For a given budget, the decision whether to deploy a higher number of units, or to invest to units of better monitoring capabilities can be directly resolved by studying the functions fib and sampling. Whereas the first was thoroughly studied in the previous sections, analyzing the effect of the sampling rate over the performance of the system is a much simpler task . With low sampling rates, GBC becomes proportional to the sum of BC values of the group members (as the number of redundant inspections reduces with the sampling rate). We can therefore, consider a guideline saying that traffic monitors with very low sampling rates can be deployed on the most central nodes in the network, even if it means deploying several monitors on the same node. However, when the overall sampling rate of monitors deployed on each node is relatively high, and then the set of monitored nodes should be chosen wisely using a more rigorous execution of the optimization algorithm.
Notice that BC and GBC based deployments have the same utility when selecting a single monitor as expected. However, GBC based strategy continuously improves the traffic coverage as more monitors are added with the marginal utility of each additional monitor slowly decreasing.
Figure 11 demonstrates the performance of our monitoring method, by showing the percentage of traffic monitored as a function of the number of monitors, for several deployment schemes: (a) Group Between’s, (b) Between’s, and (c) Random deployment. The benefits of the proposed method can clearly be seen from this chart. BC based strategy produces relatively high quality deployments for small number of monitors (less than five). However, when 10 or more monitors need to be located random deployment is on average as effective as choosing the most central intersections.
Moreover, for large numbers of monitors (more than 70-80) random deployment, although the simplest strategy, achieves coverage results that are very similar to choosing the most central intersections. This result may seem surprising but in fact it is absolutely reasonable. Central intersections tend to lay on the arterial roads and usually are quite close to each other. This results in reduced marginal utility of each additional junction joining the deployment. Using the results of Figure 11 the effect of the number of monitors over the overall percentage of traffic coverage can be observed — and used by policy makers in order to decide on the optimal monitoring strategy:
Definition1 .Let us denote the cost of an attack as CATTACK.
Definition2 .LetM(x): Z+ ! [0; 1] be a monotonous function denoting the percentage of traffic that is monitored using x monitoring units5. 5 The function(x) can be extrapolated using simulations, as demonstrated in Figure 11. Note that M(x) is domain dependant and may significantly change for different networks.
Brents Lyons is an IT professional from Exam key. Are you really looking for this s90-07a exam question exam assistance? Move ahead to take the benefit of a00-280 test and pass your exam easily .