Home Automation Without the Cloud

Current Status: Public Draft

By now, everyone and their mother has written an article about home automation – so why another one from me? The answer lies in the focus. Most of what I’ve read revolves around shiny, off-the-shelf solutions with their own apps, cloud backends, and Alexa integration. In short: exactly what I want to avoid.

I’ve expressed my general skepticism towards cloud services elsewhere. But this isn’t about paranoia over data security. Of course, it doesn’t feel particularly cozy knowing that Chinese or even US providers might have access to my heating system. And yes – my heating schedule, shutter controls, and the start times of my coffee machine, dishwasher, or washing machine could reveal quite a bit about my daily routine. But the real threat level is minimal.

Far more relevant are two other topics: resilience and obsolescence.

Resilience of Infrastructure

Home automation should work reliably – even when the internet connection fails. Anyone like me, living in digital backwater Germany with a dusty-rose-colored DSL line, knows the reality: copper cables that are decades old, nominally offer “100 Mbit/s,” but are in practice unstable and fragile. Meanwhile in Sweden, they’ve laid fiber deep into the woods – here, the official line is still that you don’t need internet at every “milk can.” Probably not even the one in my kitchen.

And when the modem component of the Fritzbox burns out every other year, it becomes obvious: this infrastructure is no stable foundation.

The problem shows itself in daily life. We launch the shutter control app, the smartphone drops WLAN because it detects “no internet” and switches to 5G. The app opens, we select the shutter – nothing happens. The server is reachable, but the device in the apartment isn’t. The same with the heating system. All the components – smartphone, router, actuators – are inside the same four walls, and yet nothing moves. Only the coffee machine responds, as long as you’re close enough for a Bluetooth connection.

And this in an environment where every function already requires its own app. Home automation based on this principle quickly creates more frustration than comfort. Alexa and the like don’t help either: without internet, they’re just as mute / moot.

So the real everyday risk of cloud automation is less about data privacy than about dependence: on an unreliable line, on servers overseas, and on the long-term survival of proprietary platforms.

Obsolescence Through External Dependencies

Which brings us directly to obsolescence. Cloud-based systems are tied to foreign servers by design. As soon as a company decides to shut down a service for cost reasons, all the associated devices stop working – even if the hardware itself is still perfectly fine. And that’s without even considering scenarios like embargoes or political sanctions.

For users, that means high upfront investments in devices that could turn into electronic waste overnight. A few examples:

• Logitech Squeezebox: Logitech discontinued the Squeezebox line in 2012. Since operation was tightly bound to mysqueezebox.com, many devices became practically useless without it. Those who didn’t switch to the local Logitech Media Server suddenly held expensive paperweights. (Heise)
• Google Nest / Works with Nest: Google shut down Works with Nest in 2023. Numerous integrations and automations stopped functioning instantly. Users had to migrate to Google Home – provided their hardware was even compatible. (Google Support)
• Apple HomeKit architecture: Apple is pushing the migration to a new HomeKit architecture. An update will soon be mandatory – older hardware that can’t handle it will simply stop working. (Heise)

And what does this mean in daily life? In the best case, individual automations no longer run reliably: shutters respond only with delay or via manual buttons, heating systems fall back to a basic local mode, a balcony solar battery only throws errors. In the worse case, entire device categories stop functioning – and since many are built into walls, it’s not as simple as uninstalling an app or flashing firmware.

In the worst case, essential systems can no longer be controlled at all: heating stuck on one fixed setting until the hardware is replaced, shutters frozen in place despite fully functional actuators, and a home battery that, without its cloud service, charges and discharges solely from grid power instead of your own solar.

This isn’t just a loss of comfort but real functional outages – quickly becoming expensive and impractical. An electrician has to come in, wall-mounted actuators get replaced, maybe parts of the installation rebuilt. What started as “practical smart home comfort” turns into a costly nuisance, entirely at the mercy of a vendor’s update schedule or business decisions.

From Early Automation to Today’s Central Units

Many households already have early traces of automation – sometimes grown over years. Basic solutions existed long before my own start, such as simple radio switches or proprietary heating controllers. I personally started with FS20 and the FHT80 series – interesting for their time, but not without flaws.

One major driver for me to move on was the constantly empty batteries and rechargeables. With standard 1.2-volt NiMH cells, even a moderate temperature fluctuation was enough to drop voltage into the “empty” range. Disposable batteries weren’t an environmentally friendly or cost-effective alternative. I even tried 1.6-volt NiZn rechargeables – with expensive special chargers – but their lifespan only barely exceeded that of throwaway batteries.

Then came Homematic and later Homematic IP, which worked far more reliably and marked the first real step out of the hobbyist niche. That was when home automation began shifting from a nerd product into something broader households could consider.

Today it has fully hit the mass market. Products like the ubiquitous IKEA smart lights prove that automation is no longer niche. But this spread has also led to a flood of apps and cloud solutions: every product line comes with its own gateway, its own app, usually with little thought for interoperability.

The result is a patchwork: multiple gateways, parallel apps, and almost no common standards. This is exactly where it makes sense to move to a dedicated local hub – independent of the cloud, expandable, and able to pull everything together.

Raspberry Pi as the Core

A prime candidate for such a hub is the Raspberry Pi. This tiny single-board computer brings together a whole bundle of advantages that have made it almost the default in this field. Power consumption is only a few watts, even 24/7, far below any traditional PC. Its size – about that of a pack of cigarettes, case included – makes it easy to hide in a server rack, on a media shelf, or right next to the router. Cost is modest, from double to low triple digits depending on model. And after the pandemic-era shortages, supply has normalized again.

Then there’s the broad range of interfaces. USB, GPIO, HDMI, Ethernet, WLAN, Bluetooth – few devices this small are as flexible. The software ecosystem is equally rich: specialized operating systems and distributions exist for everything from classic CUL-stick setups to newer protocols and integration platforms.

And typically there’s still enough headroom to run extra services on the side – a small personal website, or even your own Nextcloud instance. The Raspberry Pi thus stands as a representative for an entire class of compact, efficient, but powerful mini-computers that serve perfectly as the command center of a home.

Excursus: Wireless Integration on the Raspberry

On its own, a Raspberry Pi is not home automation – its real strength is becoming a bridge into many wireless worlds. This is where the small USB sticks come in, teaching the Pi to speak older and newer systems alike.

In Homematic, everything normally runs through the CCU (Central Control Unit), the heart of communication and logic. Buying the official CCU3 gets you a standalone box. On a Pi, you can achieve the same with the HmIP-RFUSB stick. Combined with RaspberryMatic, this turns the Pi into a full-fledged CCU replacement that supports both classic Homematic and Homematic IP devices.

Other specialized sticks cover further systems. A CUL stick (Busware CUL v3) connects FS20 components, older FHT80 heating controllers, and sensors like the KS300 weather station, which operate at 868 MHz. A JeeLink stick is useful for various 433-MHz devices like simple sensors or certain weather stations. For IKEA’s Trådfri lights, Zigbee sticks such as the ConBee II or the Sonoff Zigbee 3.0 USB dongle provide local integration without the cloud.

The key advantage: nothing has to be replaced all at once. Old FS20 sockets in the living room can continue working next to a KS300 on the balcony. At the same time, newer Homematic IP radiator controllers run smoothly, and the Zigbee under-cabinet lights in a fresh IKEA kitchen integrate right alongside.

Step by step, the Pi grows into a universal wireless hub. On a single USB hub, Homematic, FS20, Zigbee, and others coexist – all controlled locally, with no data leaving the house.

Home Assistant as Control Center

However diverse FS20, Homematic, Zigbee, and proprietary ecosystems may be, from the user’s perspective they should act like one system. That’s where Home Assistant comes in. Running happily on a Pi, it brings drivers and integrations for hundreds of protocols and devices.

One central example is household energy balance. The APsystems EZ1-M inverter provides local data on solar output from the balcony. A Shelly power meter measures simultaneous household consumption. In Home Assistant, both values combine, giving instant insight into whether you’re in surplus or pulling from the grid.

Feeding that surplus into the grid would be technically trivial – but politically, it’s hardly welcome anymore. Instead of fully embracing renewable expansion, energy policy still leans on operation and even new construction of fossil power plants. For small-scale producers, that means: current feed-in is rewarded with 0 cents per kilowatt-hour, and serious discussion is underway about charging fees for feed-in. Which makes it all the more logical to consume your solar directly rather than donating it or even paying to get rid of it.

The consequence: once surplus exists, the system either starts additional loads – washer, dishwasher, AC – or charges a battery. Even with conversion losses on a separate inverter, it’s far more worthwhile than feeding into the grid under today’s conditions. Power stays in your own loop, reduces grid draw, and gets used optimally.

On that basis, automations go far beyond timers. If there’s enough sun and little else running, the air conditioner might push from 21 °C down to 19 °C instead of idling. Conversely, it might allow 23 °C if the washer is busy. Efficiency, not just comfort, becomes dynamic.

Expansion is straightforward. A simple wireless button can signal that the dishwasher is loaded and ready, or that the scooter battery is plugged in and needs charging at some point. Home Assistant then decides when: once sun is plentiful and no other big loads are active. And if it really has to be now, a double-click communicates: “I want it charged NOW.”

Modern kitchen devices from Siemens or Bosch, currently tied to Home Connect, can also be integrated locally in the future once APIs open up – step by step eliminating external clouds.

Beyond device control, Home Assistant’s dashboard is a real asset. It doesn’t just display sensors and switches, but also external data like calendar entries. That way, a small signal light next to the trash bins can switch on if the calendar says tomorrow is pickup day. The dashboard itself shows the reminder, too.

The same logic applies to alarms. Instead of the fridge just beeping when its door stays open too long, Home Assistant can tie that signal into a general warning light:

“Hey, something important is off – check the dashboard.”

And cloud? Not entirely banished. Optional services like weather forecasts significantly expand possibilities: the system knows not just current sun but how much more is expected. Yes, this depends on internet – but only occasionally. If the line works once a day, data is cached and local logic carries on, even if the connection drops.

Beyond a Single Apartment

Home automation doesn’t end at your own front door. In a multi-apartment building, concepts can expand dramatically – provided the metering setup allows it. The key is a shared main meter: if it exists, household controllers can actually cooperate.

Apartment 1’s hub might report surplus. Apartment 2 then shifts some loads to that time, so at least part of the energy gets used locally. If the main meter dips below zero, the system knows: there’s excess. A central battery can then charge instead of wasting energy into the grid.

Electric cars slot in naturally. Charging doesn’t follow rigid schedules but adapts: when rooftop solar is abundant, or when major household loads are idle. Heat pumps or even legacy heating can join, soaking up energy when the building provides it.

In this way, the building becomes an intelligent, cooperative system. Rooftop PV, balcony inverters, batteries, heat pumps, EVs, and apartment hubs all work together – locally, without external cloud services. And through each apartment’s controller, practically every consumer in every home becomes part of the system: washer, AC, dishwasher, lighting.

A further benefit lies in data analytics. From load profiles, weather patterns, and usage habits, patterns emerge that translate into new rules. Iteratively, the system becomes smarter: devices run at ever better times, batteries charge more efficiently, and energy flows continuously improve. Out of many small automations grows a constantly improving system – enhancing both comfort and sustainability.

The effect is clear: grid draw shrinks, and with it costs. In a pricing system where the electricity rate is set by the most expensive last supplier – usually a gas plant – this model is strikingly future-proof. Every kilowatt-hour produced and consumed locally bypasses that logic and directly reduces residents’ bills.

Why I Lay So Many Cables

So far we’ve talked about wireless protocols, gateways, and bridges. Useful as they may be – my heart belongs to cables. Anyone who knows me knows I run network lines generously in every renovation, often far more than seems necessary at first glance.

Why? Because cables are simply more reliable. The 2.4-GHz band is hopelessly crowded: Wi-Fi, Zigbee, Bluetooth, and countless proprietary systems all competing. Add to that the periodic frequency reallocations by international regulators – today’s standard might be tomorrow’s interference. With copper in the wall, that risk doesn’t exist.

Cables are also predictable and durable. True copper endures for centuries. The same material that carried DSL over two-wire doorbell cables still does today. And if DSL can squeeze through that, imagine what properly shielded Cat7 will do – not just today but twenty years from now.

Perhaps my obsession comes partly from history. In the GDR, copper was scarce – hence the reliance on Al-Cu or even full aluminum lines. That shortage has left an appetite that seems insatiable. Much like bananas: I can overeat on most foods until I can’t stand them anymore – but bananas always go. And so do copper cables.

At the end of the day, the sober fact remains: only cables are truly reliable. Wireless is a convenient supplement, but not a foundation you can count on for critical systems.

Conclusion

The cloud is, and will remain, the worst option for home automation. It may look convenient in the short term, but long-term it creates dependencies that are expensive, unreliable, and risky. The products mentioned here aren’t ads, just examples from my own setup – grown over years, from FS20 actuators to Zigbee lights, from Shelly meters to balcony inverters.

Wireless has its place, and in many scenarios it’s a practical addition. But for real stability, cables always win. A properly laid network line is immune to overcrowded frequency bands, shifting standards, or the whims of regulators. Copper lasts for decades, even centuries – and if DSL still works over two-wire bell cable, you can imagine what shielded Cat7 can deliver.

Of course, I know there are even better options. Fiber is the future – flexible, powerful, practically limitless. Just not yet as universally applicable or cost-efficient. One day it will take over. Just like IPv6. And we’ve only been telling ourselves that for, what, twenty years now?