Architectural Separation of Identity and Sensitive Data

23 concepts19 papers1 proposal

When a single centralized system stores both a user's real-world identity and their behavioral history, it creates a structural vulnerability. Whether through insider threats, corporate profiling, or external data breaches, combining who a user is with what they are doing poses a severe risk to privacy.

To solve this, a prominent research direction focuses on the architectural separation of duties. Instead of relying purely on policy or standard encryption, these systems split responsibilities across independent, non-colluding entities. By ensuring that no single party simultaneously possesses both the identity keys and the behavioral data streams, the architecture itself guarantees user privacy.


The Core Mechanism: Splitting the Persona

The fundamental intuition behind this approach is to break the link between a user's identity and their actions by distributing data across distinct architectural layers.

[ User ] ───► [ Identity Authority ]  (Knows WHO you are, but not WHAT you do)
         ───► [ Service Provider ]    (Knows WHAT is done, but not WHO did it)

Consider an Electronic Toll Pricing (ETP) system, where vehicles must be billed based on their physical location history. In a traditional system, the toll operator tracks the vehicle's exact trajectory, creating a massive surveillance footprint.

To prevent this, Xihui Chen et al. proposed a dual-entity architecture in A Group Signature Based Electronic Toll Pricing System and its subsequent formal verification, Design and Formal Analysis of A Group Signature Based Electronic Toll Pricing System. Their framework splits system responsibilities between an operational service provider and an independent identity authority:

  • The Service Provider collects transactional and spatial data to calculate bills but only sees randomized pseudonyms or group signatures.
  • The Identity Authority manages user credentials and cryptographic keys but never sees the operational or tracking data.

Because neither party possesses both datasets, the system prevents unilateral deanonymization. The only way to link a behavioral trace back to a physical identity is through an explicit, auditable legal trigger that forces cooperation between both entities.


Zero-Knowledge Relays and Multi-Service Isolation

When applications require real-time data routing—such as location sharing or crowdsourced tracking—the central infrastructure can be structurally blindfolded to act purely as a message relay.

In location-sharing platforms, Marcello Paolo Scipioni demonstrated in Towards a New Privacy-Aware Location Sharing Platform that end-to-end encryption can render intermediary servers entirely unaware of raw coordinates. Building on this concept for crowdsourced hardware tracking (like Bluetooth item locators), Chinmay Garg et al. showed in Toward Secure Crowdsourced Location Tracking System that treating the service provider as a zero-knowledge relay introduces negligible latency while completely eliminating the threat of mass location profiling.

For complex enterprise applications, this isolation is achieved by partitioning data attributes across microservices. In designing the architecture for LinkedIn Salary, Krishnaram Kenthapadi et al. detailed in LinkedIn Salary: A System for Secure Collection and Presentation of Structured Compensation Insights to Job Seekers how sensitive compensation attributes and user identities are split across isolated services using distinct encryption keys. This defense-in-depth strategy ensures that even an operator with administrative access to one component cannot link a user's identity to their private financial contributions.


Decoupling Social and Spatial Networks

The challenge intensifies when data types are inherently relational, such as in mobile social networks where users want to share their locations with friends. If a single server handles both social graphs and spatial queries, it can easily reconstruct comprehensive user profiles.

To mitigate this, researchers partition the data types across non-colluding providers:

  1. The Social Server: Manages user profiles, friendships, and social ties, but has no access to real-time location data.
  2. The Location Server: Processes spatial queries and distance comparisons using randomized pseudo-identities, completely blind to who those pseudonyms represent.

The authors of PPLS: a privacy-preserving location-sharing scheme in mobile online social networks (Chang Xu et al.) and MobiShare+: Security Improved System for Location Sharing in Mobile Online Social Networks (Jingwei Li et al.) demonstrated that by leveraging homomorphic encryption and private set intersection across these split servers, systems can compute proximity alerts between friends without either server learning both the identity and the location.


Generalizing to Health, Analytics, and Governance

The principle of separating identity from behavioral logging extends naturally to high-stakes domains like medical research and smart city governance.

Digital Phenotyping and Public Health

In mobile health tracking, Scott Barnett et al. introduced a dual-key enrollment protocol in Intelligent Sensing to Inform and Learn (InSTIL).... The platform assigns unique identifiers within the primary cloud repository while storing re-identification keys in an entirely separate, external administrative store. Similarly, for tracking infectious disease outbreaks, Jianqing Liu et al. demonstrated in A Cloud-Based Secure and Privacy-Preserving Clustering Analysis of Infectious Disease that distributing encrypted health data and location data across independent cloud providers prevents any single host from correlating distinct data categories.

High-Resolution Geolocation Studies

When collecting continuous smartphone sensor data for statistical analysis, Mattia Zeni et al. showed in Improving time use measurement with personal big data collection... that separating identity-linked metadata from raw sensor streams structurally protects participants from re-identification during large-scale data processing.

Decentralized Governance and Multi-Party Trust

In decentralized environments, the separation of duties can be enforced via multi-layer architectures. Emmanuel Boateng Sifah et al. designed a blockchain framework (BEMPAS: A Decentralized Employee Performance Assessment System...) that decouples identity management, behavioral logging, and credit allocation into distinct architectural layers.

Even when an anonymous identifier must be unmasked for lawful surveillance, the separation principle holds. Masoud Naderpour et al. demonstrated in Auditable De-anonymization in V2X Communication that the process of identity resolution can be distributed across multiple independent entities, ensuring that no single authority can unilaterally perform surveillance without technical and legal consensus from the other components of the chain.

Key Foundational Work

A Group Signature Based Electronic Toll Pricing System (2011) — contributed 2 concept elements to this direction.
Towards a New Privacy-Aware Location Sharing Platform (2011) — contributed 2 concept elements to this direction.
Anonymous credit cards and their collusion analysis (1996) — 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_separation_of_duties_538ee218,
  title = {separation of duties},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/separation-of-duties/rdr_538ee21821544648b488a3d71c7c361c},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_separation_of_duties_538ee218_v1,
  title = {separation of duties},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/separation-of-duties/rdr_538ee21821544648b488a3d71c7c361c/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake