A Distributed Public Key Infrastructure Based on Threshold Cryptography for the HiiMap Next Generation Internet Architecture

Oliver Hanka, Michael Eichhorn, Martin Pfannenstein, Jörg Eberspächer, Eckehard SteinbachFuture 2011 · DOI

TL;DR

This paper proposes a novel security architecture for the HiiMap Next Generation Internet (NGI) by integrating a Public Key Infrastructure (PKI) directly into its mapping system. Leveraging Threshold Cryptography, it distributes public keys across multiple administrative domains, enhancing trustworthiness and resilience compared to traditional PKI models.

What Problem Does This Paper Address?

The current Internet architecture, designed without security as a primary concern, faces limitations as it evolves into a critical global infrastructure. While various security add-ons exist, there's a need for a more integrated and robust security foundation, especially at the network layer, to ensure trust and consistent security for all participants. Specifically, existing Public Key Infrastructures (PKIs) often rely on centralized Certificate Authorities (CAs), which can be single points of failure and trust issues. The HiiMap architecture, a clean-slate approach for NGI based on a locator/identifier-split, aims to address these challenges by providing a scalable, reliable, and mobile-friendly framework. However, securing this mapping infrastructure itself is crucial. The paper tackles the problem of how to securely manage and distribute public keys within the HiiMap architecture to ensure a trustworthy mapping service.

Methods

The core of the proposed solution lies in integrating a Public Key Infrastructure (PKI) directly into the HiiMap's mapping system. Instead of a separate PKI, public keys are stored as part of each mapping entry. To address the trust issues associated with centralized PKIs, the paper leverages Threshold Cryptography, specifically a (k, n)-threshold scheme based on Shamir's Secret Sharing.

Here's how it works:

  • Key Splitting: A node's public key is divided into 'n' pieces.
  • Distribution: These 'n' pieces are distributed across multiple independent administrative domains (regions) within the HiiMap architecture.
  • Reconstruction: To reconstruct the public key, only 'k' out of the 'n' pieces are required.
  • Client-Driven Querying: When a client needs a public key, it queries multiple regions for the key pieces. The client can choose which regions to query, allowing it to omit untrustworthy or geographically distant ones.
  • Storage Location Determination: The paper proposes a method to determine where key pieces are stored. The node's identifier is hashed, and the resulting hash is split into 8-bit segments, each representing a key storage address space (KSA). A mapping directive, obtained from a global authority, translates these KSA values into specific regions responsible for storing the key pieces. This ensures a distributed and fair load distribution.
  • Malicious Region Detection: The system can detect malicious regions. If a region provides a corrupted or manipulated key piece, the reconstruction of the public key will fail or result in an invalid format, allowing the client to identify and potentially report the misbehaving region.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • PKI Integration for Efficiency: Integrating PKI directly into the mapping system, rather than maintaining it as a separate service, significantly reduces resource overhead and simplifies maintenance.
  • Threshold Cryptography for Trust and Resilience: Moving away from single points of trust (like CAs) by using threshold cryptography for key distribution is a critical step towards a more robust and fair NGI. Distributing key pieces means no single entity holds the complete key, making it harder to compromise.
  • Client-Side Control over Trust: Empowering clients to choose which regions to query for key pieces gives them direct control over their trust in the system, allowing them to avoid potentially compromised or untrustworthy entities.
  • Decentralized Key Storage and Load Balancing: The proposed method for determining key storage locations ensures that the burden is distributed across regions, preventing any single region from becoming a bottleneck or a target.
  • Proactive Security for NGI: The approach aims to build security into the network layer from the ground up, aligning with the goals of Next Generation Internet architectures.

Key Results

The paper presents a qualitative evaluation of the proposed security approach against IETF AAA (Authentication, Authorization, and Accounting) recommendations. The system is evaluated across four levels: Total coverage (T), Partial coverage (P), Failed coverage (F), and Out of scope (OS).

  • General Requirements: The system achieves "T" for Scalability, Fail-over, Mutual Authentication, Data Object Integrity, Reliable AAA Transport Mechanism, Support Proxy and Routing Brokers, Shared Secret not Required, and Ability to Carry Service Specific Attributes. It has "P" for Transmission Level Security and Data Object Confidentiality. "OS" is noted for Certificate Transport and Auditability.
  • Authentication Requirements: The system fully supports Network Access Identifier (NAI) and Authorization without Authentication ("T"). Reauthentication on Demand is "OS". Support for Challenge-Handshake Authentication Protocol (CHAP), Extensible Authentication Protocol (EAP), and Password Authentication Protocol (PAP)/Clear-text passwords are "T*" (if required).
  • Authorization Requirements: Support for static/dynamic IP assignment and RADIUS Gateway Capability are "T*". Reject Capability is "T". Reauthorization on Demand and Preclude Layer 2 Tunneling are "OS". State Reconciliation is "T". Unsolicited Disconnect is "F".
  • Accounting Requirements: Real-Time Accounting, Mandatory Compact Encoding, Accounting Record Extensibility, Batch Accounting, and Dynamic Accounting are "OS". Guaranteed Delivery is "T*" and Accounting Timestamps are "T".

Overall, the system fulfills 92% of the evaluated requirements, partially covers 4%, and misses 4%, when neglecting out-of-scope aspects.

Limitations and Future Work

  • Malicious Region Cooperation: While the paper acknowledges that a cooperation of several malicious regions could pose a threat, it argues that the motivation for such an attack is low and that a sufficiently high 'k' value in the (k, n)-threshold scheme can mitigate this. However, this introduces a trade-off between complexity, delay, and security.
  • Unsolicited Disconnects: The system is noted as "F" for Unsolicited Disconnects, meaning it cannot initiate a disconnect from the mapping system's side. The paper suggests that higher layers might implement this.
  • Scalability of Key Storage Address Space: The paper assumes less than 256 regions for simplicity in determining key storage locations. While it states that any number up to 256 will work, scaling beyond this might require further consideration.
  • Dynamic Accounting: Dynamic accounting is considered out of scope, as the system does not integrate sessioning. This functionality would need to be handled by higher-layer services.

Future work could explore more detailed quantitative evaluations, investigate the trade-offs of different (k, n) values for threshold cryptography, and address the implementation of unsolicited disconnects at higher layers. The paper also suggests that the global authority can periodically conduct random probes to monitor regions, which could be further developed.

Summary updated

Abstract

In this article, a security extension for the HiiMap Next Generation Internet Architecture is presented. We regard a public key infrastructure which is integrated into the mapping infrastructure of the locator/identifier-split addressing scheme. The security approach is based on Threshold Cryptography which enables a sharing of keys among the mapping servers. Hence, a more trustworthy and fair approach for a Next Generation Internet Architecture as compared to the state of the art approach is fostered. Additionally, we give an evaluation based on IETF AAA recommendations for security-related systems.
BibTeX
@inproceedings{hanka2011distributed,
  author = {Oliver Hanka and Michael Eichhorn and Martin Pfannenstein and Jörg Eberspächer and Eckehard Steinbach},
  title = {{A Distributed Public Key Infrastructure Based on Threshold Cryptography for the HiiMap Next Generation Internet Architecture}},
  booktitle = {Future},
  year = {2011},
  doi = {10.3390/fi3010014},
}
Figure