📡 WiFi CSI Radar (V1.43)

Radar V1.43 is a high-performance Wi-Fi CSI (Channel State Information) sensing tool specifically optimized for the M5Stack Cardputer. It utilizes advanced radio-wave analysis to detect motion and spatial presence through walls and obstacles.

Important

Source Code Status: This project is proprietary. The source code is private. Distribution: Binary only via M5Burner.

---

⚡ Technical Highlights

• Dual-Core FreeRTOS Architecture: Networking and high-speed injection are handled on Core 0, while heavy Math and UI rendering run on Core 1 for zero-lag, real-time sensing.
• Ultra-Precise 1D Sensing Core: The original and most accurate engine. Specifically tuned for high-fidelity motion detection, capable of sensing subtle movements (like breathing or micro-vibrations) through walls.
• Hybrid Room Mapping (SLAM) with Smart Geometry: Integrated IMU logic to create a digital layout of your room. V1.43 introduces an Angle-Based Segment Merging engine that mathematically aggregates collinear steps into perfectly straight walls and automatically trims/closes polygon gaps.
• 2D Spatial Profiles: Support for training up to 21 unique spatial zones.

Note

2D Mode is currently in Active Beta. Accuracy is being refined. The system now features a default 110x110px geometric tracking grid if no room is mapped.

• External NAT Bypass: A custom injection engine that routes traffic through external DNS targets to bypass router flood protection and rate-limiting.
• Deep SAFE MODE: Intelligent power management. When charging, the device drops CPU frequency to 80MHz, suspends IMU polling, and halts all heavy CSI processing to ensure safe, cool, and rapid charging.

---

🛠 Installation

1. Open M5Burner.
2. Search for WiFi CSI Radar or Zeloksa.
3. Select version V1.43.
4. Burn to your M5Stack Cardputer.

---

🕹 Controls & Usage

• [ ENTER ]: Toggle Radar / Start Calibration / Hold to Reset Map
• [ i ]: Open Interactive User Manual (3 Pages)
• [ D ]: Toggle Web Server & IP Overlay
• [ R ]: Switch between 1D Radar / 2D Spatial Mode
• [ H ]: MAX Mode (Force 240MHz CPU & Maximum Polling) / AUTO Mode
• [ , / . ]: Manual Frequency (Hz) Adjustment
• [ - / = ]: Volume Control (Base volume normalized to 30%)
• [ [ / ] ]: Hardware Screen Brightness Control
• [ ESC / ` ]: Back / Exit Current Mode
• [ DEL ]: Factory Reset (Clear Wi-Fi credentials, works in Main Menu)

---

📖 Comprehensive User Manual

To get the most accurate readings from the CSI Radar, you must understand how Wi-Fi signals bounce in your environment. Follow these steps for each mode:

🎯 1D Mode: High-Precision Motion Detection

This is the default mode. It acts as an ultra-sensitive invisible tripwire.

1. Positioning: Place the Cardputer on a stable, flat surface. Do not hold it in your hands.
2. Start Calibration: Press [ENTER]. The screen will show a 12s countdown.
3. The "Ghost" Rule: Immediately step away from the device and stand perfectly still, or leave the room. The radar needs these 12 seconds to learn the "static baseline" of the empty space.
4. Armed State: Once the countdown disappears, the alarm is armed. Any movement (breathing, walking) will disrupt the CSI baseline, spike the graph, and trigger the alarm.

🗺️ 2D Mode: Spatial Profiles (Active Beta)

This mode attempts to locate where the movement is happening using the Web UI.

1. Switch to 2D Mode by pressing [ R ].
2. Press [ D ] to enable the Web Server and note the IP address on the screen. Open this IP in your smartphone or PC browser.
3. Train the Void (Crucial): In the Web UI, click TRAIN EMPTY. Everyone must immediately leave the room for 7 seconds. The radar must memorize the completely empty room baseline (Profile 20).
4. Train the Corners (5-Step Profiling): The radar needs to learn your Wi-Fi reflection from multiple angles. Go to the Top-Left (TL) corner of your room and open the TL menu in the Web UI. You must record 5 states for this corner:
   • Arrows (⬆️ ⬇️ ⬅️ ➡️): Click an arrow and stand perfectly still facing that direction for 7 seconds.
   • Center Button (🔄): Click the center icon, and rotate slowly 2 times in place while the radar beeps.
5. Repeat this complete 5-step process for all 4 corners of the room (TL, TR, BL, BR).
6. Once all 20 spatial profiles are trained, the tracking dot on the Cardputer and Web UI will interpolate and track your physical position in real-time.

📍 Hybrid SLAM: Room Mapping

Use the onboard IMU to draw a digital floor plan of your room.

1. Switch to Mapping Mode via the main menu.
2. Posture: Hold the Cardputer perfectly flat against your chest, screen facing outward.
3. The Walk: Press [ENTER], then walk at a steady, robotic pace. Step firmly on your HEEL with each step (the IMU uses the physical shock to count steps).
4. Turns: When you reach a corner, stop, turn exactly 90 degrees on the spot, and continue walking.
5. You will hear a beep for every successfully registered step. Click STOP in the Web UI (or press [ENTER]) when you have closed the loop. The Smart Geometry engine will automatically trim crossing lines and close the perimeter.

Warning

Environmental Variables: Wi-Fi CSI is highly sensitive. Moving furniture, opening doors, or even heavy rain outside can alter the baseline. Always recalibrate the 1D or 2D profiles if the environment changes.

Tip

MAX Mode significantly increases detection precision and resolution. Warning: This mode results in faster battery drain and slight device heating due to high CPU and Wi-Fi load.

---

🆕 V1.43 Changelog

• Total UI/UX Reconstruction: Every screen has been re-designed from the ground up for better visual depth, improved readability, and professional-grade performance.
• Advanced Room Mapping Engine:
  • Wall Length Measurements: The SLAM logic now automatically calculates and displays the length of each wall ($m$).
  • Area Calculation: Integrated Gauss's area formula (Shoelace algorithm) to provide real-time room size calculation in square meters ($m^2$).
  • Smart Geometry: Added a mathematical engine to auto-close polygons (undershoot correction), trim overlapping tails (overshoot correction), and aggregate collinear steps into clean, single-segment walls.
• Zero-Latency Navigation: Re-engineered the State Machine to use front-edge trigger detection. Interaction is now 100% instant, eliminating the "heavy" feeling of the keyboard and adding protection against phantom double-clicks during state transitions.
• UI Detail Refinement:
  • Rebuilt the [ i ] interactive manual into a comprehensive 3-page guide covering all hidden hotkeys and mode-specific instructions.

---

💬 Feedback & Suggestions

If you find a bug or have a suggestion for the next version:

1. Go to the [Issues] tab at the top of this page.
2. Click [New Issue].
3. Describe your problem or idea in detail.

---

☕ Support the Project

If this tool has been useful for your research or hobbyist projects, consider supporting further development:

• https://boosty.to/zeloksa

---

Created by Zeloksa. Optimized for Cardputer.