Byzantine fault tolerance (BFT) is an algorithm designed to allow distributed systems to tolerate faulty or malicious nodes while maintaining consensus among the remaining nodes. It allows systems or networks to continue functioning by resisting an attack like a 51% attack or a denial-of-service (DoS) attack stemming from some nodes transmitting information about false transactions. ^[1][2][3]

Overview

Cryptocurrency is characterized by its decentralized design, where a distributed network of nodes verifies and records transactions, enabling it to function independently without centralized control. To make this possible, each cryptocurrency requires a method to establish consensus, and this is where Byzantine fault tolerance (BFT) becomes significant. Byzantine fault tolerance is a feature of distributed systems that enables them to function properly even if certain nodes in a network are faulty or malicious. When a node in a distributed system provides incorrect or misleading information to other nodes, a Byzantine failure will occur. Various factors, such as malicious attacks, software defects, or challenges with nodes reaching consensus, can trigger failures in a distributed system. Given that Byzantine failures are nearly unavoidable in any distributed system, Byzantine fault tolerance helps tolerate these failures by ensuring the system can reach consensus even if some nodes provide incorrect information. ^[1][4][5]

History

In blockchain technology, Byzantine fault tolerance originates from the Byzantine general problem pioneered by Leslie Lamport, Marshall Pease, and Robert Shostak. This concept gained significant recognition following the publication of their paper titled "The Byzantine Generals Problem" in 1982. The paper illustrated a scenario involving a group of Byzantine army generals encamped outside an enemy city. Each general commanded their army and needed to communicate and reach a unanimous decision about whether to attack or retreat. However, there was a problem carrying out a joint action due to some compromised generals. This problem is called the Byzantine fault, and a system that successfully addresses this problem is considered Byzantine fault tolerant. Subsequently, the concept of Byzantine fault tolerance was adapted for use in cryptocurrency blockchain networks. In the context of cryptocurrency, these generals are represented by nodes responsible for validating crypto transactions, creating a system with Byzantine fault tolerance, which refers to the ability to continue working even when some nodes go down or intentionally try to deceive it. ^[2][4]

In 2008, Byzantine fault tolerance started gaining widespread attention when Satoshi Nakamoto released the Bitcoin white paper, introducing a revolutionary consensus approach resilient to Byzantine faults that utilizes the proof of work (PoW) protocol. Since Bitcoin's introduction, researchers in the blockchain domain have progressed these concepts, creating various blockchain consensus methods like proof of stake (PoS) that also target achieving Byzantine fault tolerance. ^[3]

How BFT works

Byzantine fault tolerance algorithms work by dividing network nodes into groups and compelling them to communicate through message exchanges. This communication allows nodes to authenticate information shared by other nodes, guaranteeing unanimous consensus among all nodes concerning the system's present condition. ^[1]

Blockchains employ Byzantine fault tolerance mechanisms through consensus algorithms to establish a collective agreement regarding the current state of the distributed ledger. These consensus algorithms regulate who has the authority to add a block to the blockchain. In certain blockchains, this involves solving a complex mathematical problem (PoW), while in others, it necessitates staking a certain amount of the blockchain's native currency to facilitate the addition of a new block (PoS). ^[6]

Furthermore, every new block to be included in the blockchain is disseminated to all nodes within the network. Each node subsequently verifies these blocks and their transactions before incorporating them into their local version of the blockchain. Blockchain technology commonly employs various BFT algorithms, such as practical Byzantine fault tolerance (pBFT), federated Byzantine agreement (FBA), and delegated Byzantine fault tolerance (dBFT), among others. ^[6][7]

In the PoW consensus algorithm, network miners engage in solving cryptographic puzzles to validate and generate blocks that store transaction data. The miner who successfully solves the puzzle ahead of others gains the privilege of including the transaction in the expanding block and receiving the block reward, but they are required to provide evidence of solving the puzzle to confirm the addition of the block. Furthermore, mining in PoW involves the use of costly computers or mining rigs. This, however, discourages miners from disseminating false information, as it would be rejected by other participants. Additionally, it lowers the risk of malicious actors taking control of the majority of nodes in the system. ^[2]

PoS consensus mechanism, on the other hand, requires the staking of a specific quantity of cryptocurrency tokens to gain permission to validate transactions. Once a network protocol accepts an individual, the transaction can be included in the expanding block for the receipt of a block reward. ^[2]

Types of BFT algorithms

Practical Byzantine fault tolerance (pBFT)

pBFT is a consensus algorithm designed to operate efficiently in asynchronous systems, where there is no defined upper limit on when the response to a request will be received. It functions by ensuring all nodes in the network have a copy of the blockchain and can validate new transactions and blocks before they are added to the blockchain. In pBFT, nodes are categorized into three groups: a leader node, which assumes the responsibility of suggesting new transactions or blocks to the network and also receives requests from the client; a set of replica nodes, which play a crucial role in validating the proposal by engaging in message exchanges with one another; and a group of client nodes, which are responsible for sending transaction requests. ^[1][8][9]

Federated Byzantine fault tolerance (FBA)

FBA is a consensus mechanism developed to attain BFT in a more decentralized and flexible manner compared to traditional BFT algorithms. Before any user requests, nodes must be pre-established and verified within the FBA network. Nodes have the autonomy to select and trust other nodes, leading to the formation of quorums, which represent the minimum number of nodes required for a solution to be deemed correct. Once a quorum is established, the corresponding block is validated and incorporated into the blockchain. The FBA consensus algorithm follows these steps: ^[10][11]

• Node Selection: Each node picks its quorum slices based on its trust in other nodes, and these slices are locally defined, varying from node to node.
• Voting: Nodes cast votes on statements, such as transaction validity or ledger version. A node acknowledges a statement if it believes a quorum supports it.
• Acceptance and Ratification: Nodes disseminate accepted statements to other nodes in the network. When a node observes that a quorum backs a statement, it considers the statement ratified and takes appropriate action.

Delegated Byzantine fault tolerance (dBFT)

dBFT is a consensus algorithm designed to establish agreement among users in the blockchain and cryptocurrency communities, although it can be complex for many newcomers. Its complexity arises from its superior ability to handle untrustworthy participants on the blockchain, surpassing other algorithms in this regard. NEO, often called the "Ethereum of China," pioneered the Delegated Byzantine Fault Tolerance consensus. This blockchain platform has the ambitious goals of creating a "smart economy" by digitizing assets and implementing smart contracts on the blockchain. The dBFT consensus mechanism functions as a state machine, where transitions are determined by a round-robin scheme to define Primary/Backup nodes and network messages. The following are dBFT states: ^[12][13]

• Initial: Represents the initial machine state.
• Primary: Depends on block height and view number.
• Backup: True if the node is not primary, false otherwise.
• Request Sent Or Received: True if a valid signature from the primary node has been received, false otherwise.
• Response Sent: True if block header confirmation has been sent.
• Commit Sent: True if the block signature has been sent.
• Block Sent: True if the block has been sent, false otherwise.
• View Changing: True if the view change mechanism has been triggered, false otherwise.
• More Than F Nodes Committed Or Lost: True if more than f nodes are locked in the committed phase or considered lost.
• Is Recovering: True if a valid recovery payload was received and is currently being processed.

Examples

Zilliqa

Zilliqa is software designed to offer high throughput with the ability to complete thousands of transactions per second. It aims to solve the blockchain scalability issue and speed by using sharding as a Layer 2 scaling solution, which splits its infrastructure into several interconnected blockchains to support more transactions. With the platform allowing for staking and yield farming, many decentralized applications (dApps), including Zilswap, Avely Finance, and LunarCrush, integrate Zilliqa into their protocols. ^{[14][15][16][17]}

The Zilliqa network provides a range of features, including smart contracting, transaction settlement, and token issuance. Its operation relies on two main principles: sharding, a method that divides the network into shards, enabling nodes to process only a portion of transactions, and pBFT, an algorithm ensuring network security and synchronization. With pBFT, consensus is reached when at least two-thirds of nodes agree on the accuracy of a record before it's added to the blockchain. This consensus process applies to specific shards, where all nodes must agree before a microblock is finalized and merged into a transaction block. Additionally, Zilliqa employs a proof-of-work algorithm to assign node identities and create shards, enhancing the security of the platform's transaction records. ^[15][16]

Stellar

Stellar (XLM), also called Stellar Lumens, is a decentralized peer-to-peer (P2P) network aimed at bridging global financial systems and establishing a standardized protocol for payment providers and financial institutions. Its purpose is to facilitate the swift and cost-effective movement of financial assets, connecting individuals, banks, and payment processors. In addition, Stellar enables users to create, send, and trade various cryptocurrencies. ^[20][21]

Stellar uses a network of decentralized servers and a distributed ledger updated every two to five seconds across all nodes while employing a consensus protocol known as the FBA algorithm. This protocol achieves faster transaction processing by utilizing quorum slices, or specific portions of the network, to validate transactions. Each node in the Stellar network selects a set of "trusted" nodes, and once a transaction gains approval from all nodes in this set, it is considered validated. This streamlined process has significantly enhanced Stellar's network speed, enabling it to handle up to 1,000 network operations per second. ^[22]

Neo (NEO)

Neo is a blockchain network designed to automate the management of digital assets through smart contracts, with the aim of establishing a distributed smart economy system using decentralized applications. The platform allows the creation of dApps for various purposes, such as decentralized exchanges (DEXs), prediction markets, and social networks. In addition to these capabilities, Neo provides users with features like a decentralized file storage system, an identity system, and an Oracle system for integrating external information, such as price data. ^[18][19]

The Neo blockchain has two native cryptocurrencies: NEO, which is utilized for voting on protocol changes, and GAS, used to cover computation costs on the network. The platform employs dBFT, a consensus mechanism ensuring blockchain security and synchronization across its distributed network of computers. dBFT functions similarly to delegated proof of stake (DPoS) and employs a real-time voting system to determine which computers running the software can generate the next block on the Neo blockchain. This means that anyone holding NEO can participate in network operations. Each NEO token, also known as a Neo coin, can be staked to represent a vote and more staked NEO equals greater voting power. All NEO owners who stake their tokens vote for the consensus nodes responsible for creating blocks. In return for proposing and adding new blocks to the Neo blockchain, these consensus nodes receive the network's transaction fees, paid in the GAS cryptocurrency. ^[18][19]