Integration Methods

The method of integration determines the role of each instrument and its measurements take in the production of the output. The types of integration are in terms of the progressive steps taken in the amalgamation of the individual processes of INS and GPS towards a completely unified system. The systems shown below represent the major steps and divisions that occur. Between each step shown there are numerous levels of separation. Even if there was only one inertial navigation system and one global positioning system there would still exist countless possible architecture types with regard to the different interactions that are possible.

Uncoupled Systems

This is the first step towards integration where each system operates individually and the final output of each system is integrated in some form. Generally the GPS output is used to reset the INS measurements at regular intervals. While this is effective in bounding the error growth rates of the INS there is no feedback of the error estimates that can be made to assist the operation of either system which prevents the opportunities for performance enhancement that is achieved in the following systems.

Loosely-Coupled Systems

In a loosely coupled system the function of one system is generally unaffected and the measurements made by this system are used to correct the errors in the other system. The common approach is to use the unaffected output of the GPS to correct the errors of the INS. The corrections for the INS are generated from the integrated solution of both systems.

Loosely Coupled System from Hjortsmarker (2005)

The loosely coupled Kalman filter works on a prediction and correction cycle. The prediction step in the filter process involves using the INS calculation of the position, velocity and attitude as a prediction. At each GPS update the Kalman filter uses the computed GPS position as a measurement by which to adjust the predicted INS position. The resulting output of the Kalman Filter is the corrections that are applied to the INS.

The main advantages of loosely integrated systems are the simplicity and redundancy of the system. The position and velocity from the GPS can be used to check the integrated solution. The GPS result is used to correct the errors in the INS each time a GPS solution is produced, which is not capable in the individual INS system. The independence of the GPS means that in the absence of enough satellites the GPS is unable to provide information to the Integrated Kalman Filter for the benefit of correcting the INS errors.

Tightly-Coupled Systems

This system is also known as closely coupled, direct integration or centralised system. In a tightly coupled system the processing of all measured quantities is undertaken in the integrated Kalman Filter. The system treats all information as being complementary data for the production of a single solution. The integrated Kalman Filter uses the pseudorange and rate data supplied by the GPS to correct the INS data. The INS sensors are used to create the prediction of the position, velocity and attitude through the use of the navigation equations of an INS. The predicted INS values are used to predict the values of the pseudoranges. The measured GPS pseudoranges are then used as a measurement by which the predicted INS pseudoranges are adjusted.  In this system as there is no separate GPS solution so the requirement to track four satellites at all times is not necessary. The pseudorange and rate data from just a single satellite can be beneficial to the estimation of the INS errors in the short term. It must be noted that the quality of the corrections quickly degrade when only one or two satellites are tracked and the accuracy of the estimates increases with the number of satellites being tracked.

Tightly-Integrated System from Titterton & Weston (2004)

Ultra-tightly Coupled Systems

The ultra tight system combines GPS signal tracking and the integrated Kalman Filter. The ultra-tightly coupled system takes the combination of tightly coupled systems a step further by combining GPS signal tracking and the integration into a single Kalman filter. The benefits of this are that the effective signal to noise ratio is improved and the more satellites are tracked the greater the improvement is. It also reduces the susceptibility to multipath errors and improves the signal reacquisition following loss of lock.