Academy for LoRaWAN®

Radio planning is a science, of course. But it’s also an art. Use it to meet the following objectives:

• Predict network performance prior to deployment
• Achieve the necessary geographic coverage (both outdoors and indoors, as appropriate)
• Achieve the desired signal strength
• Achieve a desired level of resilience
• Realize LoRaWAN geolocation services by making sure LoRaWAN packets are received by at least three gateways
• Optimize and select gateway locations
• Minimize network costs

Radio planning is a science in that it requires an understanding of physics. To be successful, you need to know how radio signals propagate, diffract, and attenuate. At the same time, it is an art in that predicting performance is extremely challenging. Environments are constantly changing, for example. Buildings are frequently erected, renovated, or removed, and actual height data is expensive to acquire. Terrain data is only approximate and can vary in detail. Interference is also quite variable. Then, there is the fact that, since LoRaWAN operates in a license-free frequency band, there is nothing to stop others from deploying gateways and devices on the same channels that you may be planning to use.

With this combination of science and art in mind, it is important to know that:

• Coverage and signal strength predictions derived from radio planning are—and always will be—approximate at best.
• Radio planning models use simplifications of the real world to generate their results. This means that the models will not be completely accurate.
• It is wise to build some safety margins into your plan to compensate for unexpected signal losses. Do not assume that signals with a strength of -120dBm will always be received by the end device or gateways, rather, plan for a signal strength of -100dBm.
• There is a decreasing return in the radio planning exercise. That is, while it may be possible to build a nearly- perfect snapshot model with 3D rendering of buildings and an understanding of the materials used, it is not necessary to go to those lengths. Indeed, an approximation based on reasonable parameters will generally give a practical and cost-effective answer.