Utility Maximization for NFV-Based Tactile Internet

In recent years, there has been significant attention and development in the Tactile Internet (TI) as a new type of network that supports typical applications such as remote healthcare and virtual reality. However, the demand for ultra-low latency and the inflexibility of traditional network architecture have presented serious challenges. In this paper, we propose a utility function for TI based on 5G network function virtualization (NFV) technology to evaluate network performance. The utility function considers the transmission of haptic and video data flows, as well as their latency and cost. We then formulate a network optimization problem with the utility function as the objective, strictly constraining latency and cost. To address this problem, we design two algorithms: the VNF deployment and routing of haptic and video flows (DRHV) algorithm tackles the VNF deployment and path selection issue, while the deep deterministic policy gradient(DDPG)-based traffic scheduling (DTS) algorithm allocates traffic resources and saves costs. Simulation results demonstrate that our algorithms achieve excellent optimization performance under the given constraints, and parameter changes align with the expected relationships and trends.