Online power aware coordinated virtual network embedding with 5G delay constraint
Solving virtual network embedding problem with delay constraint is a key challenge to realize network virtualization for current and future 5G core networks. It is an NP-Hard problem, composed of two sub-problems, one for virtual node embedding, and another one for virtual edges embedding, usually s...
| Autores: | , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2018 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/123775 |
| Acceso en línea: | https://hdl.handle.net/2117/123775 https://dx.doi.org/10.1016/j.jnca.2018.10.005 |
| Access Level: | acceso abierto |
| Palabra clave: | Mobile communication systems Virtual network embedding Core network virtualization Power consumption Coordinated embedding Delay Comunicacions mòbils, Sistemes de Àrees temàtiques de la UPC::Enginyeria de la telecomunicació::Radiocomunicació i exploració electromagnètica::Comunicacions mòbils |
| Sumario: | Solving virtual network embedding problem with delay constraint is a key challenge to realize network virtualization for current and future 5G core networks. It is an NP-Hard problem, composed of two sub-problems, one for virtual node embedding, and another one for virtual edges embedding, usually solved separately or with a certain level of coordination, which in general could result on rejecting some virtualization requests. Therefore, the main contributions of this paper focused on introducing an online power aware algorithm to solve the virtual network embedding problem using less resources and less power consumption, while considering end-to-end delay as a main embedding constraint. The new algorithm minimizes the overall power of the physical network through efficiently maximizing the utilization of the active infrastructure resources and putting into sleeping mode all non-active ones. Evaluations of the proposed algorithm conducted against the state of art algorithms, and simulation results showed that, when end-to-end delay was not included the proposed online algorithm managed to reduce the total power consumption of the physical network by 23% lower than the online energy aware with dynamic demands VNE algorithm, EAD-VNE. However, when end-to-end delay was included, it significantly influenced the whole embedding process and resulted on reducing the average acceptance ratios by 16% compared to the cases without delay. |
|---|