a laboratory requires de-coupling of the inertial solution from the GNSS solution. This is done electronically: the INS is ‘put to sleep’ and simulated inertial data is input directly to the EGI’s computer.