LoRa is a technology that enables low-energy wireless communications over very long distances. Under these conditions, the simulation of LoRa projects becomes essential. Indeed, it allows on the one hand to virtually model and evaluate the performance of projects in the field and, on the other hand, to anticipate the optimization of the often expensive costs during the implementation of these projects. In this paper, we revisit the issue of simulation for LoRaWAN networks. Our research has led us to conclude that a LoRaWAN simulator’s relevance can be assimilated to the problem of better gateway placement with the best exploitation of their parameters. Our approach consists of first studying some of the existing simulators with the aim of proposing a generic architecture model for such a tool and then, proposing recommendations for the selection of relevant simulators. To validate our model, we used the K-means machine learning technique to solve the problem of bad gateway locations in the LoRaWAN-SIM simulator. From our prototyped K-LoRaWAN-SIM simulator, we observed a gain of 10 gateways compared to the LoRaWAN-SIM simulator considered as our reference prototype. In fact, the results obtained with 25 gateways are close to those obtained with 35 gateways in LoRaWAN-SIM.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

A K-Means Based Approach for Optimal Gateway Deployment in LoRaWAN-SIM

  • Thomas Djotio Ndie,
  • Antoine Junior Tsagmo Denkeng,
  • Karl Jonas,
  • Roblex Nana Tchakouté

摘要

LoRa is a technology that enables low-energy wireless communications over very long distances. Under these conditions, the simulation of LoRa projects becomes essential. Indeed, it allows on the one hand to virtually model and evaluate the performance of projects in the field and, on the other hand, to anticipate the optimization of the often expensive costs during the implementation of these projects. In this paper, we revisit the issue of simulation for LoRaWAN networks. Our research has led us to conclude that a LoRaWAN simulator’s relevance can be assimilated to the problem of better gateway placement with the best exploitation of their parameters. Our approach consists of first studying some of the existing simulators with the aim of proposing a generic architecture model for such a tool and then, proposing recommendations for the selection of relevant simulators. To validate our model, we used the K-means machine learning technique to solve the problem of bad gateway locations in the LoRaWAN-SIM simulator. From our prototyped K-LoRaWAN-SIM simulator, we observed a gain of 10 gateways compared to the LoRaWAN-SIM simulator considered as our reference prototype. In fact, the results obtained with 25 gateways are close to those obtained with 35 gateways in LoRaWAN-SIM.