Geography Reference
In-Depth Information
Inaccuracies in the Positioning of Mobile Phones and Their Impact
on Data Quality
A mobile phone
s GPS is frequently affected by high inaccuracies in positioning,
especially in urban areas. The main cause is a reflected GPS radio signal, which
leads to an incorrect calculation of transmitter-receiver distance and subsequently,
incorrect positioning. The resulting inaccuracy in an urban environment lies usually
between 3 and 40 m (Modsching et al.
2006
). However, the noise level decreases
with the square of the distance. If the positioning system indicates, that the distance
from the noise source is 2 m, but the actual distance of the measuring device is only
1 m, the measured value should be about 6 dB higher than it would be for indicated
distance, which makes moderately loud noise sources (e.g. industrial installations)
hardly mappable.
Positioning using trilateration from GSM base stations has an error of one order
of magnitude higher,
'
than GPS (Pent et al.
1997
) and therefore cannot be
recommended at all.
One of the possible solutions may be to link positional measurements up to map
data. The position information of each measurement point can be assigned to the
nearest street [a similar principle has been tested in the project NoiseTube (Stevens
2012
)]. In the case of a chain of sample points (obtained, for example, while the
user walks or rides a bike), the corresponding section of this sample chain may be
assigned to that street segment. However, this means that detailed digital maps,
which also include sidewalks outside of the main street layout, must be available for
post-processing. For particularly wide or asymmetric streets is necessary to con-
sider on which side user is.
If the position signal is lost or user is outside of paths, stored in the reference
geodatabase, it is advisable to create an opportunity to additionally georeference
measured data (both in the field and during post-processing). If the user does not
have a GPS signal, he/she should be alerted, and given the option to record map
points or nodal points of his/her path (such as road intersections, stops, etc). Then
the map application could identify those points with places on the map and the
measured values would be uniformly spatially distributed between these points.
Effect of Background Noise and User Behavior on Noise Measurements
When measuring the general noise situation, residual noise from distant sources
does not constitute a significant problem because it has to be included in each
measurement and it need not to be separated like for the evaluation of one type of
noise source
s properties.
But in the case of measurement conducted by a moving pedestrians, background
noise loudness of his/her own footsteps is undesirable. The noise level of the walk
depends mainly on the type of footwear (shoe sole), the nature of the surface, on
which the walker goes, and frequency of his/her steps. For shoes with soft to
'
