Privacy-Protecting COVID-19 Exposure Notification Based on Cluster Events

TL;DR

This paper proposes a privacy-preserving COVID-19 exposure notification system based on cluster events, avoiding direct proximity detection, and utilizing location data and testing results to notify individuals of potential exposure.

Contribution

It introduces a novel cluster-based notification system that protects individual privacy by not linking events to specific persons or histories, differing from existing proximity-based methods.

Findings

Uses existing COVID-19 tests with quick results and high specificity.

Employs location tracking to identify cluster events without individual linkage.

Enables notification of exposure through public channels, protecting participant privacy.

Abstract

We provide a rough sketch of a simple system design for exposure notification of COVID-19 infections based on copresence at cluster events -- locations and times where a threshold number of tested-positive (TP) individuals were present. Unlike other designs, such as DP3T or the Apple-Google exposure-notification system, this design does not track or notify based on detecting direct proximity to TP individuals. The design makes use of existing or in-development tests for COVID-19 that are relatively cheap and return results in less than an hour, and that have high specificity but may have lower sensitivity. It also uses readily available location tracking for mobile phones and similar devices. It reports events at which TP individuals were present but does not link events with individuals or with other events in an individual's history. Participating individuals are notified of detected cluster events. They can then compare these locally to their own location history. Detected cluster events can be publicized through public channels. Thus, individuals not participating in the reporting system can still be notified of exposure. A proper security analysis is beyond the scope of this design sketch. We do, however, discuss resistance to various adversaries and attacks on privacy as well as false-reporting attacks.