What is Primus?
What is Primus?
Primus (formerly "PADO") is a decentralized verification and computation network based on advanced cryptography like MPC-TLS and zkFHE. Primus leverages the power of multi-party computation (MPC), fully homomorphic encryption (FHE) and zero-knowledge proof (ZKP) techniques to revolutionize the way that sensitive information is processed and protected. With a focus on seamless integration and innovative solutions, Primus empowers any web data to be validated for its authenticity, and further to be encrypted and processed by Primus computation nodes while preserving computation integrity and data confidentiality.
About MPC-TLS
There are different techinical approaches for implementing MPC-TLS, where Primus focuses two of them, namely, MPC-model and proxy-model.
MPC-model is a specialized multi-party computation protocol performed on transport layer security (TLS) standard. Note TLS, also known as the "s" in "https" to the general public, ensures that data can be securely communicated between a client and a server. MPC-model enables the TLS data to be proved with the help of a third-party, i.e., a cryptographic attestor. The attestor "virtually" joins the TLS protocol and interactively validates the encrypted response data with the client. Eventually, the attestor can be convinced of the fact that the data is indeed from the TLS server if the client behaves honestly in the whole process. Technical details about can be found here.
proxy-model, is another cryptographic protocol designed in the TLS layer. compared to MPC-model. Proxy-model offers an alternative approach where the attestor serves as a proxy between the client and the server. The attestor "blindly" forwards the encrypted TLS data back and forth between the client and the server. By executing a zero-knowledge proof protocol between the client and the attestor, the later ensures the encrypted TLS response data is sent by the server and the encrypted data is a correct response to a specified request. Technical details of proxy-model can be found here.
About zkFHE
Fully homomorphic encryption (FHE) schemes are powerful cryptographic primitives, that allow computations to be performed on encrypted data without first having to decrypt it. The mainstream FHE applications follow the so-called "outsourced computation" paradigm, allowing any computation task to be performed by a third-party worker, with the input of encrypted private data from a user, in case the user is unable to compute with his private data locally. FHE opens up possibilities for any applications like delegated data processing and collaborative computation which require data confidentiality and privacy as a main concern.
However, one drawback of traditional FHE schemes is the lack of verifiability, i.e., there is no measure to ensure the correctness of the outsourced computation. A typical example is that one can not distinguish the result of a privacy-preserving inference of a machine learning model, from a random response by a remote worker on receiving the user's encrypted data input.
Proposed by Primus Labs, zkFHE is a hybrid cryptographic technique that uses zero-knowledge proofs, to guarantee the computation integrity of the evaluation operations on encrypted data, while retaining the performance of the state-of-the-art FHE schemes.