Don't Get Distracted
summary
Builds a WiFi geolocation app for a DoD contractor that triangulates phone signal sources using a convolution of R² error and Gaussian estimation, speeds a matrix-based Java implementation from seven minutes to seconds by memoizing great‑circle distance calls, reduces median error from about 14 meters to ~3 meters with a genetic algorithm for constants, adds Kalman filters to track moving access points and multiple hotspots, attempts promiscuous‑mode sniffing (requires jailbreaking), and concludes the tech works well enough to locate people and raises ethical red flags.
AI-written summary · about
chapters (5)
- 0:00Start
- 7:56A Genetic Algorithm
- 9:34Genetic Algorithms
- 10:00Add Tracking a Moving Wi-Fi Access Point
- 23:32Avoid Unintentional Harm
related talks
29mIncentives and Games in Large Scale Mobility
15mA Closer Look: Evaluating Location Privacy Empirically
26m"Reality Mining for Compan
24mBuilding Aggregated Solutions to Scale Your Geospatial Strategies
19mDigital Trace Data to Measure Human Mobility
23mLocation Data and the Future of Media Buying
29mSoftware defined radio based Synthetic Aperture noise and OFDM (WiFi) RADAR mapping
48mUrban Data & Decisions