Low-power application design through radio focus

Even with a very low-power wireless device, wireless connection usage still has a major impact on your product. Total power consumption and battery life are impacted by your technology choice (LTE-M or Narrowband Internet of Things (NB-IoT)), transport and application protocols you use, and how often you use the link and the amount of data you transfer each time.

An important decision point for your application is how frequently it should communicate with the cloud. Every connection to the cloud spends time and power on overhead. For example, tasks the Long-Term Evolution (LTE) modem must do without providing real value for your application. Every time you send or receive data, some functions with associated overheads are triggered. The overheads are:

The nRF9160 System in Package (SiP) was designed from the ground up with the first two factors listed above in mind. However, your choices on the last two factors also have a major power consumption impact and need your attention.

LTE offers direct connection to the Internet in a global, secure, and mostly reliable way. This inherently leads to more overhead compared to simpler, local networks. See the LPWANs bridge the IoT divide blog for more context around cellular positions in Low-power Wide Area Network (LPWAN)s. By making correct and relevant decisions early in your product design, you can avoid pitfalls and inferior decisions.

If the value in your application includes real-time and low latency delivery of data, that can justify the cost of frequently setting up a connection to transfer small amounts of data. However, frequent sensing of data may not be viable due to the incurred battery size or cost, and the cost of ownership from overhead in both power consumption and data usage.

The key point is to value the data you send versus the cost of sending it. If the value of your application does not always rely on low latency delivery, you can transform your power budget and extend the battery lifetime by leveraging the local processing capacity in nRF9160. Local processing can be anything from a simple collection of data over time and bulk transfers, to local processing or decision making all the way to local machine learning or AI.

Internet protocols were originally made for transferring large amounts of data between computers, so the cost per bit transferred is much lower if you can send larger amounts each time you take the cost of connecting to the cloud, spreading the overhead cost on more bits.

Most IoT applications have a mix of critical or low latency data like alerts or alarms that needs to be sent immediately, and data collected and processed over time that can be sent in larger bulks. If the historical data can be processed locally to reduce the size, then you can also increase the value of each bit as well as lengthening the intervals between transfers. This is the key to achieving a low-power wireless product.

The first and most important decision you need to make is how to use your wireless link:

These considerations form a foundation for the choices you need to make in the following sections.