On "Building A Better IoT"

Since it does not look like there is a discussion thread on the series, here it is.

In the series @wendell goes trough a lot of topic, I try to summarize them:

  • Security in term of actual hack

  • Security in term of support

  • Security in term of who you have to thrush

  • Functionality in term of actual usefulness of the service

  • Functionality in term of easiness to use

  • Functionality in term of easiness to expand

  • Functionality in term of horizontal integration

  • Robustness

  • Reliability

(i would divide robustness from reliability; you may have a system very hard to bring offline, but has a lot of quirks; or vice-versa, the classic “authoritative server” make sure all is aligned and well, but suffer the classic single point of failure)

I am also designing in those day something similar, and put some though about the architecture of the system from a very high point of view.

I would love to talk abuot and expand on every single point, but I would like to keep this main post to collect/fix the various discussion point collected, and also have a cold take on the issue.

Meanwhile i would love your input

Hi everyone, Hi lesto.
First post here on the forums (some time channel subscriber)…

Are there any other topics at least marginally related to the IoT discussion here? Would be nice to have them linked here.

Anyways, just saw Wendell’s video on IoT…
I hacked my dumb AirCon split sometime ago along with planing some home automation but the whole project never came to life as it kinda reached the same pitfalls Wendell described in the video, somewhat messy solutions on the wireless end, no standard or concise good practices for a IoT wired network (tho borrowing the CAN approach from automotive seems nice).

My initial two cents are the openhab project, lots of repos on github… And some findings that a lot of the temperature control on split AC units seems really dumb and naive when controlling the compressor and using data from their temperature sensors (seems like units with inverters work better on that front with some PID).

For now that’s about it, I’ll probably remember some other points now that I’m revisiting this subject, its a nice one.

1 Like

I’d tip my hat towards the openHAB/Home Assistant crowd here on this one. Both systems would work well 100% offline when built out correctly. They tie in nicely with video systems as well and would be a great fit for offline, IoT.

1 Like

I remembered fiddling around with this project http://mynewt.apache.org/
It’s an RTOS designed for IoT stuff… I think it even have it’s own drivers and abstractions to handle wireless hardware. Probl’y worth to take a look…

After reading through the docs in that link it seems more geared towards Bluetooth and it’s BLE stack than anything else. Unless I missed something… Not sure that really provides much to branch out to the vast IoT radios and standards that are out there.

@marsh an RTOS is not a good solution, a standard API, probably in form of RFC, is.
This because otherwise you are just creating a new vendor lock; yes it is not a proper lock, but I think I make the point of the pro and const.

A standard API is something manufacturer, especially smaller ones, may be tempted to follow as it frees them for developing other part of the system like the controller or other sensors, and make possible even for professional installation to know what will and will not work.

Is basically the standardization that made industrialization possible all over again.

il take a look into it, I saw Microft a while ago and was promising, and now with mozilla Voice/ DeepSpeech we have high quality offline solution for TTS and STT

1 Like

I think we’re missing an implied topic here, one that’s probably implied only because it’s an integral part part of moving away from mainstream IoT.

I realize this is basically my dead horse, too, but I’m not done beating it yet.


I do not know how or why the data on when your lightbulbs are on, at what color temp, or when your laundry machines or when your fridge is empty, or what you watch on your smart TV, is valuable (but I have a few ideas for some of these), but I have no interest in allowing that to subsidize my home automation and I think many of us here agree. I also think it’s an important part of the discussion.

We should probably be paying attention to which products make “telemetry” or similar network requests and logging those products someplace public, so we can all make more educated decisions about the individual products within our IoT.

Taking this stuff offline / divesting it of internet access certainly is an option, but it’s not necessarily the only or the best option. We have to admit that that depends on context: which features we require, what integrations, and so forth, don’t we?

I would say it its included in trust.
Do you trust that company/project/developers to release safe and non breaking updates? That their tool is secure? that they wont sniff your lan? and so on