Eisenhower Matrix-Based Machine Learning Framework for Message Prioritization in LoRaWAN

Long range wide area network (LoRaWAN) is a prominent low-power wide-area network (LPWAN) technology, widely adopted in the Internet-of-Things (IoT) applications for its energy efficiency and long-range communication. However, the ALOHA-based access mechanism in LoRaWAN constrains message prioritization, making it challenging to address urgent communications effectively. We address this limitation by proposing a dynamic message prioritization framework that integrates an Eisenhower matrix (EHM)-inspired rule-based system with a multioutput neural network (MONN). Our model classifies messages into four priority levels based on LoRa features and real-time network conditions while simultaneously predicting received power (PRx) to enhance transmission reliability. By leveraging SHapley Additive exPlanations (SHAP), we ensure model transparency by analyzing feature contributions in both classification and regression tasks. Experimental results show that our approach achieves 100% classification accuracy for message prioritization and an ${R}^{{2}}$ score of 0.98 for PRx regression. These findings demonstrate that the proposed framework effectively prioritizes messages based on their criticality, enhancing LoRaWAN’s adaptability in dynamic network environments.