Dual mode for battery powered sensors

[joaoabs]

If we consider the scenarios where the sensor nodes are battery powered, you’d agree that all details matter.
I realized that in some cases, the sensor nodes may be working normally, reporting its stuff, but the information being reported is not relevant.
For instance, if you have a PIR sensor in your living room, every time it detects presence, will send that info to the controller no matter if it’s you (or your family), thus drawing precious battery. Basically the idea here would be just to report information to the controller whenever relevant.

Therefore, I thought that maybe we could implement the option of a dual mode for sensor nodes. Something like:

1. Sleep mode. Constant sleeping, where every X minutes, wakes up, contacts the controller and says/asks: “I am here/should I change mode?” If not, continues like that, if yes, changes to mode 2. Even if the sensors are triggered, there is no message sent to the controller, thus saving battery (actually the sensors may be powered off).

2. Waken mode. Active (or sleeping until interrupt generated by sensor) and reporting the corresponding event to the controller. The nodes reaches this state after waking up in the state 1 and the controller told him to change state. In this state, it can receive a message from the controller that orders him to mode 1.

Rational:
I guess most people would only activate their alarm during part of the day (usually by night), so why having the nodes reporting events during all the time (even when they are not relevant)?.
I also think that most of the people wouldn’t bother that the sensor node wouldn’t be immediately active after its request for activation in the controller. If we know that will be active in X minutes, it will be acceptable in most cases.

Use case:
Let’s imagine that the sleeping period is 15 minutes. When going to bed at night, I would activate the alarm, knowing that at the most, in 15m the sensor node will be activated. Would this delay be a real problem? How likely would be a burglar to try to get in in this period?
Considering that could extend significantly the battery duration, I would say it would be a good compromise.

What do you think? Would this make sense?

Thanks for your opinions,
Joao

[mfalkvidd]

I think it makes sense. I started implementing something similar for my office plant monitoring.
It can be done by requesting a variable from the controller, so there is no need to make changes to the MySensors protocol.

The method would be similar to what is being discussed in http://forum.mysensors.org/topic/3580/robot-mower-ignores-the-rain-mysensors-to-the-rescue

[m26872]

It's a good idea which has been around for quite a while I think. I haven't seen many implementations yet though. I agree that it should be very a feasable implementation without any particular controller or MyS support. Anyway, it looks like MyS support for "smart sleep" is coming:
https://github.com/mysensors/Arduino/pull/271

[Dwalt]

I don't know how other controllers handle arming security devices, but this is already possible with Vera. There was some discussion about it here.

[xydix]

Is there any progress in this area?
I guess this is not the same as smart sleep?