Coordinator: Francesco Verde
Funds: 8.500,00 €
Brief project description:
The DYNAMIC RESOURCE MANAGEMENT FOR SOFTWARE-CENTRIC 5G NETWORKS (DREAMNET) project aims at defining new methodologies and implementing new tools for the automatic configuration and management of self-adapting geographically-distributed nodes and devices located at the edge of 5G networks to support instantiation, migration, and orchestration of virtualized network resources, in order to provide resilient and customizable services to large numbers of mobile users. The aim of the DREAMNET project is twofold as explained in the following. The first goal of the DREAMNET project is to provide a mathematical framework for modeling migration of virtualized resources across cloud environments, which are characterized by a large bandwidth-delay product (i.e., strictly speaking, the product of the round-trip delay for data connection and the capacity of the underlying wired or wireless physical link). Probably the most basic reason behind the absence of a general treatment of resource allocation for migration of VMs in cloud scenarios is the lack of adequate models. To fill this gap, new models will be developed that link up the performance of the migration process with the main network parameters, e.g., link capacity, jitter, latency, and probability of random losses. Besides predicting the performance of migration, such models will be the basis for optimally and dynamically managing available resources by resorting to network-wide constrained optimization, where the overall VM migration process is modeled by a generalized network utility maximization problem, subject to constraints ensuring provable quality of service (QoS) guarantees. Particular attention will be paid to those solutions of the considered global optimization problems that grant self-adaptation and self-configuration capabilities to the edge nodes/devices.