Overview of my undergraduate thesis:
"Fall Detection Using RF Sensor Networks"

My undergraduate research involved using a wireless sensor network to detect if someone has fallen, which could be useful in assisted living facilities where some residents need constant monitoring.

Of course, determining someone's location, movement, and behavior is easier if you can attach a device to them or monitor them visually. However, my project falls under the category of device-free localization, which means detecting where people are without them having to wear a device.

Here's how it works:

An array of sensor nodes is set up in or around a room in a perimeter arrangement. These transmit and receive RF signals in the 2.4 GHz range, each node taking a turn to transmit. When a person is in the room their body affects the RF field, which causes variances in the signals the nodes receive from each other. By applying the appropriate algorithms to the data collected from the nodes, it's possible to determine where a person is.

The challenge for my thesis was to use this type of device-free localization to recognize a particular type of motion, namely, falling.

I developed a proof-of-concept fall detection system that does not require the user to wear a device. A two-level array of RF sensor nodes is used to determine a person's vertical position and motion inside the network, based on the received signal strength data measured by the nodes. The system essentially creates a 3-D image and uses a hidden Markov model to estimate the person's pose (standing, mid-position, or lying down) for every measurement sample. It then uses the time between standing and lying down to detect a fall.

The paper developed from this research was presented at the 2013 IEEE 24th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications in London. Or you can see the original senior thesis.

Even better, here's a nice recap of the research, based on my presentation.

This research received several mentions in the press, including the following:

Science & Enterprise
Science Daily
Fast Company
Healthline News
Scientific American
Business Standard

Overview of my master's thesis: "Maintaining Accuracy in Device-Free Localization Systems in Changing Environments"

Device-free localization (DFL) systems are used to locate a person in an environment by measuring the changes in received signal strength (RSS) on all the links in the network. A fingerprint-based DFL method, such as the type addressed in this thesis, collects a database of RSS fingerprints and uses a machine learning classifier to determine a person's location.

However, as the environment changes over time due to furniture or other objects being moved, the RSS fingerprints diverge further and further from those stored in the database, causing the accuracy of the system to suffer.

My goal with this thesis was to investigate the degradation over time of localization accuracy in RF sensor networks using a fingerprint-based method — and then see if there's a way to keep the accuracy from degrading.

The research determined that the random forests classifier performs the best as changes are made, compared to three other classifiers tested. In addition, a correlation method for selecting the channel used with each link improved localization accuracy from an average of 95.2% to an overall 98.4% accuracy using random forests.

All this is presented in the master's thesis, "Maintaining Accuracy in Device-Free Localization Systems in Changing Environments".