Offline and online power aware resource allocation algorithms with migration and delay constraints

In order to handle advanced mobile broadband services and Internet of Things (IoT), future Internet and 5G networks are expected to leverage the use of network virtualization, be much faster, have greater capacities, provide lower latencies, and significantly be power efficient than current mobile technologies. Therefore, this paper proposes three power aware algorithms for offline, online, and migration applications, solving the resource allocation problem within the frameworks of network function virtualization (NFV) environments in fractions of a second. The proposed algorithms target minimizing the total costs and power consumptions in the physical network through sufficiently allocating the least physical resources to host the demands of the virtual network services, and put into saving mode all other not utilized physical components. Simulations and evaluations of the offline algorithm compared to the state of-art resulted on lower total costs by 32%. In addition to that, the online algorithm was tested through four different experiments, and the results argued that the overall power consumption of the physical network was highly dependent on the demands' lifetimes, and the strictness of the required end-to-end delay. Regarding migrations during online, the results concluded that the proposed algorithms would be most effective when applied for maintenance and emergency conditions. (C) 2019 Elsevier B.V. All rights reserved.