For this project, we were tasked with applying a numerical method to a real-world situation. We chose to attempt to identify individual gaits based on vibration data of their footsteps. The purpose of this project is to better understand if and how a person can be identified based on their gait. This biometric data has implications in many fields, including security and medical fields. 

Because the vibrational data is very noisy due to environmental factors, and is difficult to interpret, we apply a fast fourier transform in to make the data easier to interpret. In order to reduce noise, we use a moving average filter. As an additional challenge, we chose to implement the fast fourier transform manually, using only simple MATLAB commands. 

We conducted testing on a metal grate due to increased vibration, because we didn't have a sensor that was sensitive enough to identify the vibrations on more vibration-dampening surfaces such as wood or tile. We used Phyphox, a phone app, to gather accelerometer data from our phone in order to get vibration data. We also assessed multiple participants and walking styles in order to understand the effects of participant factors and walking styles on vibrational patterns. 

Our code was capable of approximating step timings of any vibrational data. This could then be matched up to the time domain data to verify the approximation. We determined that it was very difficult and unreliable to distinguish between individual gaits. We did determine, however, that it was possible to distinguish between different types of gaits, and that different individual gaits do have slightly different frequencies and magnitudes. It is also possible to identify basic physical features of a person (ex. height, weight) based on their gait.