Private key encryption and recovery in blockchain

Mehmet Aydar, Salih Cemil Cetin, Serkan Ayvaz, Betül AygünarXiv 2019 · arXiv · OpenReview

TL;DR

This paper tackles the critical challenge of private key management in blockchain by proposing a novel system for secure encryption and recovery. It leverages biometric data (fingerprints) for encryption and a distributed secret sharing scheme for recovery, aiming to enhance both security and usability for blockchain asset owners.

What Problem Does This Paper Address?

Blockchain technology offers decentralized control over assets, but this power comes with the significant responsibility of managing private keys. The paper identifies two major pain points: the lack of secure and user-friendly methods for storing private keys, and the absence of efficient recovery mechanisms when keys are lost. Traditional methods like memorization are difficult, cold storage is cumbersome, and storing keys digitally in plain text is highly insecure. Even encrypted digital wallets require users to remember complex passwords, and losing them means losing access to assets. This paper aims to bridge this gap by providing a robust solution that combines biometric security with a decentralized recovery approach.

Methods

The proposed solution has two main components:

  • Biometric Encryption of Private Keys: The authors introduce a method to encrypt a user's private key using their fingerprint. This involves:
  • Symmetric Encryption: A symmetric encryption algorithm (like DES) is used to encrypt the private key.
  • Biometric Key Generation: The symmetric key itself is derived from the user's fingerprint data. This process involves preprocessing the fingerprint image (enhancement, binarization, thinning), extracting minutiae points, and then transforming these points using a Cartesian transformation. This transformed data is then used to generate a hash, which serves as the symmetric key.
  • Reed-Solomon Error Correction: To ensure the generated hash is consistent even with slight variations in fingerprint scans, Reed-Solomon error correction is applied to the minutiae point hashes within cartesian blocks. This allows for the recovery of the overall hash (and thus the symmetric key) even if some parts of the fingerprint data are imperfect.
  • Distributed Private Key Recovery: For recovery, the paper proposes a system based on Shamir's Secret Sharing (SSS) scheme, enhanced with biometrics.
  • Secret Splitting: The encrypted private key (or a passphrase used to derive it) is split into multiple pieces.
  • Distribution to Stewards: These pieces are distributed to trusted nodes, termed "stewards," in the network.
  • Biometric Authentication for Recovery: To recover the private key, a user needs to provide their biometric data (fingerprint) to regenerate the symmetric key. This regenerated symmetric key is then used to decrypt the recovered pieces and reconstruct the original private key. The SSS scheme allows for recovery even if some stewards' data is lost, as long as a minimum threshold of pieces (k out of n) is collected.

Datasets

Not discussed in the paper. The methodology relies on processing fingerprint images, implying the use of standard fingerprint datasets for development and testing, but specific datasets are not mentioned.

Key Insights and Ideas

  • Biometrics as a Dynamic Key Generator: Instead of storing a static password or key, the system dynamically generates a symmetric key from a user's biometric trait. This offers a more secure and user-friendly alternative to traditional password management.
  • Revocable Biometric Templates: The use of Cartesian transformation and Reed-Solomon coding on fingerprint minutiae points creates "cancelable" biometric templates. This means that if a template is compromised, it can be revoked and a new one generated, mitigating the permanent nature of biometric data compromise.
  • Decentralized Recovery with Enhanced Security: Shamir's Secret Sharing, combined with biometric authentication, provides a robust and flexible key recovery mechanism. Distributing key shares across multiple stewards prevents a single point of failure or compromise, while the biometric requirement ensures that even if shares are stolen, the private key cannot be decrypted without the user's direct involvement.
  • Addressing the "Two Bottlenecks": The paper directly tackles the two major issues in blockchain key management: secure storage (via biometric encryption) and efficient recovery (via distributed secret sharing).

Key Results

The paper outlines a methodology and proposes a system. While specific quantitative results (e.g., accuracy of fingerprint matching, success rate of recovery, performance metrics) are not detailed in this excerpt, the core contribution lies in the conceptual framework and the integration of existing cryptographic and biometric techniques into a novel solution for blockchain private key management. The authors claim their approach provides secure storage and efficient recovery.

Limitations and Future Work

  • Limitations:
    • Availability of Stewards: The initial description of the distributed recovery mechanism implies that all stewards might need to be available to recover the passphrase. However, the paper later clarifies that Shamir's Secret Sharing adds flexibility, suggesting this initial limitation is addressed.
  • Biometric Accuracy: The effectiveness of the system relies heavily on the accuracy and reliability of fingerprint recognition. False positives or negatives in fingerprint matching could lead to failed encryption or recovery.
  • Privacy Concerns: While the system aims to enhance security, the handling and storage of biometric data always raise privacy concerns. The paper mentions that biometric data should not be directly associated with biometric properties to ensure revocability, but the practical implementation details of this are crucial.
  • Performance: The computational overhead of fingerprint preprocessing, feature extraction, transformation, and Reed-Solomon coding, as well as the SSS scheme, would need to be evaluated for real-world performance.
  • Future Work:
    • Integration with Mobile Applications: The authors express an aim to integrate the proposed solution onto mobile applications.
  • White-Box Cryptography: They also plan to explore the integration of white-box cryptography for further security enhancements.
  • Comprehensive Evaluation: A full implementation and rigorous evaluation of the proposed system, including performance benchmarks and security analyses, would be necessary.

Summary updated

Abstract

The disruptive technology of blockchain can deliver secure solutions without the need for a central authority. In blockchain protocols, assets that belong to a participant are controlled through the private key of an asymmetric key pair that is owned by the participant. Although, this lets blockchain network participants to have sovereignty on their assets, it comes with the responsibility of managing their own keys. Currently, there exists two major bottlenecks in managing keys; $a)$ users don't have an efficient and secure way to store their keys, $b)$ no efficient recovery mechanism exists in case the keys are lost. In this study, we propose secure methods to efficiently store and recover keys. For the first, we introduce an efficient encryption mechanism to securely encrypt and decrypt the private key using the owner's biometric signature. For the later, we introduce an efficient recovery mechanism using biometrics and secret sharing scheme. By applying the proposed key encryption and recovery mechanism, asset owners are able to securely store their keys on their devices and recover the keys in case they are lost.
BibTeX
@misc{aydar2019private,
  author = {Mehmet Aydar and Salih Cemil Cetin and Serkan Ayvaz and Betül Aygün},
  title = {{Private key encryption and recovery in blockchain}},
  year = {2019},
  eprint = {1907.04156},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1907.04156},
}
Figure