Project Title

Monitoring ADL Activities with Raspberry Pis Using WiFi Channel State Information

Presenters

Chris HuntFollow

Academic department under which the project should be listed

SPCEET - Engineering Technology

Faculty Sponsor Name

Dr. Fangyu Li

no human subjects

Abstract (300 words maximum)

With the rapidly growing popularity of the internet of things (IoT) devices, new technologies are becoming increasingly prevalent in the average home. Advances in human motion tracking show its potential to be a versatile sensor for gesture and location detection in IoT systems, but this impact is limited due to expensive equipment and complicated installation. Many tracking systems are impractical for the average consumer, while cheaper camera-based systems are often seen as an invasion of privacy. Research into WiFi channel state information (CSI) tracking shows promises in using WiFi devices to track movement and gestures without collecting video information, but many of these systems use modified hardware and/or laptops to extract CSI resulting in system dependence on specialized and expensive equipment. This project aims to test how effective a practical, low cost WiFi tracking system would be in detecting activities of daily living (ADL) by using Raspberry Pis and the Nexmon CSI extraction software. Raspberry Pi is a small and inexpensive alternative to a laptop or modified router. This makes Raspberry Pi based tracking systems a viable candidate for monitoring ADL. A tracking system using multiple Raspberry Pis will be tested for how accurately it can detect different types of motion. These tests will start by testing the system's capability to detect movement in an area and will later move to distinguishing a single user’s common motions like walking, sitting, lying down, and opening a door. If these tests are particularly successful, then tests for detecting walking direction and detecting multiple users will be conducted. The completed project is expected to be a small system consisting of four Raspberry Pis that can detect two to three specific actions performed by a single user with a high degree of accuracy.

Project Type

Event

This document is currently not available here.

Share

COinS
 

Monitoring ADL Activities with Raspberry Pis Using WiFi Channel State Information

With the rapidly growing popularity of the internet of things (IoT) devices, new technologies are becoming increasingly prevalent in the average home. Advances in human motion tracking show its potential to be a versatile sensor for gesture and location detection in IoT systems, but this impact is limited due to expensive equipment and complicated installation. Many tracking systems are impractical for the average consumer, while cheaper camera-based systems are often seen as an invasion of privacy. Research into WiFi channel state information (CSI) tracking shows promises in using WiFi devices to track movement and gestures without collecting video information, but many of these systems use modified hardware and/or laptops to extract CSI resulting in system dependence on specialized and expensive equipment. This project aims to test how effective a practical, low cost WiFi tracking system would be in detecting activities of daily living (ADL) by using Raspberry Pis and the Nexmon CSI extraction software. Raspberry Pi is a small and inexpensive alternative to a laptop or modified router. This makes Raspberry Pi based tracking systems a viable candidate for monitoring ADL. A tracking system using multiple Raspberry Pis will be tested for how accurately it can detect different types of motion. These tests will start by testing the system's capability to detect movement in an area and will later move to distinguishing a single user’s common motions like walking, sitting, lying down, and opening a door. If these tests are particularly successful, then tests for detecting walking direction and detecting multiple users will be conducted. The completed project is expected to be a small system consisting of four Raspberry Pis that can detect two to three specific actions performed by a single user with a high degree of accuracy.