Knock to Unlock — Mac Authenticator
Accelerometer-based Mac unlocking via iPhone knock patterns
// Built with
swift CoreMotion MATLAB CoreBluetooth macOS01 Problem
Needed a convenient, secure method for unlocking a Mac using a phone-based knock pattern recognition system, eliminating the need to type passwords when the phone is nearby.
03 Solution
Built a paired iOS and Mac application using CoreMotion accelerometer data for knock pattern detection, MATLAB for signal processing filter design, and CoreBluetooth (BLE) for low-energy communication between devices. The iOS app captures knock gestures, processes the accelerometer signal to extract patterns, and transmits unlock commands to the Mac companion app.
04 Outcomes
Qualitative
- Demonstrated feasibility of knock-pattern authentication on iOS
- Pioneered accelerometer-based gesture recognition for device pairing
- Applied signal processing expertise from Shopster ultrasound work to motion domain
06 Constraints
- Must reliably distinguish knock patterns from incidental movement
- Low latency BLE communication required for responsive unlocking
- Battery-efficient accelerometer monitoring
- Must work across varying phone orientations and surfaces
07 Innovation
Accelerometer-based knock pattern recognition with MATLAB signal processing — similar to Shopster's ultrasound approach but applied to motion data. Designed filters to isolate deliberate knock impulses from ambient device movement and noise.
Novelty: high
