Energy Aware SFC Offloading On PDP Switches Based On DQN

Network Function Virtualization decouples network functions from hardware, allowing user service requests to be fulfilled by software running on general-purpose servers. This technology is critical in data center (DC) networks. Network services are delivered through Service Function Chains (SFCs) composed of different Virtual Network Functions (VNFs). Currently, the increasing energy consumption of data centers has drawn significant attention, and the energy wastage caused by SFC processing on data center servers is a critical issue. In this paper, we propose an energy-aware SFC deployment method for heterogeneous platforms, named SFC-EQ, based on Deep Q-Networks (DQN). We introduce a programmable data plane SFC offloading mechanism and use the DQN algorithm to optimize VNF node deployment and offloading strategies. This approach dynamically guides the SFC traffic between servers and P4 switches, reducing the energy waste caused by server power cycling and idle states. Experimental results show that the proposed method significantly reduces energy consumption while improving network performance. Simulation results show that our method reduces average energy consumption and processing time by 22% and 31%, respectively, compared to the greedy algorithm, and improves QoS performance by over 60%.