Abstract:
Position fusion combines locations derived from distributed sensors, monitor an entity,
to obtain an accurate estimate of the entity's position. It takes advantages of redundancy
and diversity present in the measured data. If the measured data from multiple sensors
are correctly combined, then the fused sensor data has better accuracy. This paper
addresses the problem of position fusion, based on Bayesian minimum mean square
error criterion, using measurement from multiple sensors with dissimilar accuracies.
The results show that the performance of the fused measurements may perform worse
than the performance of the individual sensor measurements. The best performance of
the fused measurements occurs when the sensors have the same accuracies. The
performance of the fused measurements is worse than the performance of the best
accurate sensor when the sensor accuracies vary widely. In this case, adopting the best
accurate sensor is recommenced and position fusion is not recommended.