In this section, we evaluate our continuous
NN algorithm. Since CPM
significantly outperforms other existing algorithms, it is used as a benchmark
algorithm in our evaluation. The following algorithms have
been implemented by C++.
All experiments were conducted on the PCs with P4 3.2GHz CPU and 2GB memory.
In accordance with the
experimental study of previous work [XMA05,MHP05], the same
spatio-temporal data generator [] is employed. Specifically, this data
generator outputs a set of data points moving on the
road network of Oldenburg, a German city. Every data point is represented
by its location at successive time stamps. Parameter data point agility
indicates the percentage of total data points that report their location updates
at each time stamp. After reaching its destination, a moving data point
randomly selects a new destination and continues moving toward it. Moving speed
may be slow, medium, and fast. Data points with slow speed
move
of the extent of space per time stamp. Medium and fast speed are
and
times faster than slow speed, respectively. Continuous
NN
queries are generated in the similar way. All queries are evaluated at each
time stamp and the length of evaluation is
time stamps.
Table below lists the parameters which may potentially have
an impact on our performance study. In our experiments, all parameters use
default values unless otherwise specified.
Parameter | Range | Default Values |
cell size (![]() |
![]() ![]() ![]() ![]() ![]() ![]() |
![]() |
number of NNs (![]() |
![]() ![]() ![]() ![]() ![]() |
![]() |
number of data points (![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
![]() |
number of queries (![]() |
![]() ![]() ![]() ![]() ![]() ![]() |
![]() |
data point agility | ![]() ![]() ![]() ![]() ![]() |
![]() |
query agility | ![]() ![]() ![]() ![]() ![]() |
![]() |
moving speed | slow, medium, fast | medium |