Enhancing quality of service in cluster-based IoT routing using Optimized Lion Routing with Deep Deterministic Policy Gradient (OLRDDPG)

Quality of Service (QoS) provisioning in cluster-based Internet of Things (IoT) networks is challenging because of constrained energy resources, dynamic topology, congestion, link instability, and increasing network density. Conventional routing and standalone metaheuristic optimization techniques have limited capability to adapt routing decisions to rapidly changing network conditions. This paper proposes an Optimized Lion Routing with Deep Deterministic Policy Gradient (OLRDDPG) framework for adaptive and QoS-aware routing in cluster-based IoT networks. The proposed framework integrates the global exploration capability of the Lion Optimization Algorithm (LOA) with the continuous policy-learning capability of Deep Deterministic Policy Gradient (DDPG). LOA is employed for candidate route generation and global route exploration, while DDPG dynamically selects and optimizes next-hop routing decisions according to residual energy, distance, delay, queue length, and link quality. The performance of OLRDDPG is evaluated using NS-3 under network densities of 50, 100, 150, 200, and 250 nodes against Glowworm Swarm Optimization (GSO), Shuffled Frog Leaping Algorithm (SFLA), Convolutional Lion Routing Optimization (CLRO), and Deep Belief Lion Optimization (DBLO). The evaluation considers energy consumption, end-to-end delay, packet delivery ratio (PDR), routing overhead, and throughput. At 250 nodes, OLRDDPG achieves 77.8 J energy consumption, 201.7 ms end-to-end delay, 86.7% PDR, and 438 kbps throughput. Compared with GSO at the same network density, OLRDDPG reduces energy consumption by approximately 24.1%, reduces end-to-end delay by approximately 29.1%, improves PDR by 14.3percentage points, and increases throughput by approximately 54.2%. These results demonstrate that the integration of LOA-based global exploration and DDPG-based adaptive policy optimization improves routing reliability, energy efficiency, and QoS performance, particularly in dense IoT networks.