Intelligent Slice-Aware Routing with Dynamic Coverage Optimization in 5G and Beyond 5G Internet of Things

Abstract

5G and Beyond 5G (B5G) networks are enabling massive Internet of Things (IoT) connectivity. Traditional routing
methods used in such networks operate independently on the real-time coverage and network slicing requirements, which leads to
higher latency, packet loss, and energy inefficiency. This study proposes a Coverage-Conditioned Slice-Aware Routing Framework
(CC-SARF) that optimally solves routing decisions and coverage reliability in dynamic 5G and B5G IoT systems. The suggested
framework incorporates real-time network monitoring to gather cross-layer terms that characterize node mobility, link quality,
traffic load, and energy levels. These observations are processed by Graph Transformer to learn spatial-temporal relationships to
forecast coverage confidence and connection reliability. A slice-aware Deep Reinforcement Learning agent identifies the most
optimal next-hop nodes to meet the QoS criteria. Delivery performance is then fed back into the learning loop that allows constant
adaptation and better routing decisions within the changing network conditions. Simulation results demonstrate that CC-SARF
significantly improves routing stability and QoS satisfaction compared with conventional methods, achieving minimal latency of
0.211 ms, higher packet delivery ratio of 99.60%, and improved energy consumption of 0.317 J in dense and dynamic IoT
deployments. Thus, the approach provides a scalable and intelligent routing solution for next-generation 5G and B5G IoT networks.