Neuro-Fuzzy ARX Approach for Dynamic RAW Duration Adaptation in Event Triggered Large Scale IEEE 802.11ah Networks

Effective resource management techniques for large heterogeneous IEEE 802.11ah networks with event-driven scenarios are demanded by the expansion of the Internet of Things (IoT). This paper introduces a new system that provides dynamic adaptation of Restricted Access Window (RAW) duration to enhance performance by combining an Auto-Regressive with Exogenous Input (ARX) model with an Adaptive Neuro-Fuzzy Inference System (ANFIS), called the Predictive ANFIS RAW Scheduler (PARS). The PARS method dynamically changes RAW group durations to meet different traffic need, including text, audio, and image. While the ARX model faithfully forecasts group duration, the ANFIS offers adaptive decision-making by using conditions including change, maintain, and alternate based on traffic patterns in real-time networks. Simulations carried out on event-driven scenarios of a large IoT environment comprising 2000 devices show that the suggested framework greatly increases channel usage by 14%, has a 24% decrease in delay, had an 18% reduction in energy consumption, and lowers collision rates by 28% compared to current static allocation strategies. This study provides a scalable, intelligent, and dependable solution for upcoming IEEE 802.11ah implementations, therefore addressing the collision issues in crowded networks. It opens the path for more efficient management of event-driven IoT networks, hence promoting developments in major IoT installations.