Magic Room: Device-Free Sensing of People using RFID Tags & Machine Learning

WINLAB Summer Internship 2024
Group Members: Shriya Das, Xiang Meng, Sam-Fone Cheung, Aly Mustafa
Advisor: Aggelos Bletsas

Project Objective

Batteryless, ultra-low cost RFID (radio frequency identification tags) tags can serve as additional ultra-low cost antennas; variations of the multipath propagation inside a room offers impressive information for estimating the number of people and their tracks inside a room; WINLAB techniques, including machine learning, will be tested.

Weekly Progress

WEEK ONE-FOUR

For the first few weeks, our group worked solely on the Plant Doctor project, also led by our advisor Aggelos Bletsas.

WEEK FIVE

​Week 5 Presentation

Progress: During our first week of work, we put 150 RFID tags along two walls of the WINLAB conference room. Before putting them on, we recorded the ​distinct EPC (electronic product code) of each tag. We spaced the tags 16cm apart to minimize any interference from multiple tags based on their wavelength. We placed these tags in rows around 4 to 5 feet off the ground.

                                            

WEEK SIX

​Week 6 Presentation

Progress: By the end of Week 6, we finished putting 219 RFID tags around the conference room. We had to avoid large areas of the room such as the whiteboards due to metallic interference and the projector wall. Of these 219 tags, we consistently received 200-210 tags and the remaining tags were read infrequently. The low RSSI (Received Signal Strength Indicator) values of some of these tags may be a result of metal studs within the walls which requires further investigation.

                       

WEEK SEVEN

​Week 7 Presentation
Progress: This week we explored MATLAB code created by the prior group and recalibrated to the conference room.

WEEK EIGHT

​Week 8 Presentation
Progress: Adapted the existing code to adhere to FCC guidelines regarding frequency usage. Additionally, we started working on a 3D mapping system for the tags.