A Collision-Aware Finite State Machine for Adaptive Payload Control in Dense LoRaWAN Networks

Abstract

Dense LoRaWAN deployments are limited by ALOHA-based random access, where packet collisions increase with the number of active end devices and the airtime of each uplink transmission. Adaptive Data Rate (ADR) reduces airtime by adapting physical-layer parameters such as spreading factor and transmission power, but it does not adapt the application payload size, which is an independent and currently underused degree of freedom for congestion mitigation. This paper proposes the Collision-Aware Finite State Machine (CA-FSM), a lightweight controller that regulates an application-layer measurement budget $M$ as a complementary control loop alongside ADR. CA-FSM observes a Channel Occupancy (CO) estimate and a Pure-ALOHA-based Packet Delivery Ratio (PDR) proxy, then selects one of five states to increase, hold, or decrease $M$ and, consequently, the payload length. We evaluate CA-FSM through ns-3 LoRaWAN simulations under device densities of 100, 200, and 400 end nodes, comparing it against a fixed-payload Vanilla-ADR baseline and three static measurement-budget baselines. The ns-3 results show that CA-FSM (labeled FSM-CS) achieves the highest mean aggregate throughput across all tested node counts while maintaining lower channel occupancy than the higher-budget static schemes. The advantage arises because FSM-CS uses large payloads when the channel permits and reduces them only when congestion requires, whereas static schemes must trade throughput for reliability or vice versa.