Information Technology Reference
In-Depth Information
Figure 7. MHs' relative locations in forward and
backward phase example.
MHs on the proposed policy may be kept at a
moderate level.
experimental results
Four of the most prominent experimental results
presented in Karydis et al. (2006) are discussed
in this section. The first experiment, examines the
traffic against MaxHop. The results are illustrated
in Figure 8a. As expected, ML produces the
highest forward traffic in all cases, whereas the
forward traffic of CM, LC and GC are about the
same. Regarding the backward traffic, ML attains
a decreased number of returning results. How-
ever, due to the absence of an efficient backward
routing policy, this advantage is invalidated. The
rest approaches, considerably improve backward
traffic, with CM performing better for MaxHop
greater than seven. From this result it becomes
obvious that, although the backward phase is in
general more demanding for all algorithms, due
to the reduction of backward traffic attained by
CM, LC and GC, the requirement for optimiza-
tion of the forward phase, is fair.
This result can be further clarified by the
results on time of the first and last results, which
are depicted in Figure 8b. As expected, increase
in available MaxHop produces longer times, since
more MHs are examined. In all cases, the increase
in time is far steeper for ML, while CM presents
an advantage over LC and GC.
Next, the impact of query range ε is examined.
In contrast, if their number is larger than CMF ,
then they are all selected. For example, consider
the case in Figure 7. Figure 7a depicts the forward
phase, whereas Figure 7b presents the backward
case. As shown, during the backward phase
some MHs have now relocated. Let CMF be 2.
When P 4 starts propagating the answer-set, it
first selects P 3 , because it belongs to the forward
path. Since this is the only such MH and CMF is
2, it also selects P 5 at random, among the other
reachable MHs.
The nodes that were selected at random in
order to fulfil CMF , are still provided with the
path of the MH that initiated the propagation of
the answer-set (for the previous example, P 5 that
is selected by P 4 , will also know the path from
P 1 to P 4 ). This way, due to mobility, it is possible
for such nodes during the backward phase to find
neighbors that appear in the forwarded path (in the
same example, P 5 finds P 2 that was in the path).
Therefore, the impact of such randomly selected
Figure 8. Traffic, number and time of results vs. MaxHop
Search WWH ::




Custom Search