real time adaptive stream-oriented geo-data filtering

The cutting-edge engineering maintenance software systems of various objects are aimed at processing of geo-location data coming from the employees’ mobile devices in real time. To reduce the amount of transmitted data such systems, usually, use various filtration methods of geo-coordinates recorded directly on mobile devices.

The paper identifies the reasons for errors of geo-data coming from different sources, and proposes an adaptive dynamic method to filter geo-location data. Compared with the static method previously described in the literature [1] the approach offers to align adaptively the filtering threshold with changing characteristics of coordinates from many sources of geo-location data.

To evaluate the efficiency of the developed filter method have been involved about 400 thousand points, representing motion paths of different type (on foot, by car and high-speed train) and parking (indoors, outdoors, near high-rise buildings) to take data from different mobile devices. Analysis of results has shown that the benefits of the proposed method are the more precise location of long parking (up to 6 hours) and coordinates when user is in motion, the capability to provide steam-oriented filtering of data from different sources that allows to use the approach in geo-information systems, providing continuous monitoring of the location in streamoriented data processing in real time. The disadvantage is a little bit more computational complexity and increasing amount of points of the final track as compared to other filtration techniques.

In general, the developed approach enables a significant quality improvement of displayed paths of moving mobile objects.