IOSG ｜ Why does FHE have better application prospects in web3

Privacy is the basic right of human and organization.For individuals, it helps people express themselves freely without having to disclose any information that I don’t want to share to third parties.For most organizations today, data is regarded as the main product, and data privacy is essential for protecting this product.The commercialization of password punk sports and data has accelerated the research and development of the original language of cryptography.

Code science is a quite extensive field. When we look at cryptography in the background of calculation, we have seen many different solutions, such as zero -knowledge certification, same -state encryption, Secret sharing, etc. These schemes have been since its birth in the 1960s.Continuous improvement.These schemes are crucial to unlock the private calculation method (the reason why the data is the main product is because people can find insight from it).To this day, the field of PRIVATE computing has developed significantly in multi -party computing and zero -knowledge proof, but the input data itself has always had privacy problems.

When the most important commodity is made public, it is very difficult to outsourcing the calculation of this data out of the calculation of this data without a legal agreement.Today, everyone relies on the compliance standards of data privacy, such as HIPAA for health data and GDPR specifically for data privacy for data in Europe.

In the field of blockchain, we better believe in the integrity of technology, not the integrity of regulators.As believers who permit and maximize ownership, if we believe that users have the future of data, we need to calculate these data without trust.Prior to the work of Craig Gentry in 2009, the concept of calculation and calculation of encrypted data has not made a breakthrough.This is the first time that someone has been able to perform calculation (addition and multiplication) on the ciphertext (ie, encrypted data).

>

AllThe working principle of the same state encryption (FHE)

So, what is this “magic mathematics” that allows computers to perform calculations without understanding the input?

Full -state encryption (FHE) is a type of encryption scheme that allows calculating calculation on the encryption data (ciphertext) instead of decryption data, which opens a series of use cases for privacy and data protection.

During the FHE process, when the data is encrypted, it will add additional data called noise to the original data.This is the process of encrypted data.

Each time it executes the same state calculation (plus or multiplication), additional noise is added.If the calculation is too complicated and the noise is added at each time, the ultimate decryption will become very difficult (which is very heavy in calculation).This process is more suitable for adding, because the noise is linearly increased, and the noise is exponentially increased for multiplication.Therefore, if there is a complex polynomial multiplication, decrypting output will be very difficult.

If noise is the main problem and its growth makes FHE difficult to use, it must be controlled.This gave birth to a new process called “Bootstraping”.Guidance is a process of encrypted encryption data with new keys and decrypting in encryption.This is very important because it significantly reduces the calculation overhead and the final output decryption overhead.Although Bootstrapping reduces the final decryption overhead, there will be a lot of operating expenses in the process.This may be expensive and time -consuming.

>

At present, the main FHE solutions are: BFV, BGV, CKKS, FHEW, TFHE.Except for TFHE, the abbreviations of these schemes are the names of the author of their thesis.

These solutions can be regarded as different languages ​​in the same country, and each language is targeted at different optimizations.The ideal state is to unify the country, that is, all these languages ​​can be understood by the same machine.Many FHE working groups are working hard to achieve the combined nature of these different solutions.Libraries such as SEAL (combined with BFV and CKKS schemes) and Helib (BGV + approximate number CKKS) help implement a FHE solution or different calculated solution combinations.For example, Zama’s Concrete library is a Rust compiler for TFHE.

>

2.FHE scheme comparison

Below is Charles Gut, Dimitris Morris and Nexarios George Chushel in its papers “SOK: New Opinions on Full -Status Cardrait” SOK: New InsightsInto Fully Homomorphic Encryption Libraries via Standardized Benchmark (2022) performance comparisons comparison with different libraries.

>

Web3 use case

When we use the blockchain and applications today, all the data are public, and everyone can see it.This is beneficial to most of the cases, but it completely limits many use cases that require default privacy or data confidentiality (such as machine learning models, medical databases, genomics, private finance, games that are not manipulated).FHE supports blockchain or virtual machines that are essentially allowed to encrypt the entire chain from the starting point to ensure privacy, and at the same time allow arbitrarily calculation on encryption data.All storage or processing data on the blockchain network is essentially safe.ZAMA has a FHEVM solution that allows EVM computing in a completely same -state environment.At the execution level, the L1/L2 project built by this library ensures privacy.Although the privacy chain has always been cool technology, the use rate and token performance have not improved significantly.

In terms of outsourcing universal computing, FHE itself is not to replace ZK and MPCs.They can supplement each other to create an unbelievable private computing giant.For example, Sunscreen is building a “privacy engine”, which basically allows any blockchain application to bring the calculation to their FHE computing environment and can feedback the calculation results.The generated calculation can be verified by ZK proof.Octra is doing similar things, but using a different type of encryption solution, called HFHE.

ZK proves that he is good at proved something when the data is not revealed, but the proofer can still access these data at a certain point in time.ZK proves that it cannot be used for the calculation of private data; they can only verify whether certain calculations are completed correctly.

MPC disperses the calculation of encrypted data to multiple machines, performs calculation in parallel, and then stitches the final calculation results together.As long as most machines that are calculated are honest, the original data cannot be retrieved, but this is still a trust assumption.Because the MPC needs continuous communication between the parties (data needs to be continuously split, calculated, and re -connected), it becomes difficult to expand through hardware.

In FHE, all calculations are performed on the encryption data without decrypting data, and this can be completed on a single server.FHE’s performance can be expanded by accelerating better hardware, more computing resources and hardware.

At present, FHE’s best cases in the blockchain field are more commonly calculated in outsourcing, rather than building a built -in FHE L1/L2.The following are some interesting use cases that FHE can unlock:

• The first generation (encrypted native): DID, casinos, betting, voting, games, private defii, private tokens, Dark Pools, 2FA, backup, password.

• The second generation (modular): “Chainlink for Privacy”, outsourcing private computing, end -to -end encryption, encrypted data availability, and verified security data storage between the blockchain and contract.

• The third generation (enterprise level): complex consumer applications, encrypted and decentralized LLM, artificial intelligence, wearable equipment, communication, military, medical, medical, privacy protection payment solutions, Private P2P payment.

Currently FHE -based industry project

The development of completely homogeneous encryption (FHE) has stimulated a number of innovative blockchain projects that use this technology to enhance data privacy and security.This section explores the technical details and unique methods of noticeable projects such as Inco, Fhenix, and ZAMA.

INCO

>

INCO is creating the pioneer of FHE and blockchain integration, creating a platform, making data computing safe and private.INCO uses Lattice-based encryption technology to implement its FHE solution to ensure that the operation of ciphertext (encrypted data) can be performed without exposing the underlying text.The platform supports smart contracts for privacy protection and allows processing encrypted data directly on the blockchain.

• Lattice-Based’s FHE: Inco uses a grid-based encryption for its FHE implementation. It is known for its post-quantum safety characteristics to ensure that it is elastic on possible quantum attacks in the future.

• Privacy Protection Smart Contracts: Inco’s smart contracts can execute arbitrary functions for encryption input to ensure that nodes of contracts and execution contracts cannot access explicit data.

• Noise management and Bootstrapping: In order to deal with the problem of noise growth during the same -state operation process, Inco has implemented an efficient Bootstrapping technology, refreshing ciphertexts, maintaining decryption, and complicated calculations.

Fhenix

>

Fhenix focuses on providing strong infrastructure for privacy protection applications, and uses FHE to provide end -to -end encryption solutions to protect user data.Fhenix’s platform aims to support the widespread application of transmission from security messages to privacy financial transactions to ensure data privacy in all computing processes.

• End -to -end encryption: Fhenix ensures that the data keeps the encryption state from the entire process of processing and storage.This is achieved by combining FHE and security multi -party calculation (SMPC) technology.

• High -efficiency key management: Fhenix integrates the advanced key management system to facilitate the division and rotation of security keys. This is the key to maintaining long -term security in the FHE environment.

• Scalability: The platform uses optimized homogenic operations and parallel treatment to efficiently handle large -scale calculations, solving one of the main challenges of FHE.

• Coordinator: Fhenix also took the lead in developing a specialized processor to accelerate FHE computing.These collaborators specifically handle the dense mathematical operations required for FHE, which significantly improves the performance and scalability of privacy protection applications.

Zama

ZAMA is the leader of the FHE field and is known for its FHEVM solution it developed.This scheme allows the Ethereum EVM computing in a completely identical environment to ensure the privacy of any L1/L2 project built by the library at the execution level.

• FHEVM solution: Zama’s FHEVM solution integrates FHE and Ethereum virtual machines to implement encrypted smart contracts.This allows confidential transactions and calculations in the Ethereum ecosystem.

• Concrete library: Zama’s Concrete library is a Rust compiler for TFHE (a variant of FHE).The library provides high performance implementation of the same -state encryption solution, making the encryption computing more efficient.

• Interoperability: Zama is committed to creating solutions that can seamlessly collaborate with existing blockchain infrastructure.This includes supporting various encrypted originals and protocols to ensure extensive compatibility and easy integration.

>

3..FHE’s key characters in Crypto and AI Infra and applications

Today, the intercourse between cryptography and artificial intelligence is in full swing.Although this intersection is not discussed in depth, it is worth noting that the innovation of new models and data sets will be promoted by the open source cooperation of multiple participants.In addition to calculation, the most important thing is the data. These data are the most important parts of this cooperative channel.The usefulness of AI applications and models eventually depends on the data trained, whether it is basic models, fine -tuning models, or AI intelligent proxy.Maintaining the security and privacy of these data can open a huge design space for open source cooperation, and at the same time allow data owners to continue to profit from training models or final applications.If these data are essentially public, it will be difficult to monetize (because anyone can access valuable data sets), so these data are likely to be strictly protected.

In this case, FHE can play a key role.In ideal, it can train the model without disclosing the underlying data set, which may unlock the monetization of the data set and greatly promote open source cooperation between the data set owner.

>

Source:Bagel network

How to enhance privacy protection machine learning (PPML)

• Data privacy: By using FHE, sensitive data, such as medical records, financial information, or personal identifiers, can be encrypted before entering the ML model.This ensures that the data remains confidential even if the computing environment is damaged.

• Security model training: Training ML models usually require a lot of data.Using FHE, these data can be encrypted to train the model without exposing the original data, which is essential for industries that process high -level sensitive information and are restrained by strict data privacy regulations.

• Confidential reasoning: In addition to training, FHE can also be used for encryption reasoning.This means that once the model training is completed, it can be predicted in the encryption input to ensure that user data maintains privacy throughout the reasoning process.

FHE’s PPML application field:

• Healthcare: Training ML models on the premise of protecting privacy can lead to more personalized and effective treatment without exposing sensitive patient information.

• Financial: Financial institutions can use FHE to analyze encrypted transaction data to achieve fraud testing and risk assessment, while maintaining customer privacy.

• The Internet of Things and smart devices: Equipment can collect and process encrypted data to ensure that sensitive information such as location data or use mode keeps confidential.

FHE’s question:

As mentioned earlier, there is no “unity” between the FHE scheme.The scheme cannot be combined, and it needs to be a different FHE solution for different types of computing combinations.The process of different schemes in the same calculation experiment is also quite cumbersome.The development of the Chimra framework is allowed to switch between different FHE schemes such as TFHE, BFV, and Heaan, but it is currently not available.This leads to the next problem, that is, the lack of benchmark testing.The benchmark test is very important for developers to use this technology.This will help save time for many developers.Considering the calculation overhead (encryption, decryption, Bootstrapping, key generation, etc.), many existing common hardware is not very applicable.It is necessary to accelerate some form of hardware, or may need to create a specific chip (FPGA and/or ASIC) to achieve FHE more mainstream applications.These models can be compared with the problems in the ZK (zero -knowledge) industry.As long as a group of smart mathematicians, application scientists and engineers are interested in this field, we will continue to optimize these two fields: FHE is used for privacy and ZK for verification.

>

4. 4..What will the future of the FHE drive look like

Will there be a FHE scheme rule owned?The industry is still conducting such discussions.Although the ideal state is a unified solution, the diversified needs of different applications may always need to be optimized for specific tasks.Is the interoperability between schemes the best solution?Mutuality may indeed be a practical method that allows flexibly handling diversified computing needs, while using the advantages of various schemes.

When can FHE be available?The availability is closely related to the progress of calculation overhead, the standard testing standard for improvement, and the progress of the development of special hardware.With the progress of these areas, FHE will become more accessible and practical.

In summary, FHE provides powerful tools for data privacy protection and security computing.Although there are still challenges in terms of interoperability, computing expenses and hardware support, FHE’s potential in blockchain, privacy protection machine learning, and wider Web3 applications cannot be ignored.With the continuous development and innovation of technology, FHE is expected to play a key role in future privacy protection and security computing fields.