Say Goodbye to "Lights Turning on Automatically at Midnight": C4002 MmWave Sensor Anti-Interference Field Test and Installation Guide

Foreword | Why Conduct This Test?

When using millimeter-wave human presence sensors, the biggest headache for many users isn't "failing to detect people," but rather the false triggers caused by being "over-sensitive":

1. Do your lights turn on automatically in the middle of the night when the room is empty?
2. Does a fan blowing on curtains or swaying plants get mistaken for "human presence"?

I recently acquired the C4002 millimeter-wave sensor. Below, we test whether its "Environmental Calibration" function actually works and if it can eliminate these "ghosts"

Test Scheme | How to Verify the Above Issues?

To verify the environmental calibration function of this module, three groups of control experiments were designed:

1. Group 1: Completely Empty Room
2. Used to verify if environmental calibration can reduce false alarms against an "ideal unmanned background".
3. Group 2: Person Sitting Still in the Room
4. Used to verify if environmental calibration sacrifices sensitivity for identifying stationary bodies.
5. Group 3: Fan + Plant Interference Environment
6. Simulating a real family living room scenario to test the practical value of environmental calibration against a complex background.

Test Preparation | Conditions and Setup

1、Device: Fermion: C4002 mmWave Human Presence Sensor - Static & Motion Detection for Home Assistant (10m)

2、Environment: Closed empty meeting room.

Important Note: The test site has lightweight partition walls (drywall), not solid concrete. This factor proved to be critical in the analysis later.

3、 Duration:

1. 3 repeated rounds per scenario
2. Duration per round: 10 minutes

4、 Interference Sources: (Simulating typical living room interference)

1. Oscillating fan
2. Green plants
3. Blowing curtains

5、 Core Metrics:

1. False Positive Rate (FPR): The number of times presence is falsely triggered/time ratio when no one is present.
2. Robustness: The change in the false positive rate after introducing interference sources.

Group 1: Completely Empty Room (The "Wall Penetration" Surprise)

Note: The test site has a lightweight partition wall structure (non-solid wall), a factor that proved crucial in the subsequent analysis.

Control Conditions: Uncalibrated VS Executing Environmental Calibration

1. Without Environmental Calibration
2. Total False Alarm Rate:1.90%
3. MOVE False Alarm: 1.90%
4. EXIST False Alarm:0%
5. After Executing Environmental Calibration
6. Total False Alarm Rate: 4.92% (Increased instead of decreased)
7. MOVE False Alarm:3.74%
8. EXIST False Alarm:1.18%

Result Analysis and Troubleshooting:

The false alarm rate actually rose after calibration, a result that is clearly counter-intuitive. After the test, we contacted DFRobot official technical support for troubleshooting. By collecting environment thresholds, we found that:

1. The environment threshold remained above 90 for long periods (Upper limit is 99).
2. It was further confirmed that the test meeting room had non-solid partition walls.

Final Confirmation: People walking in the corridor outside the wall were successfully detected by the C4002.

This indirectly confirms that the C4002 mmWave radar possesses extremely strong penetration detection capabilities (can penetrate approx. 10 cm lightweight partition walls).

If the back or side of the mmWave sensor faces a lightweight partition wall, glass wall, or public corridor, you should avoid through-wall false alarms by adjusting the installation direction, setting detection distance intervals, or using software algorithms.

Group 2: Person Sitting Still in the Room

Test Method:

1. The tester sits specifically in front of the mmWave sensor at distances of 1 meter and 1.5 meters respectively.
2. Tester sitting still approx. 1 meter directly in front
3. 1 Meter Position: Effective recognition rate 99.72%
4. Tester sitting still approx. 1.5 meters directly in front
5. 1.5 Meter Position: Effective recognition rate 99.89%

Comparison Result:

After executing environmental calibration, the detection capability for a stationary human was almost unaffected. There is no significant issue of "reduced sensitivity".

Group 3: Fan + Plant Interference (Real Home Scenario)

Control Conditions: No Calibration VS Enabling Environmental Calibration under interference

1. Environmental Calibration Disabled
2. Total False Positive Rate:5.81%
3. MOVE False Alarm: 3.48%
4. EXIST False Alarm: 2.34%
5. Environmental Calibration Enabled
6. Total False Positive Rate:4.38%
7. MOVE False Alarm:2.62%
8. EXIST False Alarm: 1.75%

Result Analysis:

In an environment with complex interference sources (fan blowing plants and curtains), environmental calibration is effective. It suppresses background noise to a certain extent, reducing the false positive rate by approximately 1.5 percentage points.

For details, see the above image.

Question 1: Does environmental calibration reduce false alarms?

1. Effective in typical home interference environments (Reduces approx. 1.5%).
2. May amplify through-wall interference in environments with lightweight partition walls.

Question 2: Does it affect stationary human recognition?

1. Recognition rates at 1–1.5 meters remain close to 99.8%.

Question 3: Is it worth enabling environmental calibration?

1. For home environments with dynamic interference such as fans, curtains, and plants, it is strongly recommended to enable it.
2. Prerequisite: The installation location must be reasonable, avoiding facing public areas or partition walls.

Overall Evaluation: The C4002 is a mmWave sensor with extremely high sensitivity and excellent presence detection capabilities. However, "high sensitivity" also means it has higher requirements for the installation environment. Compared to whether environmental calibration is enabled, reasonable point selection and installation direction are often more critical.

Appendix | Data Visualization and Field Explanations

1. Since a large amount of data was acquired, you can use the visualization webpage below to upload the CSV values detected by the sensor for display.

Main Data Fields Include:

1. time_s: Time since the test started (seconds)
2. state: Target state
3. exist_dist: Existing target distance (meters)
4. exist_energy: Existing target energy value (intensity)
5. move_dist: Moving target distance (meters)
6. move_speed: Moving target speed (m/s)
7. move_energy: Moving target energy
8. enter_exist: Cumulative times entering "Exist State"
9. enter_move: Cumulative times entering "Move State"
10. state_changes: Total number of state changes