Decentralized Proximity Matching for Digital Contact Tracing

34 concepts25 papers1 proposal

Controlling the spread of an infectious disease requires public health officials to quickly identify and isolate individuals who have been exposed to a contagious person. Historically, this meant relying on manual interviews or invasive GPS tracking. However, broadcasting a user's real-time location history to a central server creates a massive surveillance risk, exposing sensitive movement patterns and social graphs to governments or third-party operators.

Decentralized proximity matching solves this problem by flipping the architecture of contact tracing. Instead of tracking where people are, it tracks how close they are to one another. Crucially, it shifts the computation of exposure risk entirely away from central servers to the individual user's device.

Relative Proximity via Ephemeral Tokens

Traditional location tracking relies on absolute coordinates (like GPS). In contrast, decentralized proximity tracing uses short-range wireless communication—primarily Bluetooth Low Energy (BLE)—to establish relative spatial tracking. As Trivedi explains in a review of digital contact tracing technologies, focusing on person-to-person proximity rather than geographic coordinates inherently prevents the creation of a centralized movement database.

However, if a smartphone continuously broadcasts a unique hardware identifier or static cryptographic key, malicious listeners could easily log that identifier at different physical locations to track the user over time. To prevent this, decentralized protocols rely on ephemeral, rotating identifiers.

[User A] --(Generates & Broadcasts Token 0x9F3A)--> [User B (Logs Token)]
   |
   v (15 minutes pass: Token rotates)
[User A] --(Generates & Broadcasts Token 0x2B8C)--> [User C (Logs Token)]

Devices generate pseudorandom tokens that change frequently (e.g., every 15 minutes). When two users pass each other in public, their devices exchange these temporary tokens via BLE and store them in a local, private diary. As highlighted by Saeed and Trieu et al., because these tokens rotate rapidly and contain no identifying information, an observer cannot link different tokens to the same device, preserving user anonymity during routine daily interactions.

Inverting the Server via Local Matching

The core cryptographic trick of decentralized proximity matching lies in how exposure is verified when a user tests positive. In a centralized system, an infected user would upload their contact diary to a server, which would then look up and notify the exposed individuals. A decentralized architecture completely inverts this workflow.

When a user receives a positive diagnosis, they voluntarily upload only the ephemeral tokens their own device generated during their infectious window to a public central registry. Because these tokens are random strings, the central server learns nothing about who the infected person met, where they went, or what their permanent identity is.

1. Infected User uploads their own generated Ephemeral Tokens to Central Server.
2. Central Server acts as a passive bulletin board.
3. Healthy User downloads the list of infected tokens.
4. Healthy User performs local matching against their own private diary of received tokens.

Every participant's device periodically downloads the updated list of infected tokens from the central registry. The device then performs a local comparison against its own private diary of received tokens. If a match is found, the device calculates the exposure risk locally based on the signal strength (as a proxy for distance) and duration of the encounter.

As emphasized by Buchanan et al. and Chan et al., this local matching mechanism ensures that sensitive interaction logs never leave the user's device. The central authority acts merely as a passive bulletin board, completely blind to which users are downloading the keys or which devices have triggered an exposure alert. Pahwa et al. note that this architecture fundamentally prevents the reconstruction of a user's movement history or social graph by any central entity.

Mitigating Physical-Layer Vulnerabilities

While rotating tokens protect identity at the application layer, broadcasting radio signals still introduces physical-layer vulnerabilities. For instance, sophisticated adversaries can fingerprint the unique hardware characteristics of a device's Bluetooth transmitter, bypassing token rotation. To eliminate this vector, researchers have introduced asymmetric and infrastructure-based communication models.

Abtahi proposes an asymmetric protocol (DP-ACT) that splits users into active (broadcasting) and passive (silent listening) nodes. Passive users do not emit any signals, making them completely invisible to hardware tracking or eavesdropping. If a passive user tests positive, they upload the tokens they collected from broadcasting peers. This allows the system to notify those contacts without the infected individual ever having broadcasted a trackable signal.

Alternatively, the broadcasting burden can be shifted entirely to fixed environmental infrastructure. Tedeschi et al. introduce a system using batteryless, energy-harvesting Bluetooth beacons placed in public spaces. In this model, smartphones operate in a passive, scan-only mode, eliminating battery drain and preventing the device from emitting trackable signals altogether. When infected, users upload their logs of received beacon IDs, and peers perform local matching against their own beacon histories to determine exposure.

Extending Decentralization to Shared Spaces

Peer-to-peer Bluetooth tracking struggles in scenarios where transmission occurs via contaminated air or surfaces in a shared room rather than direct proximity. To capture these "presence tracing" events without sacrificing privacy, the decentralized paradigm must be extended to physical venues.

Lueks et al. developed CrowdNotifier, a protocol that utilizes Identity-Based Encryption (IBE) to allow users to check into locations (like restaurants or classrooms) via QR codes. Instead of a central server maintaining a visitor log, matching is performed entirely on-device by downloading public tracing keys associated with an infected venue-window and attempting decryption locally. Similarly, Peng et al. demonstrate that hashed location identifiers can be stored locally and matched on-device, using accelerometer data to automatically manage check-outs without invasive GPS tracking.

Real-World Efficacy and the Research Trade-off

Does this decentralized approach actually work to curb disease transmission? Large-scale empirical studies confirm its viability. Evaluating the NHS COVID-19 app, Wymant et al. demonstrated that automated BLE alerts successfully identified random contacts in public spaces that manual tracing missed, preventing thousands of cases. Similarly, Salathé et al. provided early real-world proof-of-principle data from Switzerland showing that decentralized proximity detection successfully broke transmission chains at a level comparable to manual tracing.

However, the very privacy guarantees that make these systems ethical and secure create a fundamental challenge for public health research. As von Wyl et al. point out, because data is strictly decentralized and anonymized, researchers lack the ground-truth epidemiological data needed to easily validate and optimize the system's real-world parameters. Balancing this ethical imperative for privacy against the need for public health transparency remains a central theme in modern digital surveillance design (Ding et al.).

Key Foundational Work

Albatross: a Privacy-Preserving Location Sharing System (2015) — contributed 1 concept elements to this direction.
Secure Personal Memory-Sharing with Co-located People and Places (2016) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_privacy_preserving_contact_tracing_7e81e51f,
  title = {privacy-preserving contact tracing},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/privacy-preserving-contact-tracing/rdr_7e81e51f7d7a44bca8822a7e9ee5728e},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_privacy_preserving_contact_tracing_7e81e51f_v1,
  title = {privacy-preserving contact tracing},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/privacy-preserving-contact-tracing/rdr_7e81e51f7d7a44bca8822a7e9ee5728e/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake