RLEAFS: Reinforcement Learning-Based Energy Aware Forwarding Strategy for NDN-Based IoT Networks

Named data networking (NDN) is a recently developed Internet paradigm that satisfies the majority of the Internet of Things (IoT) requirements and may eventually replace the current Internet architecture. The new features introduced by NDN, such as self-certifying contents, receiver-based service, caching, and name-based routing, clearly increase the effectiveness of data transmission. NDN additionally provides lightweight forwarding rules that are appropriate for limited devices. Because of these characteristics, NDN is a very promising for IoT communication. IoT networks, composed of a large number of heterogeneous and resource-constrained devices, benefit from NDN’s ability to handle challenges related to mobility, scalability, and security. However, deploying NDN-based IoT networks raises several issues caused by excessive interest packet forwarding. To address these challenges, we propose Reinforcement Learning-based Energy-Aware Forwarding Strategy (RLEAFS), a novel strategy for NDN-based IoT communications that leverages reinforcement learning to optimize forwarding decisions. Our Strategy integrates Q learning algorithm into path selection procedure, focusing on minimizing energy consumption and extending network lifetime while maintaining efficient data delivery. The proposed RLEAFS Strategy consists of two schemes: one designed to handle the dynamic and complex nature of real-world IoT environments, and another focused on improving the interest forwarding strategy to reduce network overhead. We implemented RLEAFS in ndnSIM to evaluate its performance against state-of-the-art NDN-based IoT forwarding strategies. The results demonstrated that RLEAFS significantly outperforms existing forwarding strategies in terms of energy consumption, network lifetime, data retrieval time, user satisfaction rates, and scalability, proving its effectiveness and robustness for IoT communications.