WSEAS Transactions on Communications
Print ISSN: 1109-2742, E-ISSN: 2224-2864
Volume 15, 2016
Modification of an Energy-Efficient Virtual Network Mapping Method for a Load-Dependent Power Consumption Model
Authors: , ,
Abstract: This paper tackles an energy efficient virtual network mapping problem where virtual nodes and links in a given virtual network have to be mapped to physical nodes and paths in a physical network so that the total power consumption associated with the mapping is minimized. The conventional method assumes that power consumption of a physical node is constant regardless of its load (constant power consumption model), and successfully reduces the total power consumption by preferentially mapping virtual nodes and links to active (used) physical nodes and paths passing only active physical nodes. However, power consumption of a physical node will become variable dependent on its load (variable power consumption model) in the near future, and the conventional method may not reduce the total power consumption because its active-node-first policy can cause large additional power consumption under the variable power consumption model. In this paper, we try to minimize the total power consumption under the variable power consumption model. In order to achieve this, we modify the conventional method so that it adopts the minimum-additional-power-consumption-first policy. The modified method calculates the actual additional power consumption associated with node and link mapping, and preferentially assigns virtual nodes and links to physical nodes and paths so that the actual additional power consumption is minimized. Simulation results clarify that the modified method can 4-40% lower total power consumption than the conventional method under the variable power consumption model.
Search Articles
Keywords: Network virtualization, Virtual network, Virtual network mapping problem, Energy efficiency, load-dependent power consumption model, heuristic algorithm
Pages: 240-250
WSEAS Transactions on Communications, ISSN / E-ISSN: 1109-2742 / 2224-2864, Volume 15, 2016, Art. #28