Abstract
A Virtual Network Function (VNF) is responsible for running codes that have been offloaded by mobile users, and are hosted in the cloud and edge servers of the 5G Internet's hybrid cloud infrastructure. The two key design goals of VNF deployment in a 5G hybrid cloud are to reduce deployment cost and minimize service latency experienced by users (i.e., to maximize their Quality-of-Experiences). However, these two service parameters oppose each other as the reduction of user service latency requires the deployment of a higher number of VNF instances, incurring additional costs. In this work, the aforementioned VNF deployment problem is formulated as a Multi-objective Linear Programming (MOLP) problem that brings a trade-off between the two conflicting objectives. Due to the NP-hardness of the above MOLP framework, we develop an artificial intelligence (AI) driven meta-heuristic Binary Gray Wolf Optimization (BGWO) algorithm for VNF deployment that achieves a near-optimal solution in polynomial time. In comparison to state-of-the-art works, the results of simulated experiments, developed by Python programming version 3.10.2, demonstrate a significant improvement in minimizing VNF deployment costs and maximizing users' QoE up to 30% and 10%, respectively.