Join a Panel

Question in: IoT and Edge Computing

Power saving techniques for design IoT Wireless Sensors


Battery life is one of the core concern when designing IoT sensor edge nodes. What are the major design practices in development very power efficient edge nodes?

Wireless Technologies
Wireless Sensor Networks
Energy Harvesting
Low Power Design
Low Power Embedded Systems
James Lai
10 months ago

2 answers


There are basic principles like using lower power electronics - perhaps a low power DSP or 8-bit processor with bare-metal firmware instead of an ARM core with OS; and using low power wireless like Lora, Sigfox, BLE instead of Wifi. However, other key considerations are minimizing the duty cycle or timing when you collect the data (from sensors) and transmit the data. Often I hear 'we need to monitor the system every 5 minutes and alert us when the sensor reading gets out of normal range'. However, this duty cycle willl require lots of processing and data transmission which costs energy budget. Perhaps the sensor collects data every 5 minutes, but only transmits when there is an alert (out of normal range or greater than 5% change from last reading), plus a daily heart-beat to make sure the sensor is still alive. Also, sensing a daily or weekly battery status is important for administrator to know when battery is running low. The other way to increase energy budget is to use energy harvesting (vibration, temp change, solar...) for applictions where they can collect extra energy. An easy way to reduce wireless transmit power is to increase the gain of the antenna with good design as oppposed to a low gain antenna crammed inside an enclosure.

John Murray
10 months ago

The war on current drain, never getting old, always in the news.
Energy harvesting is definively something something to consider. The most unique and advanced design i have seen is from Disney of all places just as the separation of identity management from telephone functions brought up the SIM going to ESIM, i believe multi-source energy harvesting is likely to be prevalent.
Scheduling transmissions, push vs. pull or report, exception conditions vs. nominal transmissions are all part of the equation. it depends on what the thing in the internet of things does.
I am not much in favor of DSP based implementation for most small sensors. if you have a finite number of states or range of states, finite logic based design might work, waveforms can be stored and selected from memory bank.
At the end of the day however, the tradeoff is response time vs battery life.

Alain Briancon
10 months ago

Have some input?