Smart wallet

Security, Detection and Analytics system

We have successfully built a smart wallet with robust security, detection and analytics funcitonalities. It has two main modes of operations, a short range mode where it communicates via a bluetooth low energy (BLE) module with the smart wallet android app and a long range mode where it communicates with a remote cloud server via any open Wifi network, seamlessly switching between the two. Short range functionality include detection by sounding an alarm and notifiaction when it goes out of range. When not in BLE range it can be tracked on the smart wallet app using the google maps API. It simultaneously runs a gait analysis algorithm using an IMU to check for the gait signature of the person carrying the wallet and pushes an alert to all interested parties if the wallet is opened and the signature does not match that of the user. Finally, it performs analytics based on spending patterns and suggests places such as restaurants and provides provides other useful information based on where the wallet was openend.


  • Every year nearly 100 million dollars is lost in the form of theft and accidental misplacing of wallets.
  • The wallets contain sensitive information in form of ID cards, credit cards and driver’s license (and maybe even library cards).
  • Around 15 million people are victims of identity theft. Fraudulent use of the identities from stolen wallets has led to financial losses of over 5 billion
  • No commercial version of the product exists with all the advanced functionality we have implemented
  • Huge market and good funding opportunities


  • Short range (in Bluetooth range): Wallet can be located when in range or notified when out of range using the Bluetooth module.
  • Long range (out of Bluetooth range): It can be located and tracked using cloud geolocation APIs.
  • Gait match: can determine if the wallet carrier is the user based on gait signature.
  • Wallet state: can determine if the wallet has been breached or not depending on open/closed state.
  • If gait matches, then user has left his phone.
  • If gait does not match and wallet is closed, alert that somebody else has wallet.
  • If gait does not match and wallet is open, alert that somebody has breached the wallet.


Technical Components

  • Feather Huzzah ESP8266
  • HC-06 Bluetooth Module
  • ADXL345 Inertial Measurement Unit
  • Piezoelectric Buzzer
  • Android Mobile App
  • Amazon AWS Cloud Database



This is the resulting page from when bluetooth is in range and current location has been logged more than a certain number of times while the wallet has been opened. The app places markers in nearby restaurants to the logged locations as suggestions.

This is the resulting page from when the bluetooth is out of range. The blue marker shows the location of the wallet and the red marker shows the location of the user.

The left plot shows the feature vectors extracted from the accelerometer data within one time window. In our case the feature vector calculated based on the angle or swing in the x and y directions. If either of the vectors exceeds the threshold, then the gait signature will no longer be matching, and the appropriate alerts will be pushed depending on the wallet state. The right plot shows the varying accelerometer data as a person is walking.

This picture shows the instantaneous accelerometer data received from the Huzzah.


  • ”Walli the smart wallet”
  • “The next-generation wallet”
  • ”Identity theft Statistics”

Our Team

Brian Chieu

EE MS, Columbia

Abhinav Sridhar

EE MS, Columbia

Yidong Ren

EE MS, Columbia


Abhinav Sridhar:
Brian Chieu:
Yidong Ren:

Columbia University Department of Electrical Engineering
Class Website: Columbia University EECS E4764 Fall '16 IoT
Instructor: Professsor Xiaofan (Fred) Jiang