5. Results and discussion
5.1. Network 1 e illustrative example
Fig. 5 shows performance of the fixed sensors (nodes 21 and
32) evaluated for the two objectives e detection likelihood
(DL) and expected time of detection (TD), and the Pareto
fronts for one and two mobile sensors. The non-dominating
solutions were computed through full enumeration of the
solution space, because of the small size of the optimization
problem. Table 1 lists the detection likelihood and expected
time of detection for the solutions on the Pareto front
(Fig. 5).
Results show that the majority of non-dominated solutions
suggest releasing the mobile sensor at nodes 10 and 13 of the
network. This is explainable since it enables the mobile sensors
to explore more of the network when released near the
source at node 10, and to detect events that fixed-sensors fail
to detect at nodes 13 and 23. This reaffirms the optimal sensor
locations suggested by Kessler et al. (1998) (i.e., placing sensors
at nodes 22, 23 and 32).