Fig 1.
An example of embedding a virtual network request on the physical substrate network, where the virtual network (left) is presented as a request graph describing the resource requirements both on nodes and links; substrate network (right) is presented as a substrate graph describing the physical infrastructure and its resources.
Fig 2.
Example of embedding the common substructure isomorphism of G1 (left) to that of G2 (right).
Fig 3.
A set of links in a graph starting from Nsource rearranged according to bandwidth in a breadth-first manner.
The adjacency links of Nsource are at the first level, the other links of the adjacency nodes of Nsource are at the second level, and so on, until all the links in the whole graph are traversed.
Fig 4.
An example of mapping virtual link Lap1 in Gv (left) on physical link lap1 in R(Gv) (right) because lap1 (the green edge) satisfies all the resource constraints, including its byproduct—physical node np1 (the green node) on the other end.
Fig 5.
An example of mapping the virtual link Lap1 and Lap2 in Gv (left) on the same physical link lap1 in R(Gv) (right) because another adjacent link of node Na (the red edge) does not satisfy the bandwidth constraint, while link lap1 (the green edge) still has sufficient remaining bandwidth to host Lap1 even after hosting Lap2, and the adjacent node of np2 (the dark green edge) has sufficient capacity to meet the demand of virtual node Np1.
Fig 6.
An example of mapping the virtual link Lap1 in Gv (left) on the physical link lap1 in R(Gv) (right) by extending the link to the next hop node np1 (the green node) of node np1’ because the remaining capacity of np1’ does not meet the demand of virtual node Np1.
Fig 7.
Updating the SM positions: (a) SM position updates in the subgroup. The SM marked with a red cross represents the region where VNR requirements are not met and must be pruned; (b) the population is divided into smaller groups with the nsource placed in two positions.
Fig 8.
SMO-based optimization process.
Table 1.
Comparing VNE_MR to VNE_TAGRD and PaCoVNE.
Table 2.
Comparing VNE_MR to VNE_MRWD and EE_CTA.
Fig 9.
Average number of open nodes.
Fig 10.
Average CPU utilization.
Fig 11.
Average number of open links.
Fig 12.
Average bandwidth utilization.
Fig 13.
Average energy consumption.
Fig 14.
Average revenue-cost ratio.
Fig 15.
Average processing time.
Fig 16.
Average number of open nodes.
Fig 17.
Average number of open links.
Fig 18.
Average node utilization.
Fig 19.
Average bandwidth utilization.
Fig 20.
Average energy consumption.
Fig 21.
Average revenue-cost ratio.