New Article at Future Generation Computer Systems

Future Generation Computer Systems, 133, 124-140, 2022.

Article “A reconfigurable resource management framework for fog environments”, N. Godinho, H. Silva, M. Curado, and L. Paquete, available at Future Generation Computer Systems.

Abstract:

Fog computing emerged to ease the load of resource-constrained devices on the Internet. It provides services and computational devices closer to the users to reduce latency and improve quality of service. For this paradigm to be used in practice, certain optimization tasks need to be solved: to allocate the resources that are available to the users and to perform service placement and routing on the communications executed by them. In this paper, we develop a framework that allows users to offload services and perform communications in a Fog environment. For this, we propose a Mixed Integer Linear Programming (MILP) formulation and a heuristic to map Virtual Networks (VNs) into the substrate network, maximizing fairness of energy and bandwidth usage on the system. Moreover, we propose a MILP formulation and a heuristic to route and offload services of an application in each VN, minimizing energy and latency. We extend both heuristics to be used in dynamic settings with increasing number of users and applications and consider a reconfiguration approach when nodes or links fail in the substrate network. To evaluate the proposed approaches, we use the YAFS simulator and several instances of VNs and services with both randomly generated values and referenced parameters. We observed that the heuristics are able to obtain results close to the optimal value in most settings.