Network planning for long range wide area networks (LoRaWAN) relies heavily on the channel models used to estimate wireless coverage and connectivity. Consequently, the quality of gateway (GW) deployment decisions may be strongly affected by the propagation assumptions adopted during the planning process. Given this motivation, this work investigates how different channel models influence the placement of LoRaWAN GWs,formulating an optimization problem that contrasts stochastic and empirical models with ray-tracing-based models. To this end, we developed a framework that integrates ray tracing (RT) simulators with a discrete-event network simulator. Using this framework to generate LoRaWAN data metrics, we employ an optimization model that determines the optimal GW placement under different channel models, received power constraints, and network scenarios. Our results show that the optimized solution is highly sensitive to the chosen channel model, even when considering the same scenarios with different RT simulators, revealing a clear trade-off between computational cost and the fidelity of the solution to real-world conditions.

翻译：远程广域网（LoRaWAN）的网络规划高度依赖于用于估算无线覆盖与连通性的信道模型。因此，网关部署决策的质量可能受到规划过程中采用的传播假设的显著影响。基于这一动机，本文研究了不同信道模型如何影响LoRaWAN网关的部署，通过构建对比随机模型、经验模型与基于射线追踪模型的优化问题展开分析。为此，我们开发了一个集成射线追踪模拟器与离散事件网络模拟器的框架。利用该框架生成LoRaWAN数据指标后，我们采用优化模型确定不同信道模型、接收功率约束及网络场景下的最优网关部署方案。结果表明，即便在相同场景中使用不同的射线追踪模拟器，优化解对所选信道模型仍表现出高度敏感性，揭示了计算成本与解对真实场景保真度之间的明确权衡关系。