integrated INS filter since the INS treats the GPS receiver merely as another

sensor.  Disadvantages are the use of two separate Kalman filters with the potential

instability of cascaded filters, and that the GPS Kalman filter provides correlated

position and velocity measurements that may not be modeled adequately in the INS

filter.

The next level of integration is the tightly coupled, open  loop integrated GPS/INS

Kalman filter which requires implementation at pseudorange level (see Figure 9 5).

This approach has more complex measurement equations but requires only one

Kalman filter mechanization.  The filter can be closer to optimal since effects of

satellite geometry and INS errors are included and INS aiding throughout a GPS

outage can be provided.  Only one naviga tion solution is computed and both carrier

and code tracking loop aiding can be done using the velocity and attitude data from

the INS.

INS UNCORRECTED

POSITION, VELOCITY

AND ATTITUDE

INS

CORRECTED POSITION,

GPS PR AND dPR

VELOCITY AND ATTITUDE

MEASUREMENTS

GPS/INS

GPS RCVR

KALMAN FILTER

Figure 9 5.  Open loop Integrated GPS/INS Architecture

The highest level of integration is the tightly coupled, closed  loop integrated

GPS/INS Kalman filter (see Figure 9 6).  In addition to having only one Kalman filter

and the other benefits of the tightly coupled, open loop filter, the main advantage

here is that the INS is continuously calibrated and will maintain the most accurate

navigation solution in case of GPS outages.

9 13