Revisiting the problem of optimizing spreading factor allocations in LoRaWAN: From theory to practice

This paper revisits the problem of optimizing LoRa network success probability by proposing an optimized allocation strategy for Spreading Factors (SFs) under both uniform and Gaussian network deployments with a single or multiple gateways. More specifically, we solve the problem of finding the best SF allocations in dense network deployments whose EDs are first assigned with the minimum SF. Theoretical models are developed to quantify the success probability of transmissions, considering the capture effect as well as intra- and inter-SF interference. A mathematical optimization framework is introduced to determine the optimal SF distribution that maximizes the average probability of packet reception. The problem is solved using Mixed Integer Linear Programming (MILP), and then evaluated using simulations. Even though optimal SF allocation strategies have been proposed in the literature, no practical insights have been discovered and no real-world deployments have been considered. To this extent, the practical benefits of using improved or optimal SF settings are discovered in this paper. Simulation results confirm the theoretical findings while they demonstrate an up to 10 percentage points improvements in Packet Reception Ratio (PRR) in the real-world use-case.