Information Technology Reference
In-Depth Information
Figure 1 is a screen shot of the user interface illustrating how the run time
component of AntScout works:
Fig. 1.
The run time interface of AntScout
The user may select any node and make it to the source or target. AntScout
will always show the currently best route in between. The interface entry
Path
shows the total time currently needed for the preferred route.
The operator may select any road segment: The interface shows the current
maximum speed and the time for traversing the whole segment. The operator
may change the maximum speed to any value simulating trac jam or acceler-
ation. The interface will respond with the new current traversing time in
Path
and possibly change the preferred route for the current user request.
3.4 Implementation Arrangements of AntScout
Due to the results of previous work at our department (cf. [14], English summary
in [10]), we decided to implement the AntNet version which is the system de-
scribed by Dorigo et al. [9]. For each node, AntNet works with a prepheromone
matrix which is updated by the ants continuously and a pheromone matrix which
is used by the ants for navigation. The prepheromone matrix is based on a sta-
tistical model using dynamic information only such as previous experiences of
ants and flexible time windows. The pheromone matrix integrates static infor-
mation into the values of the prepheromone matrix such as segment lengths.
Both, prepheromone and pheromone matrices serve like a routing table for its
node: There is a line for each possible target and a column for each adjacent
node. A matrix entry (i,j) denotes the quality of using the link to adjacent node
j if the target is node i.
The continuous generation of ants from each possible source node to each
possible target node is the complexity bottleneck of the entire method. This is
