Leveraging Digital Twins in Software-Defined IoT Networks: A Smart Agriculture Case Study

Software-Defined Networking (SDN) has proven to be a promising and prosperous direction for the Internet of Things (IoT). Numerous SDN-based architectures have been specifically designed for IoT environments and have demonstrated their efficiency. However, these architectures are often tailored to meet the requirements of a particular application or service. In such cases, the SDN controller makes decisions according to a management policy, either predefined or externally imposed, resulting in configurations that may meet the performance requirements of a specific service but often fail to adapt to other needs due to the lack of autonomous decision-making capabilities. In this paper, we explore the use of a decision-making agent to enhance the autonomy of the SDN controller. This agent relies on a Network Digital Twin, which is responsible for simulating and evaluating the physical network environment. By leveraging these evaluations, the agent assists the controller in making more appropriate and adaptive decisions, enabling it to dynamically adjust its control strategies according to varying application needs. This integration allows for more intelligent and optimal decision-making based on concrete and contextualized network evaluations. Here, we evaluate the performance of this agent in the context of Low-Power and Lossy Networks (LLNs) for a Smart Agriculture use case. Our results show that the agent meets the specific requirements of each service, including high data reliability, low-latency communication, and energy efficiency, demonstrating the benefits of combining SDN with a Digital Twin for autonomous and adaptive network control.