The IEEE 802.11 ah standard is the prominent protocol for medium range communication in the domain of Internet of Things (IoT), facilitating interaction between wireless sensor nodes and access points. Supporting up to approximately 8192 nodes connected to a single access point through single hop strategies, IEEE 802.11 ah employs CSMA/CA for the authentication and association process of sensor nodes with access point. However, the standard encounters challenges such as significant collisions and prolonged network setup times when attempting to associate an excessive number of wireless sensor nodes with a single access point, primarily due to authentication thresholds set at intervals. In this research, we propose a dynamic adjustment of the authentication threshold based on network congestion status. Learning parameters including the number of requests served successfully by the access point in the past, present and a congestion threshold are leveraged to adapt to current network congestion levels. Through evaluation using the LILD method in the ns-3 simulator, our approach demonstrates considerable enhancements in terms of collision reduction and network setup efficiency. This research endeavours to promote responsible network management practices for improving the performance and reliability of IEEE 802.11 ah based IoT deployments.