Human Motion Detection using WiFi
Human motion detection (HMD) is a domain of research in which human motion is recognized using various methods. These methods include sensor-based, vision-based and Wi-Fi based methods. In our project, we are using Wi-Fi-based method. HMD?s popularity is increasing in practical applications includin
2025-06-28 16:27:43 - Adil Khan
Human Motion Detection using WiFi
Project Area of Specialization Electrical/Electronic EngineeringProject SummaryHuman motion detection (HMD) is a domain of research in which human motion is recognized using various methods. These methods include sensor-based, vision-based and Wi-Fi based methods. In our project, we are using Wi-Fi-based method. HMD’s popularity is increasing in practical applications including smart homes, user authentication service, healthcare monitoring, and smart space management.
HMD through sensors in contrast to HMD through Wi-Fi, increase inconvenience for users since it requires them to wear smart devices on bodies. In addition, vision-based methods are easy to leak personal privacy and are limited to Line-of-Sight (LOS) condition. In our project we are overcoming these limitations by sensing HMD through Wi-Fi.
The fundamental concept around HMD using Wi-Fi sensing is that when a person moves, the motion of their body will affect the communication channel in terms of signal attenuation and phase shift. Wi-Fi signals are received using Software Defined Radios (SDR). SDRs generate IQ samples of the received signal and these samples are further processed to extract Channel State Information (CSI) values using CSI tools.
The ultimate outcome of our project includes extraction of CSI values for motion and no-motion and then creating their respective plots. Plots of these CSI values will enable us to detect motion.
Project Objectives- Dataset Acquisition: Create a dataset of CSI values for motion and no-motion
- Data Pre-processing: Create Doppler Spectrograms
- Training Deep Learning Model: Train a deep learning model through spectrograms to differentiate between motion and no-motion
First, a dataset of motion and no-motion is created, for Wi-Fi based recognition. Then the dataset is analyzed for calculating Doppler shift frequencies through channel state information (CSI). Once Doppler shift frequencies are calculated, they are used to create spectrograms, known as Doppler spectrograms. These spectrograms are then used to identify motion versus no-motion. Each activity will have a unique spectrogram, which will help in identifying it. These spectrograms will be used to train deep learning models which include Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM).
Benefits of the ProjectHuman motion detection (HMD) can be implemented using various methods like sensor-based, vision-based and Wi-Fi based methods. In our project, we are using Wi-Fi-based method. HMD’s popularity is increasing in practical applications including smart homes, user authentication service, healthcare monitoring, and smart space management.
HMD through sensors in contrast to HMD through Wi-Fi, increase inconvenience for users since it requires them to wear smart devices on bodies. In addition, vision-based methods are easy to leak personal privacy and are limited to Line-of-Sight (LOS) condition.
Technical Details of Final Deliverable- A dataset for motion and no-motion: One of our final deliverable will be a dataset which will contain CSI values for motion and no-motion.
- Spectrograms for training the model: Another deliverable will be a set of spectrograms for motion and no-motion repectively to train the model.
- A trained deep learning model: Train a deep learning model through spectrograms to differentiate between motion and no-motion.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 80000 | |||
| GPU NVIDIA GTX 750ti DDR5 128bit | Equipment | 1 | 25000 | 25000 |
| Atheros Routers | Equipment | 2 | 6000 | 12000 |
| External Intel NIC 5300 Card | Equipment | 1 | 1000 | 1000 |
| Udemy Courses | Miscellaneous | 2 | 5000 | 10000 |
| Tp-Link Routers | Equipment | 2 | 2500 | 5000 |
| Node MCU | Equipment | 2 | 1000 | 2000 |
| Antenna Vert 2450 | Equipment | 1 | 25000 | 25000 |