Which IoT platform should I use for low-energy temperature sensors to be powered by battery?

Question

I'd like to setup two temperature detectors: one outside and one inside my house, in order to compare temperature and act on the difference (opening/closing the window for example).

The issue I'm facing is which platform to take. My initial thought was to go with the Photon, but the price is quite high when I can go for a Raspberry Pi Zero for way much. On the other hand, the Pi Zero requires a lot of power, and since I plan to place one sensor outside, I was hoping to "place it and forget it" at least for a few months, with some AA batteries (one? two?) on it.

So I'm asking for your help. I'm open to other platform to implement my plan. Here's what I'm looking for in this platform:

1. Wi-Fi capable (or some transmission for the outside remove, and Wi-Fi for the inside)
2. Can live weeks, months using simple AAs
3. Not expensive. It's just a small side project.

Answer 1

ESP8266 can be a very good candidate for your project.

For increasing the battery life with ESP8226 you can follow this Open Home Automation article. Ways to save power save highlighted by this article are:

• make measurements at intervals (every 10 minutes for example), and sleep the rest of the time.
• getting rid of the LED drops consumption to 77 uA, or 0.077 mA. This means that the same battery can last for … 3.7 years!

And certainly ESP8266 is inexpensive.

ESP8226 key features:

• low-cot Wi-Fi chip with full TCP/IP stack and MCU (microcontroller unit)
• IEEE 802.11 b/g/n Wi-Fi
• 16 GPIO pins, SPI, I²C, UART on dedicated pins

More reference material at espressif.com.

Answer 2

I see at least 3 choices to make in your system design.

RF Protocol WiFi is not very energy efficient. You can mitigate this by only sending readings infrequently (measure at 1 min, transmit at 20 min). BLE or similar might be better, but you need to trade range and parts cost if you opt for something a little less commodity. If its personal domestic use, much over 2 years is probably not worth much extra optimisation.

Inside Unit You probably have different power constraints for this unit, but you don't specify. Critically, it doesn't need to be the same platform as the outdoor unit, but it doesn't sound like you need an SBC running linux here. How you want to develop your stack is maybe the driving factor here (as well as familiarity).

Outdoor Unit Currently you have a low feature requirement - just a digital interface to your thermometer. You might want a resolution of 0.25 °C or better to allow some scope in how you use it. Certainly you need a sleep mode, but the choice is quite wide. Development environment and ease of use might be as important a factor as price. You have no compute payload worth worrying about over the communication protocol, so a low clock frequency makes sense.

Other factors you might want to consider are how likely you are to expand this in the future, this might affect your choice of board (for example if you want to add a display/control unit indoors).

It's fairly clear that the only important choice here is that the outdoor/battery unit should be a micro-controller with sleep (and RF), rather than a full linux platform. Newer platforms are likely to offer better energy efficiency, but might be sufficiently expensive as to offset the benefit in this use case.