ZetaChain 백서

ZetaChain

원문

bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 5쪽 중 2쪽

ulnerable to certain 51% attacks), this kind of bridge is strong and trustless, albeit rather expensive to operate because the chain of the block headers have to be constantly updated in the smart contract. The Rainbow Bridge is also a good example of a trustless bridge, between Ethereum and NEAR. Wormhole [17]: Wormhole is also a cross-chain message delivery service, but it's not trustless. Rather, it depends on a set of validator nodes to attest the validity of the message delivered. Consider the same task of building a one-way bridge on Ethereum from Solana. When a user sends 1 SOL to a certain custody address, one wrapped SOL is issued on Ethereum. The Ethereum smart contract does not verify the transaction on Solana in order to issue the wrapped coin; it trusts that the super majority of the set of Wormhole validators are honest and correct. The security of Wormhole relies on the super-majority of the validators being honest. It appears that Wormhole relies on reputations of validators to build trust. LayerZero [13]: LayerZero is a communication layer for facilitating cross-chain message delivery. It's essentially a weaker form of Relay (see introduction about Relay). The idea is to enable Chain B to verify that a given transaction or event has happened on Chain A. If Chain B supports general smart contracts, a light client of chain A can be implemented in a smart contract so as to verify information about Chain A in a trustless manner. However, even a light client can be expensive to run in a smart contract, both in terms of computation and storage; for example the BTCRelay on Ethereum appears to be discontinued. LayerZero reduces these costs with an ultra-light client on smart contract which does not report and store the whole chain of block headers (or a significant part of it). Rather, LayerZero relies on trusting a block header without a chain of block headers that can trace back to some known trusted block. The key assumption of the security of LayerZero is that the two parties—Relayer, who provides proof of transaction, and Oracle, who provides the block header—are non-colluding. In our terminology and categorization, LayerZero is not “trustless” due to the trust needed for the independence of the two parties. We use a stricter definition of trustless as where the validity (not necessarily liveness, censorship resistance) of messages does not depend on trust in anything other than the two participating blockchains. If the relayer and oracle collude, they can defraud LayerZero by making up an invalid block header (costs about 2 Ether to compute PoW nonce which is the coinbase reward of each block), and make chain B believe that a non-existent transaction has happened on chain A. LayerZero essentially outsources its security to third-party relayer and oracle. Axelar [19]: Similar to LayerZero, Axelar provides a message passing protocol for building cross-chain smart contract apps. Axelar connects to 81 chains including heterogeneous smart contract platforms including Solana (SVM), Osmosis (Cosmos IBC), although it cannot connect to non smart contract chains directly such as Bitcoin and Dogecoin. The cross-chain message passing appears to be decentralized; there are a set of validators or observers that observe certain smart contracts on connected chains and forward its message to another smart contract on possibly another chain, by initiating a smart contract call there with the message included. Axelar also has its own L1 blockchain but it seems the programmability of the blockchain (its WASM smart contract platform) is not needed for cross-chain message passing. IBC [10]: Inter-Blockchain Communication (IBC) protocol is a TCP/IP-like protocol for communication between sovereign blockchains. IBC is an end-to-end, connection oriented, stateful protocol between blockchains. Practically, IBC usually requires fast finality chains such as CometBFT, and the blockchain must support IBC protocol such as Cosmos SDK-built chains. For the blockchains that support IBC, they can establish connections, and through these connections one blockchain can verify proofs against the consensus states of another blockchain. Each blockchain that supports IBC must run a light client that is capable of verifying proofs on the other blockchain in order for them to be connected. The IBC module must also handle production of proofs, and a separate process (relayer) must relay the packet and proof to the counterparty chain. Among the blockchains that support IBC, very strong interoperability can be established, such as coin transfer, atomic swaps, cross-chain decentralized exchanges, and even cross-chain smart contracts. The major drawback of IBC is that it requires adoption–which is a lot to ask of other blockchains, and also might not be possible for legacy blockchains. Intent [20] based cross-chain solutions: e.g., Across Protocol. Intents are a relatively new mechanism for conducting cross-chain transactions with hybrid on-chain and off-chain components. Users interact with contracts that conform to the Intent EIP [20] and “fulfillers”, i.e., third parties compete to fulfill the order for a fee. The cross-chain orders can be bridging some token to another blockchain, or they could be a swap order that results in a different token on a different chain. Note that the Intents themselves do not specify a particular way to complete the cross-chain transaction (bridging or swap), rather it specifies a unified and unambiguous way for contracts and users to signal the “intent” to trade and leave the fulfillment of the trade up to mostly private market makers. Intents are descriptive protocols for a subset of cross-chain transactions. 3.2. Cross-blockchain Asset Transfer Hop [9]: Hop is a protocol to send coins across rollups and their underlying L1 in a trustless manner. Rollups are by default siloed systems and the asset transfer between rollups and L1 can be slow and expensive. For example, optimistic rollups usually take a week to exit into L1; on the other hand, zk-rollups can instantly validate exit but it involves high computation which is expensive on L1. Hop solves the problem of moving coins across rollups by creating bridges and bridge coins, and uses AMM markets to exchange coins rather than sending coins directly. Specifically, Hop creates bridge coins for each rollup, and the bridge coins can be moved around in batches so as to decrease the cost. The bridge coin acts as an intermediary asset in transferring a coin on rollup A to rollup B. Hop uses the existing rollup bridges to do cross-rollup transactions so it does not need a separate off-chain service. Connext [3]: Connext is a trust-minimized solution for cross-chain asset swaps. The idea is somewhat like generalized atomic-swaps, using Hash Time Locked Contracts (HTLC) to ensure transaction atomicity. It uses a network of off-chain routers to create a market and AMM style pricing mechanism. The safety of user funds do not depend on third-parties, only the liveness of the system does. Compared to Hop, Connext uses off-chain services and therefore can connect beyond rollups on a single L1; compared to externally verified solutions, Connext is application specific and not general purpose. For example, it cannot be adapted to send arbitrary messages or cross chain contract calls. THORChain [15], Chainflip [16]: THORChain (along with similarly built competitors like Chainflip) is a decentralized liquidity network that facilitates AMM style native L1 coins on different blockchains, including Bitcoin, Litecoin, Bitcoin Cash, Ethereum. Notably, THORChain is not, strictly speaking, a bridge, as it does not lock & wrap coins and transact on wrapped coins. Rather, THORChain is an application-specific blockchain that maintains the pool, logic, and management of vaults on different chains for swapping. THORChain distributes the signing key using the GG20 TSS scheme and has its own implementation based on Binance's TSS library. ZetaChain is in-part inspired by the design of THORChain, and can be thought of as a simpler and more generalized platform which enables not only swapping, but a generic smart contract platform that allows arbitrary cross-chain applications to be built easily. For example, developers can implement similar functionality to THORChain as a smart contract on ZetaChain. Synapse [14]: According to public information, Synapse is supposed to be an externally verified validator set based system for cross-chain swaps. It issues AMM smart contracts on external chains, and some composite stablecoin as an intermediary asset to cross-chain. To move the intermediary stablecoins across chains it appears to use a burn and mint strategy. Detailed public information about their validator mechanism is not available at the time of writing this paper. CCTP by Circle ( https://developers.circle.com/cctp ): The Cross-Chain Transfer Protocol (CCTP) v1/v2 by circle (the issuer of USDC stablecoin) is a cross-chain asset transfer protocol, with message passing authenticated and facilitated centrally by Circle service provider. It’s similar to LayerZero, Wormwhole, or Axelar with USDC attached to every cross-chain messaging. The difference is that CCTP is not decentralized; only that its interfaces on multi-blockchains are contract calls. 3.3. Cross-blockchain Smart Contract Quant Network [18]: Functionality-wise, the Quant network and its Overledger [18] is the closest to ZetaChain. The Quant network is a centralized service that provides a standardized web-service-based access to the connected public or private blockchains, or regional legacy database ledgers. It supports general programmability triggered by events on those blockchains (transaction to/from a given address, smart contract interaction, events, state changes, etc.), via popular languages and frameworks such as Javascript, Java, Python, etc. ZetaChain aims to achieve similar general programmability, but with an incentivized public blockchain, with far reduced trust assumptions, more transparency, complete verifiability and auditability. ICP/Chain-Key [2]: The Internet Computer Protocol (ICP) has proposals to enable interoperability to the Bitcoin network via its Chain-Key technology, which is similar to the distributed threshold signature scheme. With Chain Key, ICP in principle can custody funds on the Bitcoin network. It's unclear how ICP observes the Bitcoin network, and how their smart contract platform interacts with external blockchains. HyperService [11]: HyperService proposes a cross-chain smart contract platform that is chain agnostic. It consists of two components: a high level language HSL to describe a cross-chain dApp, and an execution layer that ensures financially atomic transactions. 3.4. Blockchain of Blockchains (BoB) The most prominent BoBs are Cosmos and Polkadot. BoBs are usually frameworks that aim at tight interoperable application-specific blockchains. Polkadot, for example, provides a relay chain which handles all consensus, and Parachains which can be different blockchains with different state-transition functions. The Parachains are tightly integrated and can interoperate seamlessly via the relay-chain. The Cosmos ecosystem, on the other hand, does not share consensus, so the interoperability between Cosmos chains is less tight. Every Cosmos chain is sovereign with their own choice of consensus (typically CometBFT-based fast finality). The Cosmos ecosystem relies on the IBC protocol (see section 3.1), and special blockchains called Hubs to facilitate cross-chain asset transfers, and even cross-chain smart contracts. To enjoy interoperability in Cosmos or Polkadot, the blockchains typically need to be built on some common ground. Legacy blockchains, or new blockchains with their own consensus, cannot be part of BoBs. 4. ZetaChain Blockchain Architecture 4.1. Overview At a high level, ZetaChain is a Proof of Stake (PoS) blockchain built on the Cosmos SDK and CometBFT Tendermint PBFT consensus engine. As a result, ZetaChain enjoys fast block time (~4s) and instant finality (no block confirmation needed, no re-organization allowed). The CometBFT Tendermint consensus engine has been demonstrated to scale to ~300 nodes in production, and with future upgrades with BLS threshold signatures the number can potentially increase to 1000+. The throughput of transactions of the CometBFT Tendermint consensus engine that ZetaChain uses can reach 4000+ transactions per second (TPS) under ideal network conditions [10]. Note that the cross-chain TPS cannot reach nearly as high because cross-chain transactions latency/throughput may be limited by external chain latency/throughput, TSS key-sign throughput, and various other factors such as external node RPC speed, etc. The ZetaChain architecture consists of a distributed network of nodes, often referred to as validators. Validators act as decentralized observers that can reach consensus on relevant external state and events, and can also update external chain state via distributed key signing. ZetaChain accomplishes these functions in a decentralized (without a single point of failure, trustless, permissionless), transparent, and efficient way. Contained within each validator is the ZetaCore and ZetaClient. ZetaCore is responsible for producing the blockchain and maintaining the replicated state machine. ZetaClient is responsible for observing events on external chains and signing outbound transactions. ZetaCore and ZetaClient are bundled together and run by node operators. Anyone can become a node operator to participate in validation provided that enough bonds are staked. See Figure 2 for a high level illustration. Validators : ZetaChain uses the CometBFT Tendermint consensus protocol which is a partially synchronous Byzantine Fault Tolerant (BFT) consensus algorithm. Each validator node can vote on block proposals with voting power proportional to the staking coins (ZETA) bonded. Each validator is identified by its consensus public key. Validators need to be online all the time, ready to participate in the constantly growing block production. In exchange for their service, validators will receive block rewards, and potentially other rewards such as gas fees or processing fees, proportional to their bonded staking coins. Observers : Another set of important participants of ZetaChain consensus are the observers who reach consensus on external chain events and states. The observers watch externally connected chains for certain relevant transactions/events/states at particular addresses via their full nodes of external chains. The observers can be further divided into two roles: sequencer and verifier. The sequencer discovers relevant external transactions/events/states and reports to verifiers; the verifiers verify and vote on ZetaChain to reach consensus. The system requires at least one sequencer and multiple verifiers. The sequencer does not need to be trusted, but at least one honest sequencer is needed for liveness. Signers : The ZetaChain collectively holds standard ECDSA/EdDSA keys for authenticated interaction with external chains. The keys are distributed among multiple signers in such a way that only a super majority of them can sign on behalf of the ZetaChain. It’s important to ensure that at no time is any single entity or small fraction of nodes able to sign messages on behalf of ZetaChain on external chains. The ZetaChain system uses bonded stakes and positive/negative incentives to ensure economic safety. In practice, all above roles (except sequencer) are collocated in the same computer node, sharing software and credentials such as validator keys and bonded stakes and the associated rewards/slashing. ZetaChain is planned to transition from Proof-of-Authority at first to a fully delegated Proof-of-Stake (DPoS) model over time, and gradually delegate the governance of the blockchain to ZETA coin holders via on-chain voting. Figure 2. ZetaChain High Level Architecture. 4.2. Observers Observers are tasked with monitoring external chains for relevant transactions. Observers are continually scanning for external chain events responsible for both burning and minting the native coin (ZETA), messages & smart contract calls, as well as other events that dApps register on ZetaChain. Each observer independently observes using its own full node of external chains, and all the observations must reach consensus on the ZetaChain before being considered finalized. Once events are finalized, it automatically triggers an execution of ZetaChain logic, which can be defined as a custom Cosmos SDK module, or ZetaChain native smart contract. There are two modes of observation: Active and Passive mode. Active observation constantly scans the external blockchains for relevant transactions/events/states. Passive mode relies on a sequencer (or a small set thereof) to scan and report transactions/events, together with Merkle proof. The observers verify the proof and reach consensus on the verification on-chain. The active mode has the advantage of being always live and censorship-resistant due to decentralization, but the cost of each node is high because it needs full nodes (of external chains) for the scanning. Passive mode is much less costly, as verification can be done with a light client. Only one or a few sequence
프로젝트로 돌아가기