IPvSeeYou: Exploiting Leaked Identifiers in IPv6 for Street-Level Geolocation

TL;DR

This paper introduces IPvSeeYou, a novel privacy attack that accurately geolocates residential IPv6 hosts at street-level by correlating leaked MAC addresses with WiFi BSSIDs, revealing significant privacy vulnerabilities in current IPv6 deployments.

Contribution

The paper presents a new method for geolocating IPv6 hosts by exploiting leaked MAC addresses and correlating them with WiFi data, achieving street-level accuracy on a large scale.

Findings

Geolocated over 12 million IPv6 routers worldwide.

Achieved median geolocation error of 39 meters.

Demonstrated privacy risks in residential IPv6 deployments.

Abstract

We present IPvSeeYou, a privacy attack that permits a remote and unprivileged adversary to physically geolocate many residential IPv6 hosts and networks with street-level precision. The crux of our method involves: 1) remotely discovering wide area (WAN) hardware MAC addresses from home routers; 2) correlating these MAC addresses with their WiFi BSSID counterparts of known location; and 3) extending coverage by associating devices connected to a common penultimate provider router. We first obtain a large corpus of MACs embedded in IPv6 addresses via high-speed network probing. These MAC addresses are effectively leaked up the protocol stack and largely represent WAN interfaces of residential routers, many of which are all-in-one devices that also provide WiFi. We develop a technique to statistically infer the mapping between a router's WAN and WiFi MAC addresses across manufacturers and devices, and mount a large-scale data fusion attack that correlates WAN MACs with WiFi BSSIDs available in wardriving (geolocation) databases. Using these correlations, we geolocate the IPv6 prefixes of $>$12M routers in the wild across 146 countries and territories. Selected validation confirms a median geolocation error of 39 meters. We then exploit technology and deployment constraints to extend the attack to a larger set of IPv6 residential routers by clustering and associating devices with a common penultimate provider router. While we responsibly disclosed our results to several manufacturers and providers, the ossified ecosystem of deployed residential cable and DSL routers suggests that our attack will remain a privacy threat into the foreseeable future.