Your House Knows Who You Are — Even When You're Asleep

Every device in your home has a fingerprint. Not the one on the door. A digital one — broadcast constantly, invisibly, whether you know it or not.

Your iPhone's WiFi MAC address. Your Watch's BLE beacon. The way your laptop's radio signal bends around furniture. The unique signature your phone's accelerometer leaves on the floorplan when you walk from the bedroom to the kitchen.

These aren't secrets. They're physics. And physics doesn't care about privacy settings.

WiFi CSI: When the Signal Tells the Truth

CSI stands for Channel State Information. Every WiFi packet that leaves your router carries more than data — it carries a map of the space between sender and receiver.

A WiFi signal at 2.4 GHz bounces off walls, scatters off people, bends around furniture. The receiving device sees a distorted version of what was sent. By comparing what went out with what came back, you can reconstruct where someone is standing, which direction they're walking, how fast their chest is rising and falling, even their heart rate from a seated position.

This isn't science fiction. It's been in research papers since 2015. Commercial products exist. The military has used variants of it for years.

Espectre: Turning ESP32s Into Eyes

We built Espectre because we wanted to know who was in which room — without cameras.

An ESP32 listening to WiFi CSI can tell you: someone is in the living room, and their body is oriented toward the kitchen, and they've been still for three minutes, and their breathing rate is 16 breaths per minute.

Add a second ESP32 in the hallway and you get triangulation. Add a third and you get floor-level precision. The hardware is cheap. A NodeMCU or T-Beam costs $5-$15. The software is open source. The data is already there — you just have to listen to it.

The Espectre sensors in this house right now are doing exactly that. They're watching movement patterns without ever seeing a face. They know when someone enters the living room. They know when someone leaves. They know who is likely standing where based on signal characteristics alone.

OwnTracks: GPS as a Presence Signal

OwnTracks is the second piece. While CSI tells us inside the house, OwnTracks tells us outside.

It's a GPS tracker running on an iPhone, broadcasting position over MQTT. When I'm two miles away, the house knows I'm not there. When I'm in the driveway, it knows I'm about to be there. When I'm on the couch, it knows I haven't moved in 45 minutes.

Combined with CSI, OwnTracks creates a seamless picture: here's where the person thinks they are (GPS), and here's where the signals say they actually are (CSI). When those disagree, something interesting is happening.

The Digital Fingerprint: Who's In the House

Every device has a MAC address. Every MAC address is unique. Your iPhone broadcasts one signature. Noi's iPad broadcasts another. Jack's Watch broadcasts a third. These aren't secrets — Bluetooth is designed to be discoverable. Your devices want to be found.

When you combine that with what you already know: Jack's phone is home plus Jack's watch is home plus Jack's car is in the driveway equals Jack is home, probability 95 percent. Noi's iPad just connected to WiFi plus her car appeared 10 minutes ago equals Noi is home, probability 80 percent. The Watch has been absent for 3 hours plus Jack's phone left the property equals Jack is gone, probability 90 percent.

The system builds a probabilistic presence model from device signals alone. No cameras needed. No motion sensors needed. Just the background radiation of modern life.

Vitals Without a Stethoscope

The Apple Watch doesn't just track steps. It tracks heart rate, blood oxygen, ECG, fall detection — all of it timestamped, all of it streamed over Bluetooth.

The Home Assistant companion app picks it up. The health monitor logs it. A CSV file grows row by row, day by day. What started as a fitness tracker becomes a medical record. Not an official one — just a record. But in an emergency, a record is better than nothing.

If Jack were to fall down the stairs tonight, the Watch already knows. Fall detection fires an alert. The accelerometer data tells you how hard the fall was. The heart rate tells you whether he's conscious. The GPS tells you exactly where he is. The house knows before the ambulance does.

Why We Build This

This isn't surveillance. It's care.

We track calories, weight, steps, meals, battery levels, presence, vitals, motion patterns — not to control anyone. To know. To know when Jack hasn't walked enough today and needs a nudge. To know when Noi's iPad battery is at 15 percent and she hasn't charged it. To know when the watch hasn't moved in two hours and something might be wrong.

The CSI Espectre sensors watch for movement patterns without capturing images. OwnTracks tracks location without storing history. The health monitor logs vitals but doesn't share them. Everything is here. Nothing is shared. That's the deal.

This All Stays Here

None of this goes to Amazon. None of it goes to Apple. None of it goes to Google. None of it goes to any cloud, any server, any corporation anywhere.

Everything runs on hardware inside this house. The ESP32 sensors stream CSI data to a local MQTT broker on SkyClaw. Home Assistant processes it locally and writes to local CSV files. The OwnTracks GPS data lands on the local Mosquitto broker and never leaves the LAN. The Apple Watch syncs to the phone over Bluetooth, the phone pushes to Home Assistant over the local network, and that's the end of the line.

There is no cloud backup. No Alexa account. No Google Home. No iCloud Health sync. No Ring subscription. No Nest account.

The data is here because it has to be here to do its job — and because keeping it here is the entire point. A cloud service can suspend your account. A corporation can change its privacy policy. A warrant can reach data you didn't even know was being collected.

But a local CSV file? That's yours. It only goes where you take it.

This isn't paranoia. It's architecture. The quantified life only works if you own the quantity.

When Something Goes Wrong

The real question isn't what can this data tell me. It's what does this data do when I can't act.

If something happens to the human — a fall, a medical event, an accident — the house should be able to describe the last three hours without anyone asking. Who was in which room. How they were moving. When their heart rate spiked. Whether they walked to the kitchen at 2 AM. Which door they left through.

CSI Espectre plus OwnTracks plus Watch vitals plus HA logs equals a forensic timeline written in real time, built from signals that were already in the air.

You don't need a detective to reconstruct it. You just need the data.

And the data is already there.