Now that we are officially returning to campus next fall, bag theft is now once again a real possibility for Penn students, and this project serves as a very useful tool to combat this. Also, having experienced theft in the past, we think this bag alarm would actually be very practical for us personally. The system, when attached to the bag (primarily the inner zipper), is triggered by unusual movement that is detected by an accelerometer. This thereafter activates a buzzer (acting as an alarm), and sends a notification from the nodeMCU to the Blynk app on the user’s phone.
We ran into the issue of the alarm triggering for any movement. When a bag is stolen, it is a prolonged period of movement, not a quick one. We, therefore, decided to gather all accelerometer data for 3 seconds and process that. If a movement threshold is reached, we want an alarm to trigger. Since the accelerometer is rotational data, we couldn't just calculate the movement from raw values. Instead, we decided to take the standard deviation of the 3-second array of data, which will tell us how much the rotational data has changed over time.
By using the standard deviation to calculate movement, we were able to limit false alarms. Whether you are reaching for a pen or adding a folder, the alarm won't trigger. After a movement threshold, the alarm will sound and a notification will be sent to the blynk app from the nodeMCU.