Multi-user edge service orchestration based on Deep Reinforcement Learning

The fifth generation (5G) of mobile network offers a remarkable degree flexibility to operators, enabling them provide users with effective and tailored services. Software Defined Networking (SDN), Network Function Virtualization (NFV), edge computing have given the operator opportunity easily bring computational capacity support latency-sensitive While 5G standards defined technological architectural frameworks orchestrate services, finding resources management QoS optimization policies is still an open research issue. In this paper, we propose online orchestration methodology for multi-user service. orchestrator goal simultaneously maximize QoS, minimize amount needed. We mathematical formulation compute optimal offline policy derive approach based on model-free Deep Reinforcement Learning (DRL) framework. As novel feature, DRL agent action modeled as parametric combinatorial problem. A multi-objective reward function leads towards choice parameters such model. Our models are built, trained fine-tuned by exploiting real data. Extensive simulations in diverse scenarios show that our produces solutions small gaps ones, both save grant adequate level.