Back to Federated Learning

Federated learning — threat modelling

Detailed STRIDE + LINDDUN recipes for FL security risk analysis.

data security privacy governance

Threat modelling and risk assessment

Apply the LINDDUN privacy‑engineering framework to elicit privacy threats, then map the resulting findings to Microsoft’s STRIDE model to ensure security coverage.

Practical recipe (LINDDUN GO ＋ STRIDE)

  1. Map data flows – draw a DFD that captures FL clients, the coordinator, and all artefacts exchanged (model updates, metrics, credentials).
  2. Identify threats – for each DFD element inspect:
    • The seven LINDDUN privacy categories: Linkability, Identifiability, Non‑repudiation, Detectability, Disclosure of information, Unawareness, Non‑compliance; and
    • The six STRIDE security categories: Spoofing, Tampering, Repudiation, Information disclosure, Denial of Service, Elevation of Privilege.
  3. Assess impact – score every threat (Low / Medium / High) for both likelihood and impact.
  4. Select mitigations – map each threat to technical or organisational controls and record the residual risk.
Threat category LINDDUN focus STRIDE focus Example mitigation in FL
Linking Re‑identification via update correlation — Add Gaussian DP noise to model updates
Identifying Direct identifiers in logs — Remove user IDs before logging
Non‑repudiation Verifiable update provenance — Retain only aggregate logs
Detecting Participation inference — Add dummy (zero) client updates
Unawareness Missing consent — Deploy dynamic consent portal
Non‑compliance GDPR Article 35 DPIA — Embed privacy‑by‑design checkpoints
Spoofing — Identity fraud Mutual TLS ＋ OIDC
Tampering — Model poisoning Byzantine‑robust aggregation
Information disclosure — Data leakage Secure aggregation (e.g. SecAgg+, LightSecAgg)
Denial of service — Resource exhaustion Rate‑limit slow / malicious clients
Elevation of privilege — Unauthorised API use RBAC and short‑lived tokens

See the LINDDUN tutorial and the companion worksheet pack for step‑by‑step templates [1].

Resources:

Bibliography

  1. Wuyts, Kim, Joosen, Wouter (2015). LINDDUN privacy threat modeling: a tutorial. CW Reports. Available at: https://downloads.linddun.org/tutorials/pro/v0/tutorial.pdf