Finding Responsibility
summary
A Wi‑Fi geolocation app reconstructs every nearby hotspot in 3D space using a convolution of an R‑squared distance model and a Gaussian estimator, then optimizes three hardcoded constants with a genetic algorithm that drops error from about 45 feet to 10.3 feet; nested loops and repeated great‑circle calculations are memoized to speed up from seven minutes to a few seconds; a Kalman filter adds position/velocity/acceleration tracking and attempts at promiscuous‑mode phone sniffing reveal the end use: locating people for DoD operations.
AI-written summary · about
chapters (9)
- 0:00Start
- 4:17Gaussian Estimation
- 8:25What a Genetic Algorithm Is
- 8:51Fitness Function
- 10:18Overfitting
- 10:53Tracking a Moving Wi-Fi Access Point
- 12:48Promiscuous Mode
- 22:36Ethics and Morals
- 29:32Chelsea Manning
related talks
10mSo THAT’S how my phone knows where I am!
24mPinPoint: Localizing Interfering Radios
01h 20mLocation privacy: what problems - which solutions?
35mWhen Geo goes Wrong
16mTracking You from a Thousand Miles Away! Turning a Bluetooth Device into...
06mLeveraging Polarization of WiFi Signals to Simultaneously Track Multiple People
33mIs Your Timespace Safe? - Time and Position Spoofing Opensourcely
06mTowards Usable and Secure Location-based Smartphone Authentication