ZPI: One approach to rule them all

This blog post presents z9, a machine‑learning network intrusion detection engine that uses a Zero Packet Inspection (ZPI) approach to detect TCP/IP/UDP/ARP scans on Android, iOS and Windows. It describes the data collection methods per platform, the training pipeline (feature optimization, denoising, classifier training), reports high accuracy via confusion matrices and nmap test results on real devices, and promotes ZPI as a privacy-preserving, low-power IDS applicable to mobile, IoT, routers and other OSes while inviting peer review.