Greedy virtual network embedding under an exponential cost function

Virtual network embedding has been gaining interest due to the diverse applications of virtual networks. Especially building a cooperative testbed for future network design requires highly efficient virtual network embedding There have been many virtual network embedding algorithms proposed so far. However, these algorithms assume simple cost functions such as minimizing the maximum utilization or maximizing the number of virtual networks embedded in the physical network under given constraints. In this paper, we propose to use a more realistic cost function so that it can take into account diverse aspects of costs in the virtual network embedding Furthermore, we propose a Greedy embedding algorithm that is suitable for the proposed cost function. Through extensive simulation analysis, we show that our proposed Greedy approach performs much better compared to a Random approach. We also show that the performance is similar to a theoretical lower bound for small to medium resource demands.