Topology-Aware DRL for SFC Placement for Healthcare 4.0

The 6G network is expected to play a pivotal role in advancing healthcare services towards Healthcare 4.0 (H4.0). Network Function Virtualisation (NFV) will play a fundamental role in this transition, as it enables the flexible deployment of H4.0 Virtual Network Functions (VNFs) over heterogeneous infrastructure. The stringent quality of service requirements of H4.0 applications, together with constrained computational and network resources, make optimal VNF placement and routing challenging. In this paper, we study topology-aware service function chain placement and routing in a 5G-orientedNF V-enabled substrate network. We propose a Topology-Aware Proximal Policy Optimization (TopoAware-PPO) framework to optimize the placement of H4.0 VNFs over virtualized infrastructure. The algorithm makes step-by-step decisions based on available resources and attributes of the deployed service, aiming to minimize deployment costs, end-to-end delay, and service level agreement violations. The framework is evaluated using simulation, and the results are compared against common baseline methods: DelayAware, GreedyCost, and Random selection algorithm, making substantial performance improvements of the proposed TopoAware-PPO algorithm.