Information Technology Reference
In-Depth Information
4.2 Comparison of the Three Protection Schemes
In this section we evaluate and compare results of our heuristic algorithm for path,
link and region protection schemes for the U.S. Long-Distance Network. Here each
link was assumed to have 32 channels; 30 demand pairs formed each scenario.
4.2.1 Network Resource Utilization Ratio
Fig. 5 shows the number of links as a function of percentage of link utilization ratio
for all three protection schemes. Results from Fig. 5 prove that the smaller the indi-
vidually protected area is, more network resources are necessary to protect a connec-
tion. Fig. 6 illustrates the relative network resource utilization ratio per connection. It
shows that the region protection is almost as good as the path protection, giving about
60 % of the respective value for link protection. It is because the total number of links
used by backup paths gets bigger when the size of individually protected area de-
creases. The region protection model leads to only insignificantly worse results re-
garding relative network resource utilization ratio per connection than those obtained
by using the best model (i.e. path restoration), as shown in Table 2.
200
5,00%
160
4,00%
120
3,00%
80
2,00%
40
1,00%
0
0,00%
1
26
51
76
4
5
6
7
8
number of channels per link
link utilization [%]
8 demands per logical topology
12 demands per logical topology
path protection
region protection
link protection
Fig. 4.
Rate of additional resource utiliza-
tion obtained by heuristic algorithm, com-
pared to the optimal results of CPLEX
Fig. 5.
Number of links as a function of
percentage of link utilization ratio for all
three protection schemes
Table 2.
Link capacity utilization for three protection schemes
path protection
region protection
link protection
network resource utilization [%]
18,30
20,01
30,56
95% confidence interval [%]
1,05
1,54
2,38
4.2.2 Restoration Times
Fig. 7 shows the cumulative distribution function of restoration times for all three
protection schemes while Table 3 the respective average values.
Results show that the values of restoration times get smaller while decreasing the
size of individually protected area. They represent values of time needed to restore a
connection after a failure of a network component, according to the protocol taken
from [5]. The lower the values of restoration time are, the smaller amount of data is
lost within the period of restoration.
