Skip to main content
Version: v1

Reference Implementation

Implementation specific architecture documentation for the Aqua Data Vault 'Personal Knowledge Container (PKC)'. In this document you will find reasoning as to why certain components were chosen to build this prototype. We have not duplicated any vendor documentation. Please find the linked-list for the documentation of the components we use here:

All software components selected are open source licenses. They each have active developer communities behind them to secure future support and further development.

Wallets / Accounts​

Why do we use Ethereum Wallets / Accounts?

An identity anchor needs to be a public, private key. This allows a public component which we can share (public key) and a private component which proves ownership over that public-key. Public keys are used in Ethereum as accounts to receive and send transactions. Therefore the Ethereum ecosystem has developed standards. These have evolved to a larger eco-system involving wallet-providers which have built software and hardware wallets to securely manage those keys. We leverage one of the most vibrant developer communities and utilize these wallets. In our case they are used mainly for offline-activities (Login, Signing) which do not require connectivity to the Ethereum blockchain.

Web-Application (Chrome-Extension MetaMask)​

Was selected as it is one of the most advanced, widely used, feature-rich web-wallets. Most importantly it's a browser-plugin which allows us to interact with browser applications. MetaMask also has the ability to connect to a hardware wallet allowing for increased security and levels of assurance.

  • Large User-Base 10 Million +
  • Hardware Wallet Support (E.g. Ledger)
    • Large User-base 5 Million +
  • All related source code is open source, lot's of money behind it (future proof)
  • SSI (Self-sovereign identity): Compatible with the DID schema did:ethr:<wallet_address>
  • Fulfills SSI principles:

Witness Network​

We are using the Ethereum Network as an optional Witness Network for cryptographic time-stamping.

Why public Blockchain?​

The witness event serves the purpose of being able to prove the existence of a document without the ability to deny it. This requires censorship resistance and global coverage. The security of the network, it's distribution, and the high economical price to attack the network, are all good reasons why we want to use a public blockchain to publish highly obfuscated hashes (fingerprints). This proves the state of the data in a Data Vault (PKC).

The MetaMask wallet is able to publish the witness event verification hash to the public Ethereum blockchain. This completes our third important layer of integrity verification, which is to prove cryptographic time and existence of a document by writing it's fingerprint onto the ledger. This is not done directly, but through a process which highly obfuscates the hash. This ensures that that there is no plausible attack (even if the attacker owns a powerful quantum computer) to match the fingerprint with any personal data.

Data Vault​

The Data Vault named 'Personal Knowledge Container' is creatd by using a tool for containerization called 'Docker'. This allows us to package our application so it can be deployed to different operating system environments. Docker provides the abstraction / virtualization layer.

Inside Docker we containerized multiple services:

MediaWiki for data governance​

MediaWiki is a data governance tool which has a legacy as a collaboration tool. It powers Wikipedia, the largest encyclopedia in the world. We step into this experience, and the benefits of a feature rich, battle-proven application. It's a powerful tool to both manage and present data.

As MediaWiki has an active and established developer community, it is possible to tap into existing professional service companies. Supporting development and any required changes within the application.

MediaWiki has been modified to use workflows for data processing. This will allow for implementing and testing workflows and business logic between Personal-Knowledge-Containers.

We use all the benefits of MediaWiki to manage and present data in the browser, while we work around one of its main limitations: Security. MediaWiki was never developed to have advanced security and permission structures, as it was built for Wikipedia - a fully public website. This has several implications for us.

Security Practices:

  • We assume that we can't defend against insider attacks on the same MediaWiki instance, therefore every user has their own instance
  • If there is a need for a collaboration space, it is possible to spawn containers for multiple accounts which trust each other to work together on shared data
  • If you use a hosted PKC instance, assume the provider has access to your data as we don't use homomorphic encryption nor do we have encryption implemented
  • We are using typed and tested PHP (for MediaWiki Extension Data Accounting) to improve test coverage to increase stability and security of our application
  • To securely exchange data between PKC's, we implement a separate security software called Guardian. The Guardian protects the PKC and manages all access according to the accounts owners set permissions. The Guardian is under active development. It will receive the highest scrutiny and attention to ensure a minimal attack surface while allowing trust between Guardians so data can be exchanged.
  • The Apache WebServer present in the MediaWiki container has a technological dependency and needs regular updates as Apache was effected by critical security issues like CVE-2021-44228
  • Therefore we do not expose the Apache web-server to the internet by default and do not recommend this to ensure that there is no unnecessary attack surface. If an organization decides to expose the web-surface of a PKC to the internet, they should expect this instance to be compromised and leak all it's stored data. Therefore we re-emphasize: If PKC / MediaWiki is exposed to the web, there can't be any guarantee that data can be kept private. Therefore assume your data can be leaked on a web-facing MediaWiki instance. Organizations hosting PKC's for collaboration should do so with restricted firewall permissions, ensuring that the web-interface is only reachable for the employees authorized to see the content.

MediaWiki Extension: Aqua​

The existing ecosystem and reference implementations around MediaWiki allowed us to quickly prototype our reference implementation for Aqua. The implementation itself allowed us to further develop the protocol. The downsides are currently old PHP code and MediaWiki's legacy. In return we benefit from a well documented project.

Looking forward we want to see other client-implementations in different languages independent of MediaWiki to show the potential of the Aqua Protocol. We hope to re-use part of the PHP code base to apply it to other applications, like Word-Press.

MediaWiki Skin Choice​

Using Tweeki as a Bootstrap 4 compatible Skin for MediaWiki, we were able to establish contact with the chief maintainer of the extension, and were able to collaborate to solve various issues to improve the skin. We are coordinating with other parties to see how we can improve support for the Skin. In the future we are hoping to increase the maintainer base. Prospectively we want to use Tweeki as an abstraction layer from MediaWiki classic to Bootstrap 4, and respectively Bootstrap 5+.

Authentication​

  • For Authentication with Web-Wallets, we use https://github.com/pelith/node-eauth-server which provides us with an open source solution for a signature-response challenge. We then integrate with OAUTH2 and OIDC using existing authentication standards to achieve the login.
  • The Pelith server provides us with the Signature-Challenge and is forwarding the access token via OAUTH2 to MediaWiki. Here we check if the user exists. If this is the case, access is granted.

WebServer-Proxy / HTTPs​

  • We are using NGINX Web-Server as the proxy server. NGINX is a modern and the most popular (#1 since 2021 by share[^1]) web-server on the web.
  • We are using Let's Encrypt to provide secure connections via HTTPS if a PKC is deployed to the web. Let's encrypt serves over 260 million websites according to the vendor website.[^2] provides free TLS Certificates for secure HTTPS connections.
  • The NGINX Web-Proxy allows us to ensure that we can add additional services to the PKC and redirect the services according to our needs.

Software-Development​

  • We use https://github.com for software development. As we started with very little budget, we gratefully utilised free resources to develop our open source software. This allowed us to focus on being productive.
    • Github is owned by Microsoft and is therefore subject to US policies and Microsofts interests.
    • This itself requires us to move away to be independent or at least host a mirror instance outside of Github. The deletion of a repository or losing access could harm our development process significantly.
  • We use GPL-V3 license on the majority of our repositories.
    • We maintain the rights on branding
    • We maintain more restrictive rights on deployment scripts (service provider technologies)

Browser Choice​

We are using the Chrome-Engine as our Web-Browser of choice. It's the most dominant web-browser on the Web, and has become the quasi standard with many different forks creating an ecosystem around it. Microsoft-Edge is actually Chrome. There are non-google chrome forks which deserve attention, funding and support (like https://github.com/Eloston/ungoogled-chromium). This will provide us with a modern and state of the art browser architecture to work with.

As Chrome is the most secure and modern base to build on, we support -only- chrome. Therefore we use a chrome extension to verify the content of an Aqua file. This extension is available for download in the Google Chrome App Store, or can be built from scratch following the instructions in our repository.[^3]

This allows us to provide ease of setup, as users can easily deploy the chrome extension without further effort by adding it from the google app store.

We are using NPM / node.js for fast prototyping. This comes with severe security implications as the npm ecosystem has been prone to various supply-chain attacks over recent months [^4] .

Technology choices on cryptography​

All cryptography used in our development is based on international NIST standards and follows the international understanding for internet security. This is consistent with the perspective of the BSI in Germany, but for some specific details there are no insights provided by the ministry.

Hashing Algorithms (SHA3-512)​

PKC uses standard Crypto (SHA3-512) winner of the National Institute of Standards and Technology see https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.202.pdf . We do not use custom encryption as this comes at the cost of unnecessary risks.

This is an accepted secure and recommended standard according to BSI - Technische Richtlinie – Kryptographische Algorithmen und Schlussellängen Seite 42.

Merkle Tree[^5]​

  • We use merkle trees to be able to witness the state of all pages / files in a single hash. Used to generate a unique fingerprint of a domain (a collection of verified data hash chain's within one domain).
  • There is a lot of future potential in ongoing research to improve privacy and utility of merkle trees in the Aqua Protocol

Public-Private-Key Signatures (secp256k1)​

Used in combination with Ethereum Accounts. It is used for signing and authentication challenges.

  • For authentication with Metamask Ethereum wallets we currently use https://github.com/pelith/node-eauth-server. We are signing a unique input value with ethereum.js using the personal sign method. Once the challenge is completed successfully we receive an OAUTH token from the Eauth identity provider of the pelith Eauth server.

  1. NGINX: Most Popular Web Server

  2. Let's Encrypt

  3. Aqua Verify Page Chrome Extension - GitHub

  4. NPM Supply Chain Attack

  5. Merkle Tree - Wikipedia