Understanding Filecoin fork project Filestar in one article

PANews reported on October 20 that the Web3 physical infrastructure project FileStar released a white paper on Github today, announcing the launch of the project. In the white paper, FileStar proposed to fork and iterate based on Filecoin and realize seamless mapping of the rights and interests of existing Filecoin miners.

According to the white paper, the FileStar project will cancel the pre-mortgage and carry out a number of technical innovations. For example, it will implement a new P1 packaging algorithm (optional options include SHA512, Poseidon, Pederson and Blake2s) to improve packaging efficiency; introduce recursive zero-knowledge proof technology to achieve off-chain aggregation of messages and improve the TPS of the entire network; and introduce VRF random sampling technology in Window PoST to reduce the number of spot checks for miners.

In addition, unlike Filecoin, which only incentivizes storage proof, FileStar will gradually implement incentives for storage resources, computing resources, and bandwidth resources, and realize a distributed Internet physical infrastructure with distributed storage, verifiable computing, and measurable bandwidth.

It is understood that FileStar plans to release its source code and launch the mainnet on October 30.

The following is the introduction of FileStar Wenquxing white paper:

summary

We believe that the blockchain world has encountered a huge development bottleneck, and can only rely on huge technological innovation to push the blockchain industry out of its current predicament. In 2009, Satoshi Nakamoto released the Bitcoin white paper, opening an era of trustless electronic cash based on the Internet. Limited by the processing power of the Bitcoin network, Bitcoin has deviated from the original vision of electronic cash and has been pushed to the status of electronic gold by the wheel of history. Ethereum has given more flexibility and programmability to transfer transactions based on the idea of ​​Bitcoin, and introduced the concept of smart contract virtual machines. By providing more powerful processing capabilities different from scripting languages, Ethereum has achieved vigorous development, but it still faces huge development constraints.

Whether it is Bitcoin or Ethereum, in essence, they are in a peer-to-peer network to achieve consistency in computing and data. Because of the extreme scarcity of resources on the chain, the Bitcoin network can only process transactions of Bitcoin itself, and the Ethereum network can only process transactions of Ethereum itself and various Ethereum token symbols and limited logic. In a sense, you can regard the existing blockchain network as an extremely closed network. Whether it is Bitcoin or Ethereum, the full-node physical hardware devices they rely on have huge limitations, which inevitably restrict the processing capacity of the Ethereum network. If you regard Ethereum as a micro cloud, then this is a cloud service with global consistency but very limited processing capacity.

Filecoin is the first to break the barrier between the blockchain world and the physical world. Through space-time proof, it links a large number of high-end servers and hardware resources, and also breaks the internal cycle of the blockchain world. FileStar will go a step further on the basis of Filecoin. Unlike Filecoin, which only encourages everyone to build distributed storage infrastructure, FileStar will be committed to building distributed storage, verifiable computing and measurable bandwidth infrastructure at the same time, and will be committed to serving all blockchain projects in the future. For example, in the future, a large number of Ethereum Rollup proofs can be verified on the FileStar network. The FileStar project will further iterate on the basis of Filecoin and build the physical infrastructure of Web3.0 through continuous technological innovation.

We believe that blockchain itself only provides a set of distributed clearing and settlement protocols. If it can only clear and settle various tokens issued by the community with weak credit, then the value of blockchain will be greatly reduced. The FileStar project will be committed to connecting the physical world under the chain. Unlike the oracle that can only link simple price information and provide it on the chain, FileStar will be committed to connecting the real computing, storage, and bandwidth resources on the Internet, and forming a unified resource measurement and resource representation method, migrating these resources to the blockchain world to build the physical infrastructure of the future distributed Internet.

Project Background

On October 15, 2020, the Filecoin mainnet was highly activated at 148888, announcing the official launch of the most anticipated decentralized storage network in the industry. Filecoin proposed a storage proof mechanism, which, through a series of technical innovations such as Proof of Replication (PoRep) and Proof of Spacetime (PoST), realized the measurement and on-chain representation of real-world off-chain storage resources for the first time. Based on this technology, Filecoin attracted a large number of miners with high-end servers to join the network through a set of economic incentive mechanisms, and to a certain extent integrated a large number of high-performance computing, storage and bandwidth resources. The core concept behind the Filecoin protocol provides an important reference for the future development direction of blockchain + Internet infrastructure, and is a milestone innovation in the industry.

However, behind the innovation of the Filecoin protocol, the Filecoin project still has some regrets in its implementation, which may hinder the further development of this new decentralized storage network:

Different from Filecoin’s “decentralized” storage concept, the development and management model of the Filecoin project is more “centralized”. The development team is still modifying the consensus mechanism, key features, and economic model before the mainnet is launched, which has affected the stability of the network to a certain extent.

The distribution and release rules of FIL tokens are not friendly to miners. At the beginning of the mainnet launch, miners can only buy FIL tokens from the market at high prices to continue to maintain mining expenses. Therefore, there is a large-scale "shutdown" of miners as soon as the mainnet is launched. This is obviously not conducive to the long-term incentives for infrastructure providers (i.e. miners); except for miners, other participants in the ecosystem have not received sufficient incentives and are not willing to participate in maintaining the development of the network;

The threshold for participating in Filecoin mining is high. It requires a large number of pre-collateralized tokens and the purchase of mining machines with AMD high-end processors. These have hindered the participation of more small miners and a large number of Intel mining machines.

The architecture and technology selection of the Filecoin network itself result in limited actual on-chain processing capacity (TPS). When the network is congested, it is even unable to process proof of effective storage on the chain, and normal transfers cannot be made. This largely limits the further expansion of the Filecoin network.

The availability of data stored in the Filecoin network is relatively limited. It can only store some "cold data" with extremely low usage frequency. For "hot data" that needs to be accessed at any time, the Filecoin network is basically unavailable;

In addition, the mining software officially launched by Filecoin still has a lot of room for improvement in packaging efficiency. In a sense, this is a waste of real-world storage, computing, and bandwidth resources, and cannot achieve efficient use of resources.

Therefore, Filecoin still has a long way to go before it can truly become a usable decentralized storage infrastructure. On the other hand, storage is only a part of the Internet infrastructure. In the future, the distributed Internet will need a more complete set of incentive protocols to encourage miners to contribute more resources including storage, computing, and bandwidth.

FileStar Project Vision

FileStar hopes to become the infrastructure of web3 (the future Internet), build a distributed storage, computing and bandwidth incentive network based on the IPFS protocol, integrate Internet infrastructure resources around the world, and achieve optimal resource utilization. This is fundamentally different from Filecoin, which only focuses on storage.

The main difference from Filecoin

FileStar has implemented a more reasonable distributed storage incentive mechanism, and will gradually evolve from distributed storage to the incentive layer of the distributed Internet, achieving more refined incentives and optimizing the use of computing, bandwidth and storage resources. The advantages of FileStar are mainly reflected in

Lower mining threshold, cancel pre-mortgage at the initial launch of the network, cancel SHA256 algorithm's dependence on AMD mining machines, and encourage more miners to contribute network infrastructure resources to the entire network;

Higher scalability, by modifying the spot check logic of WindowPoST and introducing recursive storage proof technology, etc., the ability of on-chain message processing is improved in all aspects, and TPS is increased;

Efficient mining software, comprehensively improving the performance of existing Filecoin mining software and improving the overall packaging efficiency of the network;

High data availability, which can be used to store "hot data" and "warm data", and users can quickly read the stored data;

A decentralized governance mechanism with multi-party participation adopts a community-based development and management model, where developers, miners, and other participants in the ecosystem will jointly determine the direction of network development;

Reasonable token distribution, no pre-mining, no fundraising, most of the tokens will be produced by mining, and at the same time, refined long-term incentives will be provided to all participants in the ecosystem to ensure the long-term healthy development of the ecosystem;

FileStar will inherit the effective storage of the Filecoin mainnet and encourage Filecoin miners to jointly maintain the FileStar network.

FileStar's Technology Improvements and Innovations

As the first step towards a distributed Internet incentive network based on the IPFS protocol, FileStar proposes a number of technical improvements and innovations.

Cancellation of pre-mortgage

According to Filecoin's economic model, pre-collateralization refers to a security mechanism in which miners must pledge a certain amount of tokens when encapsulating each sector and generating effective computing power, and the tokens will not be returned to the miners until the end of the sector life cycle. Its original design was to encourage miners to store data for a long time and ensure data availability.

However, the current design of pre-mortgage in the Filecoin network is not reasonable: on the one hand, the amount of pre-mortgage is relatively large. At the current growth rate of network computing power, if miners mortgage, they are likely to be unable to make ends meet and unable to maintain the entire network continuously; on the other hand, most of the sectors currently encapsulated in the network are so-called "junk sectors", and there are not many "order sectors" containing valid data. It is meaningless to ensure the availability of these junk data through pre-mortgage. Pre-mortgage is the main reason for the current shutdown of miners.

In fact, in order to ensure the security of the data, Filecoin has also designed post-mortgage, that is, all mining rewards can only be gradually unlocked if miners continue to save the data. The security and availability of the data have been largely guaranteed.

Therefore, FileStar will completely cancel Filecoin's existing pre-mortgage, retain the post-mortgage of rewards, and make the following improvements to the sector's mortgage rules to further ensure security.

First, FileStar allows junk sectors to have a shorter life cycle. The main significance of junk sectors is to prove that there is corresponding valid storage space available in the network. Currently, the life cycle of any sector in the Filecoin network is at least one year, but in fact, long-term storage of junk data is a waste of resources. The junk sectors that do not store valid data in the FileStar network will support a shorter life cycle to avoid storage resources being occupied by junk data for a long time.

Secondly, the order sector of FIleStar requires a mortgage storage fee. The order sector stores the user's valid data. By mortgaging the storage fees paid by the user, miners will be encouraged to store the order sector first and ensure its data availability.

Canceling the pre-mortgage will allow miners to join the network for mining at any time, and continue to contribute computing power to jointly maintain the security of the network.

New hash algorithm

The SHA256 algorithm used in Filecoin relies heavily on the instruction set optimization of AMD processors, and a large number of Intel mining machines that do not support related instruction set optimization are at a great disadvantage and are basically unable to participate in mining. This choice has greatly dampened the enthusiasm of Intel miners and also brought great limitations to the future development of the Filecoin network.

FileStar aims to become the incentive layer of the future Internet infrastructure, which will inevitably require more diverse hardware to join the network. Therefore, it is necessary to ensure that the mining performance of x86-based mining machines is at the same level while ensuring security, and to encourage diversity. Currently, FileStar is verifying hash algorithms including SHA512, Poseidon, Pederson and Blake2s. FileStar will evaluate the security and actual performance of these hash algorithms on different platforms and select the most suitable algorithm to support Intel mining machines or other high-performance mining machines.

Recursive zero-knowledge proof technology

Proof of Replication (PoRep) is an important part of Filecoin's storage proof. Combined with zero-knowledge proof, PoRep can quantify storage resources and generate corresponding proofs on the chain. In Filecoin's PoRep proof mechanism, miners need to submit two proofs to the network for each sector they encapsulate. The corresponding messages are PreCommitSector and ProveCommitSector. In fact, in the existing Filecoin network, most of the on-chain messages are submitting these two proofs. However, the on-chain message processing capacity (TPS) of the Filecoin network is very limited. When the network is congested, a large number of proof messages will occupy most of the on-chain resources, and ordinary messages will not be packaged. This also leads to the "selfish mining" behavior of large miners, and the proof messages of small miners are basically unable to be put on the chain.

FileStar proposed a recursive zero-knowledge proof (Recursive ZK-SNARK) technology to solve the above-mentioned TPS bottleneck problem and message on-chain problem.

The basic principle of Recursive ZK-SNARK is to perform off-chain proofs of several sectors generated by miners within a certain period of time, form a Merkle tree, and generate an aggregated proof. In the end, only one proof needs to be submitted to the network to complete the proof chain process of multiple sectors at the same time. In this way, the proof messages that each miner needs to submit will be significantly reduced, thereby improving TPS and achieving network expansion. In addition, by adjusting the degree of proof aggregation, the network's message processing capabilities can also be adjusted to meet the needs of different development stages of the FileStar network in the future.

WindowPoST + VRF mechanism

After completing PoRep, miners need to provide proof of spacetime (PoST) to prove that the data is continuously stored. Each sector encapsulated by miners in Filecoin will be spot-checked every day, and miners need to submit WindowPoST proof correctly, otherwise their pledged FIL will be confiscated. For miners with large storage capacity, the number of proofs submitted every day is very large, and with the further development of the network, more and more WindowPoST will need to be submitted in the network, which may eventually cause network congestion and reduce the network's ability to process ordinary messages.

FileStar introduces a random check mechanism into the WindowPoST check mechanism, which greatly reduces the frequency of each miner submitting WindowPoST proofs, without having to submit proofs of computing power multiple times every day. Ordinary random check functions may be predicted, thus affecting the security of the network, because if miners can determine the time when they are checked, there is a possibility of cheating. FileStar uses a verifiable random function (VRF) to further improve the security of random checks.

Efficient mining software

FileStar will also optimize the existing open source mining software to comprehensively improve the mining efficiency of mining machines and maximize the use of mining machine computing resources and storage resources. The optimization is mainly focused on the task scheduling module and the zero-knowledge proof module.

Task scheduling optimization. Under the same hardware and software conditions, different task scheduling strategies will directly affect the packaging efficiency of the mining machine. Filecoin's current mining software has many defects in task scheduling, which greatly affects the growth of effective network storage. FileStar will release mining software with task scheduling optimization to improve the mining efficiency of the mining machine.

Zero-knowledge proof optimization. In Filecoin, both PoRep and PoST have adopted a large number of zero-knowledge proof algorithms, but there is still a lot of room for optimization in the generation process of zero-knowledge proofs. FileStar's mining software will greatly optimize the efficiency of zero-knowledge proof generation and release it to all miners.

Combining the above two optimizations, FileStar's overall network mining efficiency will be significantly improved compared to the existing Filecoin network, which means that with the same hardware investment, FileStar will gradually become the largest distributed storage network.

Filecoin computing power mapping

The Filecoin mainnet has nearly 600 PB of effective computing power, and it is still growing rapidly. Miners participating in Filecoin mining are the first batch of infrastructure contributors to decentralized storage, and will also be the mainstay of future web3 infrastructure. FileStar will map the rights and interests of all effective computing power on Filecoin in the early stage of its launch. If miners participate in FileStar mining, they will have the possibility of receiving rewards that match the Filecoin mainnet computing power. The specific mapping rules will be announced when the FileStar mainnet is launched.

Decentralized governance mechanism

The FileStar development team is responsible for maintaining the FileStar project, but the development and management of the entire project will adopt a community-based model. In the FileStar community, anyone can submit code, and they are required to submit complete test code. All newly submitted code will be merged into the test network after being fully tested, and will be launched on the main network after running stably for a period of time, so as to ensure the security and stability of the main network.

The FileStar project fully respects the opinions of all community participants in the ecosystem. The development and launch of each new feature must be decided by a vote of community members. Developers, miners, and ordinary users in the FileStar ecosystem can participate in the vote and jointly decide the direction of network development.

Token Economic Model

The native token in the FileStar protocol is STAR, which is mainly used to pay for message fees and storage fees. Miners who participate in mining can obtain STAR mining rewards and fee rewards. In order to attract miners to contribute storage, computing, bandwidth and other resources, and at the same time encourage more ecological participants to join, FileStar has designed a more refined token incentive model.

The total amount of STAR tokens is 2,000,000,000 STAR. There is no external fundraising and the team does not pre-mine. The token distribution rules are as follows (the specific distribution details may be fine-tuned when the mainnet is launched, which shall prevail):

70% is generated by mining, and the output is decreasing day by day, and the output is halved every 6 years.

30% is used to incentivize miners who provide storage resources

15% is used to incentivize miners who provide computing resources

15% is used to incentivize miners who provide bandwidth resources

10% is used to incentivize miners who provide other valuable computing, including zero-knowledge proof services, AI, big data and other valuable computing, which will be decided by the community through voting.

30% is used for long-term incentives for other participants in the ecosystem

15% is allocated to the FileStar Foundation, unlocked in 5 years, for long-term maintenance and iterative upgrades of the project

7% allocated to community developers

5% allocated to enterprises and applications in the FileStar ecosystem, unlocked in 5 years

1% allocated to media

1% is allocated to exchanges and wallets that support FileStar

1% is used to handle legal relations and ensure that the project complies with local laws and regulations

Project Release Plan

The first draft of the white paper was released on October 20, 2020

On October 30, 2020, the first version of Filestar code was released, which implemented improvements based on filecoin, eliminated collateral, and fixed several bugs in filecoin.

On February 30, 2021, most of the content in the white paper was implemented, including a new encapsulation Hash algorithm, recursive zero-knowledge proof to achieve off-chain message aggregation to improve the TPS of the entire network, and the introduction of random sampling Window PoST+VRF, etc.

The Verifiable Computing Network will be launched on August 30, 2021 to incentivize some Filestar miners to switch to the Verifiable Computing Network.

The measurable bandwidth network will be launched on December 30, 2021 to encourage some Filestar miners to switch to the measurable bandwidth network

Future Work

The goal of the first phase of FileStar is to realize an incentive network for distributed storage. Its purpose is to attract a group of high-quality servers and provide abundant storage, bandwidth and computing resources. After the first phase of the network is stable, FileStar will provide more abundant incentive mechanisms to encourage the construction of web3 infrastructure. The directions include:

Propose a computing resource and bandwidth resource certification mechanism to further attract different basic resource providers and realize distributed computing network and distributed bandwidth network;

Incentivize computing resources in the network to provide zero-knowledge proof computing services to other nodes;

Provide verifiable computing services to other nodes in the network.

Summarize

Based on the core innovation of Filecoin, FileStar has implemented a better distributed storage network incentive layer: by canceling pre-mortgage, adopting new hash functions and other improvements, the threshold for miners to participate in mining has been comprehensively lowered; combined with recursive zero-knowledge proof, PoST+VRF and other technical innovations, the TPS and message on-chain problems have been solved, greatly improving the scalability of the FileStar network; the optimization of mining software has significantly improved mining efficiency and improved the utilization efficiency of hardware resources in the network.

FileStar adopts fairer token distribution rules and decentralized community governance, which greatly improves the fairness of the protocol, contributes to the long-term development of the project, and ultimately realizes a distributed storage, computing, and bandwidth incentive network based on the IPFS protocol, becoming an important part of the future web3 infrastructure.

Unlike Filecoin, which only incentivizes people to provide distributed storage, FileStar will incentivize people to provide verifiable computing and measurable bandwidth resources based on distributed storage, and ultimately form a huge differentiation with Filecoin. We believe that with the overall development of the blockchain industry and the full development of the Filestar team, a world of distributed Internet is slowly unfolding before our eyes.