Virtual Network Embedding Based on Graph Entropy.

Area of Science:

• Computer Science
• Network Engineering
• Graph Theory

Background:

• Embedding virtual networks in large substrate networks is computationally intensive.
• Current methods face challenges in efficiently searching vast resource spaces, even for small virtual networks.
• Identifying substrate network areas with high structural similarity to virtual networks can reduce search complexity.

Purpose of the Study:

• To propose a novel approach for optimizing virtual network embedding in substrate networks.
• To enhance the efficiency and maintain the quality of virtual network embedding.
• To leverage graph entropy for matching substrate and virtual network structures.

Main Methods:

• Dividing the substrate network into substructures based on node importance.
• Calculating the graph entropies of these substructures.
• Comparing the graph entropies of substrate substructures with the virtual network.
• Preferentially embedding virtual networks into substrate substructures with the closest entropy.

Main Results:

• The proposed method significantly improves the efficiency of virtual network embedding.
• The quality of the embedding is maintained without substantial degradation.
• Experimental results validate the effectiveness of the graph entropy-based approach.

Conclusions:

• Graph entropy comparison offers an effective strategy for optimizing virtual network embedding.
• The approach successfully balances efficiency gains with embedding quality preservation.
• This method provides a scalable solution for complex network embedding problems.