Abstract
Virtual Network Functions (VNFs) in cloud servers of Fifth Generation (5G) network systems are responsible for executing offloaded codes from mobile users. Placement of VNFs in the cloud is very complicated to get on-time execution service due to many reasons including users' mobility and resource heterogeneity, which often cause VNF relocations from one data center to another. Minimizing service delay (i.e., maximizing user Quality-of-Experience) for the user applications and the number of VNF relocations are the two main design goals of VNF placement problem; however, they do oppose each other. In this paper, we have formulated the above problem as a Multi-objective Integer Linear Programming (MILP), which is proven to be an NP-hard one. The proposed optimization framework trades-off between the number of VNF relocations and user Quality-of-Experience. We then develop an Artificial Intelligence based meta-heuristic Ant Colony Optimization (ACO) algorithm to achieve sub-optimal placement of VNFs within polynomial time. The performance analysis results, carried out in Cloudsim, depict that the proposed system outperforms the state-of-the-art works significantly in terms of user satisfaction and VNF relocation overhead.