Just yesterday, it occurred to me that an IoT thermostat would be nice (if and only if it doesn't mean that a server failure results in me freezing). Right now, I have the choice of heating the house all day (thus wasting heat = energy = money while I'm not home), or coming back to a cold house, turning up the thermostat and then having to wait 1-2 hours until the room is pleasantly warm. (My apartment is understaffed in the heater department and I cannot change that easily since I'm renting.)
Those have existed for years, apartment buildings are just usually cheap and lazy and don't install them.
I've got a Honeywell model [1] I can set from anywhere via an app. It also has scheduling it falls back to if I haven't overridden it via the app or faceplate. My winter weekday schedule: 66 at 6am, 62 at 9am, 65 at 4pm, 62 at 10pm.
It's quite easy and cheap to make a wifi-connected thermostat yourself. I did it, it's great.
I made a web interface so that I can control it with my phone when I'm not at home.
It is based off an ESP8266 placed in the boiler which controls a relay to tell the boiler whether to heat or not, and an ESP8266 in my office that has a temperature probe and a rudimentary user interface, and runs a web server to allow remote temperature inputs, and controls the ESP8266-relay over WiFi.
EDIT: Failure of the system could result in you freezing, but that's true of anything that controls your heating, and at least this is a device you understand are in control of, and it doesn't rely on anything outside your house. If it fails and you're freezing, just short the terminals of the relay while you debug it.
In cold climates, the problem with failed thermostats isn't frozen people, it's frozen pipes, which can happen when the people aren't home.
An easy way to insure against such failure is to put your homemade thermostat in parallel with an old fashioned thermostat which is set to a constant low temperature.
A seemingly common safety is to wire the custom thermostat in parallel with a simple commercial one so that the commercial one kicks in if the temperature gets too low.
If you are in Europe, you can also buy the ESP8266 from Olimex with relay, box and eventual lipo plug. You then just have to flash your code on it. Only downside is they are not properly shielded electromagnetically so they can only be sold as Kit for experienced users (as certification costs would dominate the price). Also, if your house happens to burn, you can be pretty sure the insurance companies will pin it on your uncertified DIY appliances.
I don't know of actual precedent, but there are numerous examples of lipo house fires (which aren't fully covered). And semi related personal one-datapoint anecdote I've experienced water damage and robberies in the past and they were a lot more prompt to try to find loopholes to not have to pay. I can't afford to set a precedent, so I stick to fused low voltage off the shelf products.
Use an off the shelf wireless plug in switch and a Telldus transmitter then you don't do any high voltage or high current wiring at least if you are controlling a standalone heater.
Thermostats with scheduling capabilities have existed for a long time and they don't need the internet to do it. Unless you're referring to when you're out of the house when you would normally be home?
Are there not thermostats on the market that let you program schedules? If you do want something a bit more connected, thermostats in the Lyric line from Honeywell[1] can give you things like scheduling and geofencing (e.g. "oh, you're on the way home, let me start cooling...") Of course, a geofence isn't going to function if you're offline, but the thing should definitely remember a time schedule to use as a backup when connectivity is lost.
When I go on vacation for a week there is no way to say "go to max energy savings while saving my pipes mode until just before I get back." I can reprogram the entire week, but then I lose what is the correct programming for normal until I reprogram it. Worse my house is uncomfortable when I get back - usually when I'm tired and want to jump into bed asap (if I know I'll get home about noon I'll look for a tourist thing to kill some time long before I get home).
I don't always have the same schedule. When I'm home sick I don't want to reprogram it. Some weekends I'm at home all day, some I'm out all day - which program is right?
IoT thermostats also through privacy and reliability concerns in. The old fashioned mercury thermostats from the 1950s still work as good as new. While your IoT thermostat work next year - some do not.
All the above is made even harder because my house is headed by an air source heat pump: if I'm returning at 9pm it might or might not be better to start heating my house at 11am when the day is warmest. The algorithm depends on the weather which is best predicted the day of.
My Honeywell t-stat has a schedule and a hold function which can be either timed or permanent. If you want to lock a temperature, you can tap the hold button a couple times to set it to "permanent hold" and set the temperature to whatever you want.
thermostats are trivial to change if you have any electronics or handy skills, just don't put new holes in the wall that aren't hidden by the mount plate if you're renting. It's basically like replacing the shower head while renting.
I am honestly not sure why this is getting down voted. The nest thermostat does continue to work as intended when offline. Of course you won't be able to use the app while you are away to control it.
You get timer-based units that reduce the temperature at night (optionally also when you are at work). They cover a good chunk of the usecase and will give good amounts of savings. No Internet, mobile app or machine learning needed.
Sadly, that all got balled up almost immediately. The offer came with a mailer for exhausted cartridges. I put them in the mailer, put the mailer in the box (rural route). My spouse mistook it for incoming mail, ripped it open, said "Why is HP sending us empty printer cartridges?"
Fortunately HP had a button to click on the web site to get another mailer mailed to me. I clicked, got the mailer (which was entirely different from the one that came with the offer) and sent the cartridges off.
That was weeks ago. No new ink. My credit card continues to be debited monthly, but the printer is defunct and now what?
These things are so, so prone to getting in a wad and leaving you frustrated. More work than actually buying new cartridges as needed.
I said "no more" to printer cartridges over a decade ago. I rarely print things, and when I do, it's usually black ink on white paper.
I was always running out of black ink, or, because of infrequent usage, my inkjet cartridge/head would be clogged. I was so sick of it. I just wanted to hit "print", and have the printer spit out the page - like I could at my employer.
What did my employer use?
A laser printer. So I started looking into it. The last time I had, such printers were still expensive, but they had come down in price. But even better, because of this, businesses and others were getting rid of their "old and slow" b/w laser printers and upgrading to faster color ones. I eventually found my first laser printer at a computer surplus outlet.
It was an HP Laserjet 6M - I paid $100.00 for it, but it didn't have a cartridge. It was a gamble, but I asked the place if I could return it if it didn't work, they said "sure".
I went to a local ink cartridge and printer ribbon store I knew about (they were an old-time fixture here, they even sold typewriter ribbons! Sadly, they no longer exist), and told them my dilemma - I needed a cartridge to test with, I didn't know if the printer worked. They gave me a mostly used cartridge from their refill pile, and told me to bring it back and they'd refill it for me if it worked! Grateful, I took it home and plugged it in.
It worked perfectly! I ran the self-test, and the printer only had about a 25000 page count - a baby! These old LaserJets are tanks, and that was a very low page count. I took the cartridge back, and got it refilled. Total score!
Some time went by, eventually I upgraded the memory to the max (8 MB using common SIMMs) and got the PostScript board for it (so now it is technically a 6MP). It still chugs away, prints when I want it to, I just have to keep feeding it paper - which is rare, because I don't print much.
I'm also only on my third cartridge; I can get them fairly cheap from multiple places, and still a lot of business use these tanks because they are so damn reliable. It isn't a fast printer, but I don't need this thing to spit out a book a minute, just a page here and there; I can wait.
One of the best computing purchases I ever made.
Years later I found a 5P at a Goodwill with the cartridge. I fired it up in the store, and it too had a very low page count (around 15K). Bought it for $25.00 and stuck it in my shop - I also bought the postscript board for it as well. My original intention was to use it for printing toner transfer PCB art, but costs for PCB fabbing have dropped low enough that it isn't really worth it. So I just keep it if and when my 6MP dies...
I figure I'll pass away long before these workhorses do.
I rescued a LaserJet 4L in the late 90s that was being thrown out for constantly grabbing multiple sheets and jamming. Gave it to my parents - with a thorough cleaning, it didn't grab or jam any more and gave them 10+ years of good service (might still be going even now, I dunno.)
IoT furniture + tags. Put stuff [with tags] into cupboards and whatever, and when you're looking for something you can locate it [via that damn' app], so cheap & green tags would be nice.
Similarly tagging stuff that you bring yourself to any group event (let's say you are an event organizer and you hand out things and you want to easily account for them).
Sole pressure monitors in shoes would be great to detect problems with stance/gait/load, etc.
I've thought of doing this for all the stuff kids need for school, plus our own keys etc.
Rather than "IoT" tags, I'm of the opinion that we should use dirt-cheap (a few ¢ each, rather than $10+ for a full "IoT tag") RFID tags, and have a reader that can be accessed by a phone over Bluetooth (or have a few permanently installed around the house). Just need to drive the cost of the readers down.
Unfortunately a lot of people don't realise how useful until they actually try them. For example getting home and having the lights turn on that you'll need automatically.
Before we had to navigate our way to the other side of a open plan room in the dark to find a light switch, now the motion sensor triggers the Hue bulb.
Plus we have a dimmer without needing an electrician to install one, and can set other colors of light for the mood or to improve the atmosphere of a movie.
You can get started with Philip's Hue for $50 (on Black Friday) and at that price point it is well worth it for a lot of benefits/circumstances.
> For example getting home and having the lights turn on that you'll need automatically.
This would be easier to determine if I lived alone. When someone comes in, someone else could be sleeping, watching something in the dark, etc. There's a decision tree for which lights to turn on and how bright.
What I would expect the lights to do is more complex than I would want to spend the time perfecting. I like tinkering, but not with things that I actually rely on. That's why I haven't tried it. Well, that, and because I'm not willing to give an appliance an internet uplink.
If you're more technically inclined you can get started for much less.
You can get an ESP8266 individually for $5, and its only a few more dollars for a relay or a couple of MOSFETS depending on what you want to control. You can use the Homie for ESP8266 framework to take care of the boilerplate, and build functional device that communicates over MQTT in less than an hour.
Most open source home automation software supports MQTT and you can use habridge to emulate Phillips Hue devices if you want to integrate it with an Echo or other commercial device.
Before we had to navigate our way to the other side of a open plan room in the dark to find a light switch
That's not a matter of trying them, it's whether you live in a house with a crappy installation. A switch on every entrance is a basic requirement. I agree that if you're renting, fixing those problems yourself for $50 can be useful, though.
Perhaps but I bet they aren't as flexible. You're paying for that flexibility and functionality (multi-color, dimmable, etc).
For one example imagine if you're away on holiday but want to give the impression somebody is home, you'll need to invest in a timer, with Hue you've already got that.
My ideal version of this kind of monitoring is an interface that lets me teach an app what doing things in the house sounds like - new activity name: shut door and lock > record a few runthroughs; then log when it happens, offer push notifications, etc. I have an induction range that helpfully makes lots of beeping noises while many gas ones do not (but also turns itself off), so that would also be possible.
IoT seems to be overengineering for these use cases. Doors that automatically lock when closed, gas that automatically turns off after a certain time (mechanical timer), solves the problem without the need for adding always-on electronics. These features are standard in a large number of apartments in Korea.
GP was talking about monitoring. Answering the question, "did I lock the door", doesn't require an Internet-connected lock with electronics and actuators. A simple backscatter status reporter using the tech presented in this article would be more than enough. Ditto for the gas cooker.
--
And speaking of smart locks - in the current IoT - that is, cloud-connected - version, they are just idiotic. Such things shouldn't be vendor-locked and their core functionality should be available over LAN. Even that PCMag article doesn't seem to recognize it, as they don't have an "works without Internet" row in their table.
I'm definitely not an advocate. Anytime someone asks me about smart devices/IoT, my recommendations always defaults to "generic" devices. I think we give up too much of our privacy and control to vendors - especially when there are solutions that don't require cloud servers.
Smart locks and some HASS script can do the lock the door thing.
Although you might just have it monitor your housemates smartphone location and lock your doors if all phones have left the house. If you don't want it to lock, it will just push message you the lock state.
If you shop at stores that have a modern cash register, buy gas at most gas stations, drive through most traffic lights, etc. you are using the IoT whether you know it or not.
Those were around before IoT was even a word, POS registers and gas stations haven’t changed much since the 90s. Traffic lights ditto, many haven’t been upgraded since the 80s.
If that is what you mean by IoT as just a rebranding of things that have been around for a long time?
As someone who makes industrial electronics, it's frustrating when people redefine the term IoT. It originally meant anything on the internet that isn't a general purpose computer (or a phone), but now some people define it to refer only to a subset of consumer products.
That's given rise to the term IIoT (industrial IoT), but it's still troublesome since some discussions of IoT refer to both kinds and some don't, and because the exact same piece of hardware can be IoT or not IoT depending on whether it is installed in a home or a business.
Surely discussions of IoT security should refer to both industrial and consumer products!
IoT never had that meaning. It stems from CMU smart devices work (1982) and ubiquitous computing (1991). Gas station and retail POSs as well as smart traffic lights predate that and aren’t related. Maybe some people have decided to take IoT literally and redefine it to include existing technologies, but that definitely wasn’t the original intention of the word.
> Gas station and retail POSs as well as smart traffic lights predate that and aren’t related.
> Maybe some people have decided to take IoT literally and redefine it to include existing technologies
I'm pretty sure it's normal to take new descriptive terms and group preexisting things into them where applicable. For example, just because the terms "car" and "motor car" came into use circa 1895 doesn't mean that nothing prior to that date can be considered a car.
Whether the definition actually works for the devices mentioned is still a valid question though.
> that definitely wasn’t the original intention of the word.
That's an odd thing to argue. We aren't talking about people who compute things when we talk about computers, are we? The original meanings of words are fairly inconsequential if over time the accepted meaning is something else.
Could you give some examples? I've yet to hear a single IoT thing I'd be willing to pay money for.