Slashdot Mirror
Ask Slashdot: Can You Have A Smart Home That's Not 'In The Cloud'?
With the announcement of Google Home on Wednesday, one anonymous Slashdot reader asks a timely question about cloud-based "remote control" services that feed information on your activities into someone else's advertising system:
In principle, this should not be the case, but it is in practice. So how hard is it, really, to do 'home automation' without sending all your data to Google, Samsung, or whoever -- just keep it to yourself and share only what you want to share?
How hard would it be, for instance, to hack a Nest thermostat so it talks to a home server rather than Google? Or is there something already out there that would do the same thing as a Nest but without 'the cloud' as part of the requirement? Yes, a standard programmable thermostat does 90% of what a Nest does, but there are certain things that it won't do like respond to your comings and goings at odd hours, or be remotely switchable to a different mode (VPN to your own server from your phone and deal with it locally, perhaps?) Fundamentally, is there a way to get the convenience and not expose my entire life and home to unknown actors who by definition (read the terms of service) do not have my best interest in mind?
Yesterday one tech company asked its readers, "What company do you trust most to always be listening inside your home?" The winner was "nobody", with 63% of the votes -- followed by Google with 16%, and Apple with 13%. (Microsoft scored just 3%, while Amazon scored 2%.) So share your alternatives in the comments. What's the best way to set up home automation without sending data into the cloud?
How hard would it be, for instance, to hack a Nest thermostat so it talks to a home server rather than Google? Or is there something already out there that would do the same thing as a Nest but without 'the cloud' as part of the requirement? Yes, a standard programmable thermostat does 90% of what a Nest does, but there are certain things that it won't do like respond to your comings and goings at odd hours, or be remotely switchable to a different mode (VPN to your own server from your phone and deal with it locally, perhaps?) Fundamentally, is there a way to get the convenience and not expose my entire life and home to unknown actors who by definition (read the terms of service) do not have my best interest in mind?
Yesterday one tech company asked its readers, "What company do you trust most to always be listening inside your home?" The winner was "nobody", with 63% of the votes -- followed by Google with 16%, and Apple with 13%. (Microsoft scored just 3%, while Amazon scored 2%.) So share your alternatives in the comments. What's the best way to set up home automation without sending data into the cloud?
https://wolfpaulus.com/journal... https://jasperproject.github.i... Neither use Google Voice, and all processimng stays inside the PI, you can also buy RELAY boards that plug into the PI to support home automation. http://www.seeedstudio.com/dep... example above, but there are many others.
Karl Denninger, the guy who writes market-ticker, has done just that, and for the same reason subby has expressed.
His post expressing his reasons for rolling his own -
https://market-ticker.org/akcs-www?post=231376
And where to get it - http://homedaemon.net
Runs on a Raspberry PI 2
B-but can the cloud be very small; on your own server in your own home?
... that is the simplest answer. Anytime you want to connect your appliances to the internet that's going to leak data, so so someone would have to design a home automation kit thats completely self contained on an internet network not accessible by regular protocols, etc. AKA the more customized you make it the more out of the way someone would have to go to get the system working with other devices outside of it.
Only because the corps say so. The reason the corps say so is because there is no money in hardware. Any ould fool can get a few PCB's made and a few components slapped on. The money is in monthly fees and selling data to advertisers. There is no technical reason the server can't reside in your own home or be p2p based, these home automation companies are simply inserting their cloud service as a sort of parasitic middleman.
The trend in the tech industry for years is to unnecessarily drag some element of the service onto a "cloud" service for this reason. Not just in home automation. Part of the problem is that in this tech boom a lot of the guys who would have been writing open source software for decentralised home automation have been hired by the cloud companies
Let's you share only what you want
Depending on the device maker, you may also be able to selectively allow outbound access for firmware patching while still blocking all the other data farming, although you may need to do a little digging into the config and/or traffic capture to do this. Devices will often use the same domain for everything though, and all too often the same hostname, so you might need something capable of URL level filtering to get this working.
Of course, none of that does anything to really protect you from some of the abysmal security that many IoT type devices have on them; e.g. backdoors or other exploitable interfaces that are available over WLANs that enable you to access the device remotely and extract the pre-shared key for your WLAN (see above about putting all this stuff on a dedicated WLAN?), change configuration options, and so on. It's also worth noting that sites like Shodan will also let the bad actors geolocate devices that have known vulnerabilities to them so they can go for a far more targetted war-driving session than used to be the case where it was more of a "see what is out there, and maybe get lucky" exercise.
UNIX? They're not even circumcised! Savages!
to inside of your home"
ftfy
personally I suspect google could get the data to saratoga springs w/least latency but w/amazon they could just use redshift & save the bandwidth/on-prem storage. apple MIGHT put on a pr show for you though...
I can't say about using proprietary, premade devices like Nest, but if you're willing to use Arduinos/ESP8266/whatnot and do a bit of programming you can use an OpenWRT-based router to run an MQTT-broker, or you can use a separate device like e.g. a Raspberry Pi for that and then Arduino/ESP8266/whatever for toggling of relays or logging power-consumption or temperatures or whatever you want automated. You don't actually have to connect any of the stuff to the Internet at all, or you can use an MQTT-client over an SSH-tunnel, or write your own front-end using Apache2 and PHP or a billion different other ways if you want it reachable from the Internet, too -- you have full control over what can and what can't be done over the Internet or if any of it can be accessed from the Internet at all.
This is, however, obviously the hard, DIY way of doing it. If you want an easy plug-and-pray system I have no idea if there even exists anything that doesn't share your stuff with 3rd parties. I, not-so-surprisingly, am in favour of the hard way that doesn't share everything with random, greedy 3rd-parties.
Since this "smart" home stuff began to emerge, I've always wondered what the great thing about it was. I personally do not mind having to leave the chair to turn on the lights, or having to carry physical keys with me to unlock the door. Nor do I mind having a "dumb" fridge where I have to think of the stuff to buy myself.
As a proper slashdotter, I spend a big chunk of my time in front of a screen, so I'm no way non-digital. Still I don't see any benefits in a "smart" home.
Most devices you would need for a smart home (e.g., thermostats, locks, light switches, etc.) are relatively simple, so if you are *really* determined to have a smart home without watchers, why not start making the smart devices yourself? If you get a working model put together, I am sure you could easily start a successful kickstarter campaign and bootstrap a business with it. Win/win for everyone.
The problem with the entrenched players is that they all have a vested interest in making everything cloud-enabled. They want your data -- it is far more valuable than the profit they make from selling the devices, even if they don't sell it to others.
"They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety."
It's been around for decades.
https://en.wikipedia.org/wiki/...
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
OpenHAB is the answer. It has tons of support, but it's still not ready for prime-time if you're not a programmer. Version 2.0 looks promising. I've been running it for two years and it's great. I have used both ZWave and Insteon devices with it.
No. You can't. Illegal. Terrorism. Pedo. DMCA. Whatever. If you ever ask yourself if something is illegal, then it is. If you aren't asking, then you're a criminal.
It's only been around since the '70s.
https://www.x10.com/x10-home-a...
I thought that was all locally controlled.
I am very small, utmostly microscopic.
OpenHAB
And several others.
I use homeGenie. It's free, supports tons of protocols. I have it publish events on all my IOT devices to my MQT server. And have a Java app that subscribes to topics and tells my things to do stuff. None of it is connected to the cloud but I do publish some stuff to external services, like gmail...
Or is there something already out there that would do the same thing as a Nest but without 'the cloud' as part of the requirement? Yes, a standard programmable thermostat does 90% of what a Nest does,
There is, the company is Connexus Controls . We provide HVAC control systems for new installations and retrofit. We provide remote access similar to the way the Nest and others do, but unlike the others, there is no centralized server, your data stays in your home, and the system will function perfectly fine with or without network access. We will provide access to our control API for anyone that wants to tinker with the system, opening up a whole world of opportunity.
I wish I had a good sig, but all the good ones are copyrighted
As a developer of custom hardware and software, I'd LOVE to make products in this space. However:
1) Most people are trained to look for cheapest prices for devices, which are (for the most part) made in third-world sweatshops.
2) To provide a competitive price, you have to manufacture in volume in third-world sweatshops.
3) Due to lack of functioning IP protections in third-world countries, manufacturing there means instantly creating many competitors you cant compete with.
4) If you're willing to give up most of the world markets, you can still only compete against imports by spending lots on lawyers for ITC import games.
In their defense, "cloud" components provide a way to monetize the product in a manner somewhat resistant to third-world knockoffs and late shift runs to your competitors, as well as provide a user-friendly front end that you can tune without requiring the customers to update software, which is always a nightmare. That said, there is NO moral defense against the wholesale "all your data belongs to us, we can sell anything to anyone as long as we anonymize (sic) it" games that are played today. That said, for most modern corporations there are no such thing as morals.
I'm not aware of realistic ways to bring such products to market that are price competitive AND can provide sufficient income stream to recover initial investments, cover ongoing operating costs for a small team, and turn even a modest profit. Not in this world.
A duct tape and bailing wire DIY shouldn't be too hard. Tricky part will be a smooth consistent niceness.
Quick google shows X10 to be alive and well, with RF or wired access to the devices. A webserver-with-API-to-X10-controller bridge device shouldn't be too hard to do with a Pi or similar acting as the bridge hardware, so that can get you on your local network - a quick google shows you should check the Pi and a project called Heyu. Rent a Linode or similar VPS for internet based control if you can't get a static method of addressing your home network when you are away or if your service provider blocks the ports you want to use
Don't blame me, I voted for Kodos
HomeSeer, which is closed-source and has a price tag, but it does not use the cloud. All processing happens on a PC of your choosing (or a prebuilt box you can get from them). Easily the best choice for non-programmer types.
Then build your house on the ground.
I have lights, hi Rez cameras, facial recognition, doors, locks, spa control, garage and alarm system. NONE of that is in the cloud and I control it directly with my phone, get alerts etc. There are plenty of options.
Google, Apple, MS, and others have the mindshare to bring concepts like "smart homes" to the masses, but home control isn't their end game.
Alternately, look at a company like Lutron that has been doing whole home automation and integration products for decades. They don't care about your data, they aren't interested in profiling you ... they just want to sell you the best control products possible, full stop.
Of course, they may also not cater to your need for conspicuous technology. Then again, conspicuous tech gets dated pretty quickly. Do you really want to change out all of the lighting controls, shades, hvac, etc with the same frequency that you update your smartphone, or even your computer?
I ran into this exact quandry when my home thermostat died late last year. Didn't want Nest/Google nor anyone else getting my data. Furthermore, I'm rural so broadband is expensive and sometimes flakey, so that was another reason to want a local-net-only solution, but there were no good ones I could find.
So I just built my own smart thermostat based on a Raspberry Pi: https://github.com/chaeron/thermostat
Since then, I've added battery-powered remote sensors (temp, humidity, pressure, water detection, etc.) to monitor key areas in and outside of the house. These were based on Arduinos (to keep power usage low) and the open source MySensors framework (https://www.mysensors.org/)
Fun project, works great, no security issues or cloud crap. ;-)
More recently I've built on that experience and am working on a dirt bike telemetry project which is based on an Arduino Mega with a GPS module, 9-axis motion sensor, hall effect sensors (for wheel spin/RPM detection) and more. The device captures real time info about the location and movement of the bike (eg. pitch, lean angle, etc.) and then can be used to overlay gauges and a moving 3D bike model over top of headcam footage captured during a ride.
Chaeron Corporation
If it is in the cloud, it is not smart. Having your house report everything you do to people you don't know with interests you don't share may or may not provide useful features, but it is definitely not smart.
In my opinion, the true problem is the hijacking of the word "Smart".
People still tend to use the colloquial meaning similar to "Able to be controlled intelligently, or using additional intelligent logic", while Companies have hijacked this meaning "Able to gather intelligence and advertising".
A Company does not believe it can make money on a "Smart" product without adding that additional income from advertising revenue.
I rather enjoy the benefits of intelligent smart-systems, but like many other readers have already pointed out, DIY systems are about the only way this can be done. Even the control-system "App" (for lack of a better term) is custom-written.
GNU saw corporate influences dominating the OS market; so they built their own OS. Now, we have corporate influence dominating IoT; is GNU going to build their own IoT?
For something like thermostats, fire alarms, and cameras, some off-site access is an essential part of their functionality. For others (switches, etc.), it's a convenience. There are plenty of high-end for business use that don't use the cloud, but they are going to cost you: dedicated data lines, secure off-site facility, off-site server hardware, maintenance staff, backup, etc. Cloud integration is just something that turns an expensive high-end product into a cheap mass market product. But if you're willing to spend the money, or the time to roll your own, you can get the hardware and have a setup that's completely under your control. For rolling your own, you can get home control software that runs on Windows, and open source libraries for Linux, for the major commercial systems.
None of the consumer grade wireless stuff really works all that reliably yet anyway. Some of the WiFi stuff works reasonably well (cameras, thermostats) and is useful. X10, ZWave, and Zigbee are nice enough for turning on a few lights, but I wouldn't trust them with anything more. WeMo is just crap.
You can probably get something to work for one person with extensive tuning. But speech and language processing at product scale requires beefy processing and pricey servers... So something you can tinker with that is hit and miss: Doable without the cloud. A product other people would: Needs the cloud.
I know that the thermostats at http://www.radiothermostat.com... can be configured to disable the cloud function (or I guess you could change its URL to a home server), and they have a JSON-based web interface that a custom home server could use.
#naabhaprzrag, #sverubfr-000, #agi-fcbafberq, negvpyr[pynff*=' negvpyr-ary-'] { qvfcynl: abar !vzcbegnag; }
There are these chaps at Castle OS who claim to do just what you ask:
http://www.castleos.com/CastleHUB.aspx
ur awesome
:Are you a James Bond supervillain?
You are welcome on my lawn.
Taking a break from tinkering with Raspberry Pi + breadboard + sensor kit to post this. Would not be that hard to control the wires a thermostat is connected to with GPIO. The question is just how smart and safe your solution is going to be. Value of cloud is machine learning based not just on your home, but all homes, and human follow up to enhance the software when there is a trend with no automated solution. Non-networked solution will never be as good, and thus will not be a consumer product.
In my opinion the only way to avoid the big companies collecting your information is DIY.
At most on the outside you should ever need is a Dynamic DNS provider, and there are dozens of those you can use and script to send your outside accessible IP (assuming you don't have static address[es]).
A lot of this is actually fairly simple programming with basic IO sensors. You could build a thermostat like a Nest with a PiZero, a basic thermal sensor, a couple of relays and some of your own time. Sure, it won't look pretty, but it doesn't have to if you're using your smart phone to control it anyway. A basic web interface that is Mobile friendly isn't all that hard.
I'm sure there is already an open source project that does this with a full breakdown of how to do it to save you some time.
And a quick google search away I found this http://www.stuff.tv/features/how-build-homemade-nest-thermostat
I have a Vera Lite. It can be entirely functional without a connection to the internet. I have door locks and a thermostat which talk on the Z-Wave protocol. When one of the doors is locked using the button on the outside the Vera changes to an "Away" preset.
I Don't Work Here
B-but can the cloud be very small; on your own server in your own home?
Not unless you want to spend a lot of money, and hundreds of hours of your own time.
Look, the economics of this is simple: By producing data that can be monitized, the cloud companies can reduce the up-front price. Most people go with the cheapest option. This reduces costs even more, since NRE can be spread over more units. It would be very difficult for a non-cloud company to compete with that. People that care about their privacy, and are willing to pay extra to protect it, are a niche market.
My home automation system uses an Amazon Echo and a Samsung SmartThings hub. The Echo is cloud based. I would prefer a non-cloud solution, but to be honest, I would not be willing to pay much more for it. I don't really care that much if Amazon knows what time I turn off the lights.
Removing spying background services on an open system like Android is easy: either don't install the Google stuff (or remove it), or disable it selectively:
1. Root the phone (it is YOUR phone, you're the boss).
2. Install a service manager like https://play.google.com/store/...
3. Open it, go to system, open Google Play Services.
4. Disable AdvertisingIdNotificationService, AdvertisingIdService, AnalyticsIntendService, AnalyticsService and AnalyticsUploadIntendService.
Now open Google Settings and see that your device does not have an advertising ID anymore. The above method kills most, however some apps collect their own data and don't let it go via Google so watch out what you install.
Most european manufacturers joined the KNX consortium www.knx.org. The standard was born in the '90s when the sun was shining without clouds, so full of modules to choose from, full of non-cloud smart home servers.
It's not as shiny and takes more work than using the big name solutions (ie: Nest/Google), but the options are out there.
The RadioThermostat works over your choice of WiFi, Z-Wave, or Zigbee (pick up to 2 protocols). The cloud service is easy to setup and includes a convenient app, but the API is fully documented and compatible with a number of open source home automation servers. You can easily disable the cloud service if desired, or reconfigure it to point to a server of your choosing (local or remote).
I'd say it'll be another year or two for the various open source home automation servers to mature to the point that they'll be as convenient to setup as the commercial systems. For now, unfortunately, it's still very much a developers-only world of roll your own solutions, unless you happen to be using the same devices that are supported by a single software package.
I currently have a Z-Wave lights, RadioThermostat, a discontinued system providing leak and motion detectors over a custom serial protocol, TI SensorTags, RF blinds and LED lighting that I plan on controlling via my home server. The hardest part I'm finding is actually controlling the RF devices, while I've given up on the ebay LED light strip controllers and am working on custom Arduino-based controllers there.
Eventually I want to add a Smart lock to my front door to give me keyless entry, but I've yet to find a smart lock that I feel is both secure and not dependent on a third-party server literally holding your (master encryption) keys hostage so you can't freely create your own digital keys at will (looking at you, Kevo).
From: When a stranger calls.
"The call is coming from inside the house!"
IIRC, they had the whole motion light thing too, but I digress.
Actually, the cloud stuff is nonsense, and this IoT shit just needs to stop and die.
when you can build your own for hundreds less with less effort? Do you have money to waste?
Have a look at fhem.de. I have been using the fhem server with various home automation systems without a need for a cloud.
You need to be looking into open protocols, and implement them using free and open source software:
https://www.ietf.org/proceedin...
It's only a matter of time before the cloud controls those too. I'm not exactly awaiting the time I'm awoken by Alexa saying "Shh, just let it happen."
You think that the savings are passed on to the customers? That's cute. And I suppose you're expecting to get a pony for your next birthday as well.
There are no savings from doing that provided to the customer, that's money that goes to pad the balance sheets of the company selling the devices even as it significantly reduces safety and security.
That's bullshit. Speech recognition was at like 97% or so for years before people had always on connections. And it gets even easier if you're dealing with commands and have people using fixed commands. Sort of like what Google does with OK Google. If you add House Activate or something similar before the command, then the system just has to see if what you said matches a known command.
The only thing that's at all tricky about it is setting it up so that it doesn't activate in response to the TV or radio.
This point is best exemplified by IP based webcams. You used to be able to buy web cameras that you attached to your home network and configured to access via an internet address. Now most off the shelf cameras force you to run through an intermediate cloud service of some kind. IP based webcams have not disappeared and I still think true techies would prefer them, but the mainstream market gives the impression that there is no other way than relying on the cloud.
You have to pay a monthly fee to get the Control 4 system into the cloud. It's apparently a very popular system probably because they setup a very strong dealer network and require the dealers to do any changes or modifications - it's basically a closed system to the home owner so its really stupid. Want to add a new zone or piece of equipment? Gotta get it from a dealer.
Pick the right controller to start, Vera is a good starting place and Openhab is more than happy to control it later on. Neither of those need internet access to work. My HA system has little to no internet access. I VPN in from phones to run it remotely and use a bit of custom code for geofencing. Now parts of the system have internet access my alarm panel use it as backup to talk to the monitoring company. Openhab is allowed to talk to weather and some other bits via my proxy. My garage door remote works fine without internet access because the local API works. Same goes for some wifi based wink stuff with my HA system IP bits all having their own vlan.
As to hardware I've tried a nest it did nothing my $22 new (it was a realy good deal) off ebay zwave thermostats do (besides look cooler) that I actualy use. There are a ton of way overpriced ha solutions all with there own ecosystem. Zwave gear pretty much just works and does not expect to talk to the outside world I've got 100+ devices around a large home (mostly AC dimmers and relays) with a smattering of sensors for occupancy the also do temp and several other bits. A hardwired alarm to do the actual alarm stuff and keep my insurance guy happy. CCTV is all internal and encrypted at rest via public keys (it can write but it can not read).
Design wise it's important to keep to only one devices realy should be talking to the world, thats the one thing that needs to keep getting updates. For me thats openhab.
No sir I dont like it.
I built a smart thermostat about 12-15 years ago.
It's really not that hard.
Get an RCS TR-16 thermostat.
Hook it up to a PC (or, today, a raspberry pi).
Read the specs on the protocol and write a small daemon to listen for requests and take appropriate action based on them. For good measure, add sanity-checking of request parameters (don't allow it be set to cool below, say, 65, or heat above 72).
Use netcat or telnet to talk to the port it listens on.
It's really not difficult at all.
These days it'd make sense to build yourself a simple gui for your phone/tablet.
That's bullshit. Speech recognition was at like 97% or so for years before people had always on connections. And it gets even easier if you're dealing with commands and have people using fixed commands. Sort of like what Google does with OK Google. If you add House Activate or something similar before the command, then the system just has to see if what you said matches a known command.
The only thing that's at all tricky about it is setting it up so that it doesn't activate in response to the TV or radio.
^This. Mod parent up. Natural language parsing and speech recognition has been improving for years, and even Apple has finally allowed "offline recognition" options for their base system.
Going to the cloud makes it *easier*, since it vastly increases the number of samples and allows them to not care about processing resources at all and be generally shit programmers unless their project eats up too much of the internal balance sheet.
We all have computers far more powerful than are necessary to do this in our pockets. Add a desktop system to act as a central unit (not an unreasonable requirement) and to offload any particularly difficult recognition task to and it's entirely possible to have it all work internally.
Hire a Linux system administrator, systems engineer,
Another big issue is responsiveness of rules/triggering/devices.
I have a smartthings hub and sometimes simple motion triggers or routines fail and usually it's due to some sort of network issue. If I lose internet connectivity then a lot of my smart things become dumb things. To be fair things seem to be improving albeit slowly but I am still tied to Samsung and whatever they want to do with their services - it looks like they are more focused on TVs and Refrigerator hubs at the moment. I have no confidence that they or any provider will keep things going through the lifetime of a particular product - Revolv anyone? At least the IOT devices themselves are relatively transferrable.
Have been looking at CastleOS as a possible alternative. Not sure where they stand with Zigbee integration - my devices are a mixture of Z-wave/Wifi/Zigbee
Also they are windows ?10? based so there is the whole potential spying thing there too but at least the operation is not cloud reliant.
E.
I don't "expect" it, I observe it.
No, but you apparently do.
You couldn't be more right!.
I remember dictation working really well on a windows 98 computer using dragon naturally speaking without any internet more than 15 years ago.
No, you don't observe that. What you observe is that newer items with more power come out for about the same price as they used to. Texas Instruments is a good example, the price points don't change, but the capabilities have gotten a bit better over the years. But, you're still looking at the same price.
In markets where there's a bit more competition you'll see that the companies will occasionally be forced to reduce prices in order to compete, which lasts until they buy out the competition and put an end to it.
They're not passing savings onto the customer and I've rarely, if ever, seen products be reduced in price because they're starting to spy on the customer. They're certainly not passing on the money that they're making on an ongoing basis otherwise eventually these devices would pay for themselves.
The sheer mental gymnastics necessary to make your point valid is mind boggling. Mostly because the savings aren't passed onto the consumer, they're kept as corporate profits.
no you don't.
Your post rings of maternal basement dweller
At least the future is here for sprinkler timers!
RainMachine is openwrt based, opensource, allows you to have the connection over the cloud or to just do your own port forwarding (or VPN).
It also only uses open services to gather info about weather in your area to plan the watering needs.
IT does not look as featureful as e.g. it's fully cloud-dependent competitor, but it also would keep working in absense of Internet or once the competitor's cloud offering is eventually withheld too, I imagine.
I am really glad I discovered this thing a couple of months ago and I hope somebody comes up with a thermostat that is similar in principle.
Do you find the Echo/SmartThings combination effective? I am considering doing the same thing.
I've given up trying to figure out who is listening to what, when, and what they are doing with it. I've gone "dumb home." I use a double-pole single-throw knife switch like they have in the old Frankenstein movie to turn my burner on and off. The crackle of electricity and the smell of ozone is very invigorating in the morning (and I get to yell "It's aliiiive" for the neighbors). I don't think Google has figured out a way to monetize it yet, but I hear they are working on it.
E Proelio Veritas.
> The winner was "nobody", with 63% of the votes
How this isn't near 100% is completely baffling to me.
pr0n - keeping monitor glass spotless since 1981.
Incidentally, when I got into IoT and "that stuff", the first thing I did was starting to write my own software to control lights. So far only made it to a generic library, with a reference implementation etc, partly because all the smart-home equipment is too expensive ... and companies don't always share (local) APIs .... and I'm lazy ... but at least I got it pushed to Github.
But really, only buy stuff that has a local API (i.e. can be accessed directly via (W)LAN), so you're not 100% dependent on cloud-stuff.
I use a Homematic CCU2. It works perfectly without any cloud.
I'm just posting this because nobody else did.
I wanted exactly the same thing and I can really recommend Domoticz (open-source which can run on a Razberry Pi) in combination with a RFXCOM (which can communicate with a lot of (also inexpensive) stuff on the 433Mhz band).
It isn't expensive either, the whole kit costs around â150.
I personally bought a complete set here since I live in the Netherlands: https://www.sossolutions.nl/domoticz-starter-rfxcom-e-2b
I'm using some rather inexpensive sensors from klikaanklikuit.nl
Lastly I made my own app with tasker to control everything on my smartphone. And I'm using Pushover for notifications.
Might sound complicated but it really wasn't as everthing can be integrated very easily with Domoticz. It works great for me personally.
Currently mostly used in surveillance cameras.. ..
Normally using the technology you create point-to-point (normally strong authenticated and encrypted) connections from app's directly to devices..
No cloud-in-the middle
Here's some examples:
http://www.nabto.com
http://www.tutk.com
http://www.danale.com
Basically its VoIP technology made for IoT...
I've mostly seen it used in products from Asia.
I had to set up some smart thermostats for a customers vacation home that does not have Internet access. Next to impossible to find something that worked with the systems he had that did not need to be connected to the cloud.
I haven't kept track, but ISPs used to shit bricks if you tried to run a home server (without paying for a business class connection). Their (somewhat legitimate) reasoning was that home servers were more likely to be hacked and used for things like anonymous e-mail relays for spam. But most botnets work just fine with any Windows desktop OS, so that reasoning is no longer valid. Still, with everyone trying to sell cloud services, rolling your own will still meet some resistance.
Most home automation/security systems and services are subsidized by the reselling of your data once it hits the cloud. So if you try to run your own private service, be prepared to pay the full (unsubsidized) price for your hardware.
Have gnu, will travel.
It could be done. The embedded boards are not expensive, and one can get a Raspberry Pi Zero or an Arduino for pretty cheap, and build from there. Someone who wants to spin their own boards and even make their own ASIC is doable. If one wants as small a physical size as possible, it may take about $500,000 to fab an ARM SoC ASIC with a decent CPU, but it isn't impossible. We had X10 offering some form of home automation for decades now, and it never needed a cloud.
The trick is to use a hub and spoke arrangement, with 1-2 redundant, hardened hubs, then devices that communicate with those. This reduces the attack surface to needing to be physically near the LAN, or compromising the hardened hub, and with VMs or containers, one can separate functionality like firewalling from device management and software updating to ensure better defense in depth.
Security of IoT devices is a solved problem. Consoles come to mind as something that has had years of attacks from all sides, without a single breach, and a console is pretty much a computer. It is just getting device makers to actually bake security in, as opposed to strapping in on later on.
Do you find the Echo/SmartThings combination effective?
No. There are a lot of problems and incompatibilities with various IoT devices. "Google Home" will supposedly have a built-in Z-Wave hub, so it will not need a separate bridge like the Amazon Echo does. You might want to wait a few months for it to be released. But Google Home is designed to mostly work with Nest devices (which are way over-priced) so I am not sure how well it will work with third party devices.
Home automation is still in the "early adopter" phase. So you should go ahead if you want some toys to play with, but not if you are looking for something that will actually save you time and money. Something that your mom would use is still a few years away.
Easy. It will be just like eureka/s.m.a.r.t. I think Watson became public but if not. You can study and easily replicate it with few thousand dollars of memory. Or mimicking preexisting programs to private/triple encripted local lan
There is already a "smart home" system that doesn't require being in the cloud. Crestron. But it is not a roll-your-own solution, which I suspect most people here would want.
I install them professionally. You buy AMX or Crestron, and have it installed and programmed. BOOM non cloud based home automation that works fantastically.
This has been the case for well over 30 years now, and most rich people have been enjoying it.
Non cloud based reliable stuff has existed for a long time now.... Dirt cheap home automation for poor people? That is a new thing from the recent decade. and "cloud" is how you extract money from those poor people unwilling to spend $20K on their home automation system,
Do not look at laser with remaining good eye.
Leviton Omnistat (formerly or still currently HAI) thermostats. I've had one for a decade. It works well.
It talks RS-232 over three wires (use one of those cheap PL-2303 USB adaptors to connect to your computer). Powered from the 24vac furnace loop (or run a separate 24vac circuit and use relays if needed).
They have a simple protocol and various folks have published scripts to talk to them. My setup polls the thermostat every minute to read the temperature and get the furnace status. It updates the time-of-day clock on the thermostat every night so the thermostat is only a few seconds off. This also takes care of DST changes.
I can write whatever interface I like to manage the thermostat settings throughout the day (currently just loading a text file because that's my kink, but if you like flashy web things, you can do that too).
The setpoint programming is in 15-minute increments and has weekday/saturday/sunday programs, so once you load the programming to the thermostat, it operates independently.
Follow the money. The cash is taking the data slurped from every device and putting it up for sale to anyone who wants it and who is willing to pay for it. Do you think that generic flashlight (or fleshlight app, as I see fairly often posted on Slashdot) really needs battery status, call volume, text access, access to root, and access to so many things in the Android manifest? No. Any info the app slurps up is heading its way with your name on it to some warehouse.
Everyone wants a piece of this pie. At least two ISPs I know of used to actively insert headers into HTTP requests so websites can definitely fingerprint an IP address or computer. If the device connects to the Internet, it will be telling all it can find out to the mothership like someone who has been "softened" with xkcd's $5 wrench.
Again, follow the money. A company can make a smart toilet with a remote controlled flushing app, just by using the same SSID for the app and the toilet. Or, they can have all that info (when is the toilet flushed, how much water, how much corn is in the poop) be saved away and make two sales, one to the end user (who really is just a second product), and one to the real customer (whomever is willing to pay the bills for it.)
Want to know how to do home automation right? Get an Arduino or other embedded item and work from there. Until there is a company that actually can just have things work without demanding all data be shipped out, the best way is to do it yourself. Hell, a friend of mine with an RV took an embedded Linux controller, a touchscreen LCD panel, various electrical measurement devices and electric valve controls, and can control/monitor his whole rig just by using a web page (which has client cert authentication, and only allows IPs from one VPN.) If one person can do this, it is really a matter of won't, rather than can't when it comes to IoT device security (or lack thereof.)
I've constantly tried maintaining requirements for my SW to work from behind
a proxy. Game manufacturers are probably the worst, but a huge hit in privacy -- I like the idea of a "smart home", -- but I want to control it from my home computer --- not a "smart phone". As near as I could tell, most of the home automation products will only work / can only be operated from one of their apps that you can get for various smartphones. None of them that I looked at had any way to record, control or analyze the info at home.
Even a stupid company that provided a smart plug to measure electrical usage had to be directly on the outside net (or "think that it was using NAT") to contact the manufacturer's website, then to monitor your data, you had to pay monthly service fees for various levels of access to your own usage data.
The state of smart "anything" (including the mandatory smart-electric meter required by my area-monopoly-electric provider (PG&E/CA/US) sucks. My electric usage rates went up by about 20% when they installed their new meter, but they won't allow me real-time access to the output to verify usage data -- I am only allowed website summary data some amount of time later.
Before they installed that, I looked for a smart meter that would allow me a serial line tap on getting data, that would also give integrity guarantees.
Basically, it came down to any device that allowed me real-time access to verify actual usage data, was not approved for use on their electrical network.
So now, they can monitor my usage, supposedly down to the their profiled-device database, but I'm not allowed to verify the meter's accuracy when their previous meter had supposedly been 20% inaccurate for decades?
Similar problems and constraints seem to apply to other "smart-enabled" (meaning externally monitor-able by a 3rd party) devices. The presumption and requirement that I have a smart-phone is becoming increasingly annoying -- including constant prompts from Google, Amazon, Steam, etc to secure my account with my SMS-enabled smartphone. Only google had an optional verification step (at one point, dunnow about now), that allowed them to call you at a POT and listen some numbers and type them into a website, but
that option is not available for all of their services and is not available at all for many other companies' services.
It varies from country to country but my guess is most homes are about 30 years old. Every service in them usually is working on the orginal services because they use simple mechanical principles that are timeless. The doors are on hinges, the lighting has mechanical switches, the water has mechanical valves etc. They are all maintainable and still working 30 years later. How many electronic devices do you own that are 30 years old and if they break can they be fixed. How much of the tech you own today can be expected to be working and maintainable in 30 years? How many of the cloud services do you expect be operation for 30 years?
At one time I worked for a home automation company doing an advance full home automation system aimed at new home builds. It used a local Linux server running on an embedded platform with no moving parts so could be expected to have a long maintenance free standalone life. I did worry about the longer term if it failed. The point become more relevant as that system, installed in a huge mansion, is not easy to understand from a service POV and the company that developed it is history now.
I want to do some home automation in my current home. I have the hardware/software/financial skills and resources to DIY it but what about when I sell my home? I don't want to support it, and indeed my current home is quite new and I am not, so anything I install could be around after I die. Putting in home automation now may actually reduce the resale value as it is a 'risk' for a buyer.
My thoughts are:
- No cloud solutions, who knows when they will go out of business. The intelligence in the system must be standalone and local.
- Use mature standards e.g. Cat 5 cables with RJ45 connections to interconnect remote modules back to a central point.
- Put electronic modules in accessible places. This and the above cabling make it possible to replace the whole electronics with a new generation in future if needed.
- There needs to be more mature and open standards, like X10, So it has a long term future.
I will do home automation but very cautiously and piecemeal with the above point in mind.
The newer Samsung phones self-disable when rooted, unfortunately. I can't find a good replacement for my Note2 that has wireless charging.
Mycroft just released (Python, version unspecified) code they say you can run on your Raspberry Pi; Mycroft is an Alexa-like system, differences being it's open, the s/w is free so you can build your own, and the hardware is pretty open too.
There is cloud STT (Speech-To-Text) going on, but they're interested in local STT according to an email they sent around to those of us on their mailing list. My GPS (ca. 2013) does non-cloud general STT, so there's working code out there.
Speaking as an owner of both an Echo and an Echo Dot, I'm very hopeful that Mycroft will join them, perhaps even replace them.
Echo's huge drawback is that it doesn't have a local operating mode via LAN ports, nor local STT and TTS. Not to mention the absurd requirement that you put up an SSL server just to make the simplest possible function work.
I've fallen off your lawn, and I can't get up.
Your examples (Google and Apple) show that you can optimize a good local only solution if you have large scale data collection in the cloud from which to get good data. These models are just so data hungry that it is impossible to make a good product that other people would actually pay to use without a good feedback loop.
Natural language parsing and speech recognition has been improving for years, and even Apple has finally allowed "offline recognition" options for their base system.
Just to go on the Apple theme, Apple had "Speakable Items" in OS 9 and OS X for many years which worked very well.
A few years ago I started a switch from X-10 to Insteon and evaluated some of the systems mentioned in other posts. I settled on Home Control Assistant http://www.hcatech.com/ as it lets me control everything locally but allows remote access if I want; the client app on my phone knows what to use based on what network it's on. The price if very reasonable and support has been great, and it's really well documented.
It does need a Windows PC to run on but I had one destined for the recycle bucket that works fine.
Seeing how well this works there is absolutely no reason for "cloud" in home control - if you can set up a PC and install software on it. For some people that is a barrier to use.
I do however have some Honeywell thermostats as for the money they do a good job versus the alternatives that existed. Their web app is easy for everyone in the house to use. I figure someday someone might figure out good local control for them or I'll replace eventually replace them. But it's tough to argue with something that is low cost (compared to things like Nest) and just works.
It can be done cheaply, but that's only if you don't include the time to set it up as part of the price. Home automation without a prepackaged cloud solution is going to mean getting in there with a soldering iron and at least a bit of scripting.
Digging deeper, it depends:
So the rest of the misunderstanding in this question is to do with clear-ish requirements and specification. If you want it to hook up with Amazon or Google stuff, you're probably heading for the (one of the) 'commercial clouds' and some lock-in, for example. Personally I'm an anti-lock-in, anti-proprietary-cloud nazi, I do not feel that Amazon, Google and Microsoft are my friends and have my best interests at heart.
On y va, qui mal y pense!
Yes, completely agree that the MySensors.org libraries for home automation can run locally. I use many $1 nRF24L01+ 2.4GHz non-WiFi radios which form a mesh network, with optional authentication. I have the sensors for temp, pressure (a single pressure sensor in the home is so high resolution that every home situation, door window, cooking etc can be determined!), humidity, door, window, light, RTC, ultrasonic movement, IR detection, etc - nearly everything costs a dollar on ebay.
you can run the sensors on cheap clone nano-arduino's (also costing a dollar on eBay) and they feed-back to a (real) $40 Arduino Uno hub for reliability This acts as a serial gateway and connects with an Raspberry Pi3 running the high level control software. The RPi3 is UPS'ed with a cell-phone emergency charger - with ethernet connection to a home router (to detect if the mains power goes out) it is great fun, many congrats to the Swedish MySensors for their RF24 library and pioneering work in the Open Source IoT systems.
no you don't.
Your post rings of maternal basement dweller
That's because mom uses it to keep him down there.
Just another day in Paradise
The newer Samsung phones self-disable when rooted, unfortunately. I can't find a good replacement for my Note2 that has wireless charging.
The Nexus 6 has wireless charging, and has an unlocked bootloader.
I am working on a project that in theory should be perfectly able to do home automation, the Hackerfleet Operating System (HFOS). ;) way, but i recently found out, it should work nicely on your Android device as well.
It is supposed to run on small energy efficient machines (RPi etc) and provide a host of services in an offline-compatible, yet federated (think of freifunk/mesh
It basically consists of a backend (Python) and a rather heavy frontend client app (HTML5) and focuses on small group collaboration - the missing parts for better teamwork without the need to offload your data to anyone else.
Some of the goals i have on my list are already partially working:
* Sensor evaluation with fancy charts, graphs etc - originally aimed at sailors and their ship's sensors!
* Maps - again, one of the heavy topics because of the sailor-orientation, it will do waypoint planning, weathermaps (grib), and a lot of other goodies, offline via caching
* Project management - not so much focused on software but actually on my private needs - and yours?
* Shareable item reservation for rooms, tables, working equipment etc - need this for c-base, the local hackerspace
* Library management for books and similar media. Small, not so heavyweight as other solutions but still useable for a federated hackerspace library
* An etherpad + wiki collaboration tool as well as a chat module
* Switchboard for simple on/off switches
* Simple and modular "If-this-then-that"-like control system
* Garden watering automation
* Robot remote control - we've already tested a lot with our robotic vessel MS0x00 ( https://duckduckgo.com/?q=ms0x00&t=ffsb&ia=videos ) which is now and then being used for automatic water depth and other measurements
Since i work more or less alone on the project and because i am currently preparing the release of a rather major overhauled version (1.1, sporting a modular component/app-system and a lot of frontend technology upgrades, release is planned for the next few days), the project is not yet in a really useable state.
So, I'd be more than happy to welcome new HTML5 (Angular) and Python developers to join the fun! ;)
Also, folks who can happily indulge in writing really complex application testing systems
Check out the repository on github: https://github.com/hackerfleet/hfos and join us on #hackerfleet@irc.freenode.org :)
Cheers!
riot
Those with smart homes in the cloud will be the first to be enslaved.
At the current state of the art at least, cloud based voice recognition simply works better than anything you can implement on an affordable local system. There are two reasons.
One is that the cloud system can devote massive amounts of resources intermittently when you need them to recognize a voice command, but give those resources to other people when you do not. Current day voice recognition systems really aren't very intelligent; they work by comparing massive amounts of data with the recorded voice data, and by doing recognition of words in the context of the surrounding words rather than one at a time.
The other is that the cloud system can massively crowdsource the problem of training the system. They can use the experience of their millions of users to make it work better. A locally based voice recognition system cannot match that.
We can make inexpensive local systems that can recognize a small number of sentences. We can make expensive ones like Dragon NaturallySpeaking that need the computing power of a decent PC or Mac and a couple of gigabytes of RAM to work decently, and even with all of that it doesn't work as well as the cloud-based personal assistants that are now ubiquitous. For example, Alexa can process voice requests to play any artist on Spotify, from a library of over 35 million songs and tens of thousands of artists, and usually gets it right. Nothing you could run on your home entertainment system could come close; even Dragon wouldn't be much help because you need a specialized database of performer names to make that application work well.
Explain to me again what I'll gain - which I consider to be of value - from this "home automation" concept.
Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
That's bullshit. Speech recognition was at like 97% or so for years before people had always on connections. And it gets even easier if you're dealing with commands and have people using fixed commands. Sort of like what Google does with OK Google. If you add House Activate or something similar before the command, then the system just has to see if what you said matches a known command.
Ooh! Ooh! Can I have my command be "Wonder Twin powers, activate!"???
Look, the economics of this is simple: By producing data that can be monitized, the cloud companies can reduce the up-front price. Most people go with the cheapest option. This reduces costs even more, since NRE can be spread over more units. It would be very difficult for a non-cloud company to compete with that. People that care about their privacy, and are willing to pay extra to protect it, are a niche market.
I think the solution is probably to do what the amazon fire does and charge a premium for ad-free. The monitization really doesn't produce a huge amount of money. Even facebook, one of the kings of data mining and selling ads, only makes like $1/month/user. It should be relatively simple for someone like nest to sell a $200 box that replaces the "cloud" for people that want local control.