Information Technology Reference
In-Depth Information
Geographic reference systems
. Movement traces are located in space-time, hence
their management and analysis requires a measurement framework of up to three
interrelated spatial dimensions.
Temporal reference systems and sequence
. On top of the above spatial refer-
ence systems, movement inherently requires a fourth, temporal dimension. This
dimension is typically directed and keeps progressing, but can alternatively be
conceptualized as being cyclical or branching (Frank
1998
,
2001
). Consequently,
observations of moving objects have a sequence that can be time-stamped and
typically can't be reversed.
Permanent instead of sporadic change
. Moving objects by definition are in motion,
and the change of position is the norm, not the exception. This is fundamentally
different to conventional spatial objects, where in essence a static world is assumed
that undergoes sporadic change or up-date events. Hence, many techniques that
rely on supporting data structures (such as indices, trees, aggregations) are funda-
mentally challenged because the entities permanently rearrange.
Movement traces are complex objects
. Whereas the moving objects themselves
are most often modeled as simple moving point objects, the traces they leave in
space-time are complex. Relations such as distance or similarity are consequently
more complex than relations between simple data points in conventional feature
space.
Implicit relationships
. Just as for relationships amongst spatial objects, relevant
relationships between moving objects are often implicit and must be materialized
first using metrics and operations considering up to four dimensions. Examples
include spatio-temporal topological relations such as meet, line up, diverge or
converge.
Overlap
. Since gregarious animals and social human beings tend to use similar
spaces at similar times, movement traces often cluster in space-time, resulting in
overlapping data, clutter, and dense areas with severe information overload.
Spatial dependency and heterogeneity
. Positional fixes along a trajectory feature
highly autocorrelated attributes. This poses challenges around descriptive statis-
tics, sampling granularities and data compression. At the same time, since move-
ment always happens in a potentially heterogeneous space, movement trajectories
are also expected to adopt certain aspects of spatial heterogeneity.
Knowledge of movement is inherently uncertain
. Even at finest tracking granulari-
ties we can't monitor the complete movement trace of a moving object, hence our
discrete view of it must always be uncertain. Uncertainty arises from positional
uncertainty of the localization technologies used and the ignorance about what
happened in between observed fixes.
Most information describing movement is derivative.
Just as geographic informa-
tion is mostly derivative (Goodchild
2001
), so is movement information. Although
more and more sensors claim direct sensing of movement properties such as speed
or orientation, many measurements describing movement are the result of compi-
lation, calculation, and interpretation, mostly hidden from the user. Even if it was
transparent, there are typically many ways of deriving movement descriptors and
the reasons for choosing one over the other are often not transparent.
