FIDEM: A Standard-Compliant Framework for Secure Binding of MUD Profiles to IoT Devices

Source: arXiv:2605.29654 · Published 2026-05-28 · By Alessandro Lotto, Savio Sciancalepore, Alessandro Brighente, Mauro Conti

TL;DR

This paper addresses the critical security vulnerability known as the MUD-Binding problem within the Manufacturer Usage Description (MUD) standard for IoT device network policy enforcement. The issue arises because devices announce URLs referencing MUD profiles without cryptographically binding the profile to the device's identity, enabling malicious devices to advertise valid but incorrect MUD URLs, thereby causing misapplied network restrictions. Prior solutions either require heavyweight PKI infrastructures, break standard compliance, impose excessive manufacturer involvement, or neglect secure profile updates.

FIDEM is a novel, standard-compliant framework that secures the DHCP-based MUD URL issuance protocol by establishing a cryptographic binding between IoT devices and their MUD profiles through Schnorr-based zero-knowledge proof (ZKP) authentication. This approach avoids PKI reliance, minimizes manufacturer overhead, and supports secure profile updates. The authors formalize a realistic threat model capturing supply chain compromise and the use of legitimate devices as cryptographic oracles. Empirical evaluation on constrained devices (ESP32-S3 and ESP32-C6) shows FIDEM incurs minimal overhead (~5ms latency, ~20mJ energy) compared to insecure DHCP baseline, while improving over certificate-based methods by a factor of 20 in speed and 35% in energy.

Overall, FIDEM represents a practical and scalable advance toward secure, efficient, and interoperable MUD deployments in heterogeneous IoT environments, addressing a key vulnerability hindering widespread MUD adoption today.

Key findings

• FIDEM incurs approximately 5ms additional latency and 20mJ extra energy consumption compared to standard insecure DHCP for MUD URL issuance (∼7% overhead).
• Compared to certificate-based X.509 MUD URL issuance, FIDEM achieves around 20 times faster authentication and 35% lower energy consumption on ESP32 devices.
• FIDEM's Zero-Knowledge Proof (ZKP) binding mechanism requires no PKI infrastructure, eliminating dependency on certificate management and manufacturer active involvement during verification.
• The proposed MUD Management Model (MMM) reduces complexity by grouping devices into classes sharing the same MUD profile and secret keys, significantly lowering key management overhead.
• FIDEM formally resists advanced attackers capable of supply chain compromises and exploiting legitimate devices as cryptographic oracles, as analyzed with ProVerif.
• FIDEM supports secure MUD profile updates via a publish-subscribe signature verification mechanism, ensuring policies remain trustworthy over the device lifecycle.
• FIDEM is fully compliant with the MUD standard, avoiding interoperability issues observed in prior non-compliant schemes.
• Existing defenses relying solely on device authentication do not guarantee MUD-binding; FIDEM addresses this gap fundamentally.

Threat model

The adversary can compromise legitimate IoT devices or introduce counterfeit devices that appear legitimate, exploiting the lack of cryptographic binding between MUD URLs and device identities. The attacker can misuse legitimate devices as cryptographic oracles, performing protocol operations without key possession, thereby enabling attacks such as spoofing MUD URLs pointing to unauthorized profiles within the same manufacturer domain. The attacker cannot physically extract secure device secret keys or compromise manufacturer private keys but can interact over the network using legitimate device credentials to mislead the MUD Controller into enforcing incorrect network policies.

Methodology — deep read

The authors begin by defining a detailed system and threat model for MUD-enabled IoT networks. The adversary can compromise legitimate devices or introduce counterfeit devices, exploiting the lack of binding between MUD URLs and device identity to trigger incorrect network policies.

Data/Environment: Experiments were conducted on two common constrained IoT reference platforms, the ESP32-S3 and ESP32-C6, representing typical hardware examples running the DHCP-based MUD URL issuance.

Architecture & Algorithm: FIDEM builds on the existing MUD standard infrastructure and focuses on securing the DHCP extension used for dynamic MUD URL issuance. It introduces a cryptographic MUD Management Model (MMM) that groups devices sharing identical network profiles into classes. Each MUD class is associated with a secret elliptic curve (EC) key Kc provisioned securely on devices and a corresponding public key Xc published in the extended MUD file.

The core novelty is a Schnorr-based Zero-Knowledge Proof (ZKP) authentication protocol embedded into DHCP option 161. During network join, the device sends a DHPC Discovery message including its MUD URL and a ZKP commitment R derived from a random secret. The MUD Controller responds with a challenge C. The device computes and returns a ZKP response proving knowledge of the secret key Kc without revealing it, binding the URL to the device securely. This ZKP flow prevents attackers from spoofing or redirecting to unauthorized MUD profiles.

Training regime is not applicable; instead, implementation details include cryptographic parameter selection (curve, hash function), reliance on secure hardware for key storage, and integration with existing DHCP message sequences.

Evaluation Protocol: The authors evaluate the scheme using the ProVerif formal verification tool to prove security properties under their threat model. Performance evaluation measures computational latency and energy consumption against baseline DHCP extensions and X.509 certificate-based approaches.

Comparison baselines include insecure DHCP MUD issuance and certificate-based issuance using TLS. Metrics focus on latency (ms) and energy (mJ) on real hardware.

Reproducibility: The authors provide open-source prototype code to promote adoption and validation. The dataset is not applicable. The approach leverages publicly available MUD standard files extended with cryptographic fields.

End-to-End Example: On device join, the IoT device fetches the MUD class parameters in its onboard key material, samples random r, sends R=rG with its MUD URL via DHCP Discovery, receives challenge C via DHCP Offer, computes ZKP response Z=r + H(Kc) mod n, sends Z in DHCP Request, enabling the Controller to verify the binding using public information extracted from the MUD profile.

This process allows the Controller to confirm the device owns the secret key associated with that MUD profile without needing PKI or manufacturer involvement, preventing spoofing.

Overall, methodology is a thoughtful cryptographic design integrated with practical IoT deployment realities and standardized protocols, grounded in formal security proofs and hardware-constrained performance evaluation.

Technical innovations

• Introduction of a Schnorr-based Zero-Knowledge Proof (ZKP) protocol embedded within DHCP messages to cryptographically bind MUD URLs to IoT devices without revealing secrets and without PKI.
• Proposal of a novel MUD Management Model (MMM) grouping devices into classes sharing MUD profiles and keys, reducing key and profile management overhead at scale.
• Design of a standard-compliant extension to MUD files to include public cryptographic parameters necessary for ZKP verification, ensuring backward compatibility.
• A secure MUD profile update mechanism based on a publish-subscribe model with periodic signature verification by the Controller, preserving security during dynamic profile changes.

Baselines vs proposed

• Standard insecure DHCP MUD URL issuance: latency ≈ baseline vs FIDEM: +5ms (~7% overhead)
• Standard insecure DHCP MUD URL issuance: energy baseline vs FIDEM: +20mJ
• X.509 certificate-based MUD URL issuance via TLS: latency 20× higher than FIDEM on ESP32 devices
• X.509 certificate-based approach: 35% more energy consumption than FIDEM

Figures from the paper

Figures are reproduced from the source paper for academic discussion. Original copyright: the paper authors. See arXiv:2605.29654.

Fig 1: MUD architecture and file retrieval procedure [10].

Fig 2: System and threat model. D1 is compromised,

Fig 3 (page 3).

Fig 4 (page 3).

Fig 5 (page 4).

Fig 6 (page 4).

Fig 7 (page 4).

Fig 8 (page 4).

Limitations

• Assumes secure hardware enclaves on IoT devices to protect secret keys; devices lacking such protections may weaken guarantees.
• Does not consider adversaries with physical extraction capabilities or side-channel attacks on cryptographic material.
• The use of the same key per MUD class poses risks if one device in the class is compromised; key revocation and rotation mechanisms are not deeply discussed.
• Excludes out-of-band communication channels (e.g., Bluetooth) that may be attack vectors outside network-based MUD enforcement.
• Performance evaluation presented on two ESP32 variants; other hardware classes with lower capabilities may face different cost tradeoffs.
• The MMM approach assumes distinct network profiles map well to device classes, which may not capture all deployment heterogeneity.

Open questions / follow-ons

• How can key revocation and rotation be efficiently managed at the MUD class level, especially after device compromise?
• Can the approach be extended to devices without secure hardware, potentially leveraging remote attestation or hardware-based protections?
• What are the impacts and defenses for out-of-band or cross-protocol MUD URL injection attacks outside DHCP?
• How can the MMM adapt to finer-grained or dynamically changing IoT network profiles beyond coarse device class grouping?

Why it matters for bot defense

For bot-defense and CAPTCHA practitioners, FIDEM emphasizes the critical need to cryptographically bind device identities to policy profiles to prevent spoofing and impersonation attacks at network admission points. While MUD profiles restrict device network behavior, without secure binding, an attacker can manipulate policy enforcement by advertising other devices’ profiles. The cryptographic framework and zero-knowledge proof design principles in FIDEM illustrate how lightweight, standard-compliant protocols can provide strong authenticity guarantees even on constrained devices without full PKI.

Practitioners working on CAPTCHA and bot-defense can draw parallels in ensuring client authenticity and binding to policy compliance in large-scale systems. The paper’s recognition of advanced adversaries exploiting legitimate devices as oracles highlights the necessity for challenge-response interactive authentication mechanisms resistant to relay and impersonation attacks, which resonate with CAPTCHA logic. Moreover, the focus on minimal manufacturer involvement and backward compatibility illustrates the practical constraints to consider when deploying bot-defense systems in heterogeneous, decentralized device ecosystems.

Cite

@article{arxiv2605_29654,
  title={ FIDEM: A Standard-Compliant Framework for Secure Binding of MUD Profiles to IoT Devices },
  author={ Alessandro Lotto and Savio Sciancalepore and Alessandro Brighente and Mauro Conti },
  journal={arXiv preprint arXiv:2605.29654},
  year={ 2026 },
  url={https://arxiv.org/abs/2605.29654}
}

Read the full paper

• arXiv:2605.29654 (abstract)
• arXiv:2605.29654 (PDF)