Home automation - sensor network design

[florint]

I want to design my own home automation system and I want to get "the big picture" on the implementation. I've seen lots of projects (garage door controllers, ambient sensors, etc) but the problem is that each of these is encapsulated in their own world. I want to take a step back and understand how other people integrate these components. Here are some of my thoughts:

- I presume that there should be one centralized "brain" - maybe OpenHab (running on a Raspberry Pi) where all the information from the sensors is fed in and where you can make decisions based on that data. (If you tried multiple home automation software frameworks, do you recommend anything else over OpenHab?)
- If we use RFM69 to send the data from the sensors there must be a receiver connected to the Pi takes that data and makes it available for consumption for OpenHab. What physical interface board do you use? Is it plug-n-play for Pi/OpenHab?
- Do people usually multiplex data from a few sensors on one Moteino (RFM69)? Suppose I have a room where I have a bunch of sensors (humidity, temperature, motion, etc). Can I multiplex their data on a single RF link (or ethernet wire)? What hardware & software do you use to do that?
- If the above questions were about feeding-in data from the sensors ("inputs" from OpenHab's perspective), what about outputs? If I want to trigger actions (turn on lights, siren, etc) how do you do it from Pi?  Is it possible to multiplex to different "outputs" on the same RF link? What hardware do you need to demultiplex it at the destination? What hardware do you need to plug into the Pi?
- Did anybody tried to integrate their security cameras with the rest of the system? Most of the pre-packaged commercial solutions for security are garbage. I want a completely local solution (no cloud storage) for my video feeds and ability to process the live video stream from the camera (for running image recognition for example - yes, I know that I need more powerful hardware than a Pi for that :-) ). Did you find any affordable cameras ($100 or less) that support ONVIF and can work with open source recording software?

Thanks in advance!

Florin

[HeneryH]

https://lowpowerlab.com/guide/gateway/ has tackled this already except the video surveillance system.

Video survelance (especially event detection and triggering) is too much for a pi.  I use a commercial product that runs on a PC for a $50 license.  http://blueirissoftware.com/  Cameras can be had for very cheap but their night time video quality varies significantly and you usually get crappy quality for cheaper cameras.   Video camera choice discussion is probably better had on a forum dedicated to that.

[Felix]

BTW The next version of Gateway will have much more support for wifi nodes and other non-RFM69 based nodes.

[TomWS]

BTW The next version of Gateway will have much more support for wifi nodes and other non-RFM69 based nodes.

BLE?

[Felix]

BLE?

The Gateway app is "wireless hardware agnostic", so beyond serial it does not care as long as the basic protocol is followed (ID:space_delimited_tokens).
For BLE you can just it up to serial and use it that way maybe? Or is this in addition to everything else? In that case maybe supporting multiple serial ports would be an option - right now only 1 serial is supported.

I actually really like the idea of multiple serials. It would open up the possibility of running a mix of LoRa/RFM69/BLE or other technologies (even non-wireless?), and also possibly a mix of frequencies (ex 2 hardware radio/gateways with different frequencies, only 1 Gateway interface).

[TomWS]

I was thinking in terms of the myriad of BLE profiles that existing sensor devices have, eg, Heart Rate Monitor, Blood Pressure Monitor, iBeacons, etc.

[Felix]

BLE is going to be short range with poor penetration, maybe good for an open space short range application.

[florint]

Sometimes it's useful to have a short range sensor, it can help localize a person in the house (kid is downstairs in the family room) if you have a smart watch or a phone.

[Felix]

There is a gentleman in Australia developing an application for monitoring elderly people. He is using a MoteinoM0 with RFM69 radio modules. He tried other things and concluded they are not working well, and converged to sub-ghz technology to overcome the hurdles of 2.4ghz. This story repeats again and again, it's how I started Moteino. I tried everything else and WiFi was the natural first choice but it just doesn't cut it. It may work for very short distance and very few obstacles.