Multi-Objective Virtual Network Functions Placement and Traffic Routing Problem

Presently, the Internet infrastructure continues to be improved to satisfy the growing demands of users. Installing hardware devices in the network's server center faces many limitations, such as difficulties in device replacement, high investment costs, and operating expenses. The emergence of Network Function Virtualization (NFV) technology has addressed these drawbacks. This technology replaces dedicated hardware with software running in a virtualized environment, offering a more flexible and scalable solution. However, resource allocation is a critical concern in NFV networks. Proper resource allocation can enhance service quality and reduce network deployment costs. This paper investigates the network resource allocation problem to achieve two objectives: i) reduce network deployment costs and ii) decrease the latency of service chains in the network. We propose the modification of Non-dominated Sorting Genetic Algorithm 2, called NSGA2-H, to solve the multi-objective problem. The distinctive feature of the proposed algorithm, compared to the traditional NSGA2, is the use of a selective mutation operator. This mutation occurs on individuals at the first front, resembling a shift operation to enhance the algorithm's exploration capability. Experimental results on real-world datasets have demonstrated the effectiveness of the proposed algorithm.