This work focuses on managing upstream traffic in multigateway Low-Power Wide Area Networks (LPWANs) to support efficient traffic from thousands of nodes. As LP-WANs face challenges such as increased collision probability and network saturation due to node densification, our research introduces a distributed and probabilistic traffic control protocol. This protocol aims to effectively manage network traffic, reduce collisions, and mitigate saturation issues, ensuring better performance and scalability in densely populated IoT environments. The protocol enables nodes to dynamically adapt their traffic to meet application needs, such as transmitting a defined number of measurements (K) within a designated time frame. This adjustment remains unaffected by the node count and network topology, focusing instead on the feedback message's destination to the network nodes, which is crucial for dynamically adapting traffic intensity and reducing collisions. We explore two feedback transmission strategies: a synchronous one, where all gateways transmit feedback simultaneously to all nodes, and a round-robin one, where one gateway at a time sends feedback to nodes within its coverage area. Based on simulation results, our evaluated strategies achieve substantial performance improvements over the Baseline LoRaWAN. Specifically, they demonstrate a network lifetime increase of up to 93.12%, a success rate increase of up to 96.34%, and a packet delivery ratio increase of up to 14.97%. These findings highlight significant enhancements in both efficiency and reliability compared to traditional LoRaWAN configurations.