Evolution of Blockchain Consensus Algorithms: A Review On The Latest Milestones of Blockchain Consensus Algorithms

Hussein et al.
Cybersecurity (2023) 6:30 Cybersecurity

https://doi.org/10.1186/s42400-023-00163-y
REVIEW Open Access
Evolution of blockchain consensus

algorithms: a review on the latest milestones
of blockchain consensus algorithms
Ziad Hussein1* , May A. Salama1 and Sahar A. El‑Rahman1
Abstract
Blockchain technology has gained widespread adoption in recent years due to its ability to enable secure and trans‑
parent record-keeping and data transfer. A critical aspect of blockchain technology is the use of consensus algorithms,
which allow distributed nodes in the network to agree on the state of the blockchain. In this review paper, we exam‑
ine various consensus algorithms that are used in blockchain systems, including proof-of-work, proof-of-stake, and
hybrid approaches. We go over the trade-offs and factors to think about when choosing a consensus algorithm, such
as energy efficiency, decentralization, and security. We also look at the strengths and weaknesses of each algorithm as
well as their potential impact on the scalability and adoption of blockchain technology.
Keywords Blockchain, Consensus, Proof of work, Proof of stake, Decentralization
Introduction blockchain technology, signaling the emergence of a new

Blockchain is one of the most promising emerging tech- scientific field and an innovative distributed technology.
nologies in the 21st century. It offers significant benefits, The consensus algorithm is a protocol or mechanism
such as decentralization, non-tampering, non-forgery, that is used to achieve agreement among the nodes in
and traceability, making it ideal for storing and secur- a distributed network (Xiao et al.2019). In a blockchain
ing important anti-counterfeiting data (Guo and Yu network, the nodes are computers or devices that store
2022). It also has the potential to solve security issues and maintain a copy of the blockchain, and the consen-
related to data tampering and loss in traditional central- sus algorithm is used to ensure that all the nodes have
ized endorsement agencies, as well as improve the effi- the same view of the blockchain and agree on the order
ciency of transaction processing in various fields, such of transactions. This is important because it allows the
as finance, medical, Internet of Things, property rights network to maintain a single, consistent, and tamper-evi-
protection, privacy protection, etc. Blockchain technol- dent ledger of transactions without the need for central
ogy has been gaining attention for its ability to create authority. The drawbacks of blockchain technology have
great value. It originated from the 2008 paper “Bitcoin: A started to materialize. For example, the ability to manage
Peer-to-Peer Electronic Cash System” written by “Satoshi redundant transactions is incompetent in terms of per-
Nakamoto” (Nakamoto 2008). The creation of the gen- formance and scaling of network (Ammous and Saifed-
esis block in Bitcoin in 2009 marked the official birth of ean 2016). Moreover, sometimes the data throughput is
too small and the capacity of defending the blockchain
from malicious nodes is limited. One of the most sub-
*Correspondence:
Ziad Hussein
stantial problems of blockchain technology is the power
[email protected] consumption; there’s an urge need for the intensive con-
1
Electrical Engineering Department, Faculty of Engineering‑Shoubra, sumption of the computational power (Denisova 2019).
Benha University, Cairo, Egypt
This problem is considered a stumbling stone in the
© The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/.
Hussein et al. Cybersecurity (2023) 6:30 Page 2 of 22
evolution path of this field. Solving this problem is con- respect to different criteria. In section 4, we will dem-
sidered one of the most important research points. Con- onstrate some future improvements. Finally, section 5
sensus algorithms are not a new topic, in fact, it predates concludes our research and focuses on the difference
the era of blockchain itself. One of the leading influences between our paper and others’ researches.
in this field can be tracked back to the late fifties of the
last century, when it had been proposed one of the con- Blockchain overview
sensus algorithms when it had been used as a part of In this section we are going to discuss blockchain archi-
probability function (Xiong et al. 2022). Consensus algo- tecture, how it works and illustrate the core of the block-
rithms are classified under two main categories; one of chain technology.
them considers the existence of malicious nodes and one
does not. The one it does is called nodes and one does Architecture
not. The one that does is called Byzantine fault-tolerant The blockchain is a type of Distributed Ledger Technol-
and the one that does not is called non-Byzantine fault- ogy (DLT) (El loini and Bahl 2018) that allows the secure,
tolerant. One of the most well-known non-Byzantine transparent, and unalterable storage of data. It consists
was introduced in (Oki and Liskov 1988), while in 1989 of a network of computers, called nodes, that preserve a
Lamport proposed the Paxos algorithm. In 2008, (Naka- shared, aligned record of trades. These transactions are
moto 2008) introduced Bitcoin using the Proof-of-Work pushed into blocks, which are linked together in a lin-
(PoW) algorithm which, on the contrary, considered the ear chain, with each block containing a timestamp and
malicious nodes. There are various types of consensus a link to the previous block. This arrangement allows
algorithms, and each one has its own strengths and limi- for the creation of a secure, decentralized database that
tations. Some of the most widespread consensus algo- is repellent to tampering and revision. The blockchain
rithms used in blockchain technology are Proof-of-Work architecture is aimed to be decentralized implying that
(PoW) (Gervais et al. 2016). This is the earliest consensus it is not regulated by a single central authority. Instead,
algorithm used by Bitcoin, and it relies on miners (nodes the network is preserved by a network of nodes that work
in the network) to compete to solve cryptographic prob- together to authenticate and record transactions. This
lems to validate transactions and create new blocks in decentralized structure permits the transfer of digital
the blockchain. PoW is secure and decentralized, but it assets, such as cryptocurrency, without the presence of
requires a lot of energy and can be sluggish and incom- intermediaries, such as banks or other financial organi-
petent. Proof-of-Stake (PoS) is a fresher consensus algo- zations (El loini and Bahl 2018; Wright and De Filippi
rithm that allows nodes in the network to authenticate 2015). The blockchain is also created to be secure. Each
transactions and create new blocks based on their stake block in the chain is secured using cryptographic tech-
(Bentov et al. 2014), or the amount of cryptocurrency niques, making it too complicated for anyone to modify
they hold. PoS is more energy-efficient and scalable than the data once it has been recorded. Additionally, the
PoW, but it can be more susceptible to centralization decentralized nature of the network means that no single
and attacks by wealthy nodes. Delegated Proof-of-Stake node can gain control of the network and manipulate the
(DPoS) is a variant of PoS where the nodes in the net- data (Zhang et al. 2019). The transparency of the block-
work vote to elect a small number of “delegates” who are chain is another key characteristic of its architecture.
responsible for validating transactions and creating new Because the network is decentralized and accessible, any-
blocks (Xu et al. 2019a). DPoS is faster and more scala- one can see the transactions that have been logged on the
ble than PoW or PoS, but it can be less decentralized and blockchain. This allows for better clarity and liability, as
more vulnerable to corruption or collusion. users can see exactly where their assets are and how they
Overall, consensus algorithms are a vital part of block- are being used (Sunny et al. 2020). The block header is
chain technology, and they play a major role in guaran- a vital part of the blockchain architecture (Puthal et al.
teeing the security, decentralization, and scalability of 2018). It is the first item that is accumulated in a block
blockchain networks. Different consensus algorithms and contains several important pieces of information,
have several trade-offs, and the choice of algorithm can including the following:
have significant outcomes for the assets and performance
of a blockchain network. In this section we have intro- • Reference link to the prior block, also known as the
duced the topic, while in section 2 we are going to pre- “parent” block. This link is what creates the chain of
sent an overview on the blockchain. In section 3, we are blocks. The blocks are connected through hash codes
going to explain some of the most substantial consensus (Fu et al. 2021).
algorithms, and various comparisons will be conducted • Timestamp, which specifies when the block was
among different types of consensus algorithms with built.
• Proof of work, which is a mathematical problem root node to verify the integrity of the data (Liu et al.
that must be cracked in order to insert a new block 2021).
to the chain. This proof of work is what permits the The block body is the part of a block that stores the
network to reach consensus on the state of the block- actual data of the transactions that are being recorded on
chain and inhibits the chain from being altered. the blockchain. The block body typically includes the fol-
• Merkle root, which is a hash of all the operations in lowing information (Ismail and Materwala 2019):
the block (Merkle 1988). This allows users to validate
the truthfulness of the transactions without having to • A list of the transactions that are being recorded in
transfer the entire block. the block. This will typically include information
• Nonce, (number used once) which is an arbitrary such as the sender and recipient of the transaction,
number that is used in the PoW calculation, and the the amount of the transaction, and any other relevant
miners are trying to find it. It is a 32-bit number that data.
usually takes 10 min to be guessed (Baldominos and • The cryptographic signatures of the transactions,
Saez 2019). which are used to verify the authenticity of the trans-
actions and ensure that they have not been tampered
A Merkle tree, also known as a binary hash tree, is a with.
data structure that is used in the blockchain to effec- • Any other relevant data, such as transaction fees or
tively verify the integrity of large sets of data. It is other metadata.
named after Ralph Merkle, who created the idea in
the 1980s (Merkle 1988). The formation of a Merkle In the block body of the Merkle tree, all information
tree is defined by the way that the hashes of the data of the transactions is being processed. Every leaf node
are arranged and mixed. In a Merkle tree, the separate of the tree stores the transaction information, and it
pieces of data are hashed and positioned in a binary is paired by a hash calculation and combined to gener-
tree structure, with each leaf node comprising the hash ate the hash until obtaining the root node of the tree.
of a single piece of data. These leaf nodes are then com- The hash value of the tree is able to detect any tamper-
bined in pairs, with each parent node containing the ing because if any leaf has been tampered with, that will
hash of its two child nodes. This process is repeated definitely change the hash value of the tree root (Ismail
until there is only one final “root” node, which contains and Materwala 2019). The Merkle tree structure is the
the combined hash of all the data in the tree. Figure 1 failsafe of the blockchain because it ensures the security
depicts the Merkle tree. The most important benefit of of the information in the blockchain. In the blockchain,
using a Merkle tree is that it permits effective authen- the word “node” is broadly being used. It is simply a
tication of the reliability of large sets of data. In the machine that performs computations. The node in the
blockchain, a Merkle tree can be used to certify that blockchain is behaving in P2P style (Li et al. 2018). The
an operation has been incorporated in a block without network is observing and coordinating the operations of
having to transfer the entire block. Since the root node the nodes in a decentralized conduct. The key task of the
of the tree contains the combined hash of all the opera- node in the blockchain is to check the information valid-
tions. Therefore, a user only requires transferring the ity and store the correct data. Nodes can be classified
into three types: the mining node, which is accountable
for creating and issuing the new blocks; the broadcasting
nodes, which is sending the information of transactions
and receiving limited amount of data; and the complete
node, which is responsible for issuing transactions, veri-
fying the data, and propagating transactions (Perard et al.
2018).
Working theory of blockchain

The blockchain (BC) is a decentralized, distributed data-
base that is used to preserve a constantly expanding list
of records, called blocks. Each block contains a times-
tamp and a link to the prior block, making it challeng-
ing to modify the data once it has been Preserved. This
structure permits the creation of a secure, visible, and
Fig. 1 Merkle tree architecture
unchanging ledger of trades that is handled by a network
of workstations on the internet, rather than a single main models are based on variables such the network type, the
authority (Singhal et al. 2018). It is valuable to know the consensus process employed, and the degree of decen-
fundamental concepts of decentralization consensus tralization. The following are some of the most well-
and Cryptography. Decentralization means the block- known blockchain structure types and classifications
chain is not controlled by one main authority. Instead, it (Guo and Yu 2022):
is retained by a network of computers, known as nodes,
that work simultaneously to prove and record transac- • Public vs. private: Access to a public blockchain is
tions. This decentralized structure allows the efficient open to everyone, but a private blockchain is only
and secure transfer of digital assets without the need available to a select number of individuals.
for mediators. The usage of mathematical algorithms to • Permissioned vs. permissionless: A permissioned
encrypt and secure data is called cryptography. In BC, blockchain needs users to be authorized in order
cryptographic methods are used to safeguard each block to engage in the network, whereas a permissionless
in the chain and verify the authenticity of transactions blockchain allows anybody to join in the network and
(Singhal et al. 2018). This makes it difficult for anyone validate transactions.
to alter the data once it has been recorded on the block- • Decentralized vs. centralized: A decentralized block-
chain while consensus is in the process of attaining an chain is one that is dispersed among a network of
agreement on the status of the blockchain (Xiong et al. nodes as opposed to a centralized blockchain, which
2022). This is done by using (PoW), which necessitates is one that is managed by a single entity.
the nodes to compete to solve a mathematical problem • Federated vs. consortium: A federated blockchain is
to attach a new block to the chain. The first node to solve one that is managed by a number of entities, but a
the problem is permitted to add the new block, and the consortium blockchain is managed by a number of
other nodes then validate that the block is acceptable carefully chosen entities.
before adding it to their own copies of the blockchain.
Eventually BC is a decentralized secure and transpar- Blockchain can also be described as a layered system.
ent system for saving and validating transactions. It uses The majority of Blockchains can be designed with the fol-
cryptographic methods and consensus-based tactics lowing layers: network layer, data layer, consensus layer,
to maintain a secure and unchanging ledger. The basic incentive layer, contract layer, and application layer as
mechanics of how the blockchain acts can be condensed shown in Fig. 2.
as follows (Singhal et al. 2018): The data layer, network layer, consensus algorithm,
incentive layer, contract layer, and application layer
• A transaction is started by a user and disseminated to are some of the layers that to make up the blockchain.
the network of nodes.
• Public vs. private: Access to a public blockchain is
open to everyone, but a private blockchain is only
available to a select number of individuals.
• Permissioned vs. permissionless: A permissioned
blockchain needs users to be authorized in order
to engage in the network, whereas a permissionless
blockchain allows anybody to join in the network and
validate transactions.
• Decentralized vs. centralized: A decentralized block-
chain is one that is dispersed among a network of
nodes as opposed to a centralized blockchain, which
is one that is managed by a single entity.
• Federated vs. consortium: A federated blockchain is
one that is managed by a number of entities, but a
consortium blockchain is managed by a number of
carefully chosen entities.
Classes and structure of blockchain

There are several concepts and classifications for the
Fig. 2 Layered system of blockchain
construction of the blockchain. These classifications and
Data blocks, a linked list, Merkle trees, and other data

structures that make use of timestamps, hash func-
tions, and cryptography make up the data layer. It
provides the framework for blockchain management,
organization, and data storage. All nodes in the chain
are connected via the network layer using a peer-to-
peer network mechanism, allowing them to transact,
send, and verify data. The blockchain’s basic technol-
ogy, the consensus algorithm, determines which nodes
have the right to record transactions and enables them
to swiftly agree on the information included in a block.
This ensures the consistency and security of the data
Fig. 3 Categories of consensus algorithms
while also improving the blockchain’s computational
efficiency. By incorporating rewards and punishments
into the blockchain’s distribution mechanism, the
incentive layer incentivizes nodes to provide services. Consensus algorithms
The contract layer consists of smart contracts and algo- Consensus is the process of reaching agreement among
rithms that are executed automatically when certain a group of people or entities on a specific decision or
conditions are met, allowing for the customization of action. In a blockchain, consensus is used to guaran-
blockchain transactions. The application layer com- tee that all nodes on the network agree on the current
bines the underlying structure, script code, and smart state of the network and the authenticity of transactions
contracts to allow blockchain to be used in a variety of (Xiong et al. 2022). This is vital for preserving the secu-
real-world scenarios. Based on data access authority, rity and integrity of the blockchain. Different blockchain
blockchain can be classified into three types (Sheth and platforms use different algorithms, such as proof of work,
Dattani 2019): public, consortium, and private. Public proof of stake, or proof of authority to achieve consensus
chains allow nodes to join and leave the network with- among the nodes on the network.
out requiring permission, but they are slow in terms of
transaction processing speeds and low overall perfor- Concept of consensus (historical background)
mance. Examples of public chain applications include Under certain conditions, Edmund Eisenberg and David
Bitcoin and Ethereum (Yang et al. 2020). Consortium Gale conducted a study in 1959 on how individuals with
chains require nodes to register and be approved by a subjective consciousness in the same space can reach a
central organization before joining the network. They consistent consensus probability distribution. This is
have lower degrees of visibility than public chains and known as the consistency problem, and it is also referred
are suitable for enterprises or companies that form con- to as the consensus problem. The study at the time only
sortia. Nodes in consortium chains do not completely focused on scenarios where the number of nodes was
trust each other and require consensus algorithms to limited and trustworthy, but this had limitations and
reach agreement. Private chains are controlled by a was not suitable for open internet scenarios. Satoshi
single internal entity and only allow access to selected Nakamoto later extended the consistency problem to
and verified participants. They have the lowest degree the internet environment with open scenes and massive
of decentralization but the fastest transaction pro- nodes in the Bitcoin system, proposing the Byzantine
cessing speeds. Table 1 compares the characteristics Generals problem, which is known as the Byzantine Gen-
of the three types of blockchain. In brief, this section erals problem. This issue, known as Byzantine failures, is
reviewed various blockchain concepts and classifica- significant in the field of blockchain technology. The con-
tions based on variables such as network type, consen- sensus algorithm ensures data consistency among nodes
sus process, and degree of decentralization. Public vs. for a specific proposal. Different consensus algorithms
private blockchain structures, permissioned vs. permis- have varying capabilities for ensuring that nodes receive
sionless, decentralized vs. centralized, and federated balanced accounting rights. An excellent consensus algo-
vs. consortium were also covered. Blockchain’s layers, rithm can keep the blockchain network active and pro-
which include the data layer, network layer, consensus vide a steady stream of effective computing power to the
layer, incentive layer, contract layer, and application entire network, whereas a poorly designed algorithm
layer were explained as well. Furthermore, blockchain can cause the entire network to become easily paralyzed
types based on data access authority were considered in when attacked. Consensus algorithms are classified into
the review. non-Byzantine fault-tolerant algorithms, Byzantine
fault-tolerant algorithms, DAG based, and Hybrid. Fig- group of learner nodes indicating that the proposed
ure 3 shows the different categories of consensus algo- value has been accepted by the majority of acceptors.
rithms, in addition to the well-known algorithms under • When the accepted value is received, the learner
each of these categories. nodes update their local copies of the data item with
the accepted value.
Non‑Byzantine consensus algorithms • If a learner node receives a different value from a
Non-Byzantine error is a type of system failure that master node for the same data item, it must repeat
occurs in distributed systems without the presence the process from step 2 to reach consensus on the
of malicious nodes. This can include issues such as correct value.
machine downtime and node reporting errors (Xiong
et al. 2022;Han and Gao 2020). Non-Byzantine fault- Paxos’ algorithm uses a series of messages exchanged
tolerant algorithms are designed to handle these types between nodes to reach consensus on a proposed value
of errors, but they cannot guarantee the security of data (Lamport Lamport 2001; De Prisco et al. 2000). The algo-
and system stability when malicious nodes are present. rithm is designed to be fault-tolerant, meaning that it can
Therefore, non-Byzantine fault-tolerant algorithms are continue to function even if some nodes in the network
typically only used in closed environments with high fail or behave maliciously. This allows the algorithm to
credibility between nodes, such as consortium chains or ensure that all nodes in the network agree on the same
private chains. These algorithms offer high performance value for a given data item, Fig. 4 shows the algorithm
and strong tolerance for non-Byzantine errors. flow.
Paxos’ algorithm VR consensus algorithm

The Paxos algorithm is a consensus algorithm that ena- Viewstamped Replication (VR) is a distributed consensus
bles distributed network nodes to reach agreement on a algorithm that allows nodes in a network to agree on the
proposed value (Lamport 2001). It was first introduced order and integrity of transactions in a distributed data-
by computer scientist Leslie Lamport in 1998 and it has base. Each node maintains a log of transactions and a
been used in many distributed systems. Unlike other current “view” number, which indicates its current state
consensus algorithms Paxos does not depend on a cen- (Liskov and Cowling 2012). When a node wants to pro-
tral authority to organize the consensus process. Instead, pose a new transaction, it sends a request to a designated
it uses messages exchanged between nodes to reach primary node, which is responsible for ordering the
consensus on a proposed value. Paxos is fault-tolerant, transactions and broadcasting them to the other nodes.
which means that it can remain functional even when Figure 5 illustrates the VR algorithm. If the primary node
some nodes in the network fail or behave maliciously fails or becomes unavailable, the other nodes can initiate
(De Prisco et al. 2000). The algorithm uses three types of a change of view process to select a new primary node
nodes: proposers, which propose values for data items; that will continue the process of ordering and broadcast-
acceptors, which evaluate and accept or reject proposed ing transactions, using the log and view numbers from
values; and learners, which receive accepted values and
update their own local copies of the data. There are many
variations and optimizations of the Paxos algorithm, but
it remains a powerful tool for achieving consensus in dis-
tributed systems. The Paxos algorithm is a complex algo-
rithm, but its basic workflow can be broken down into
the following steps (Lamport 2001):
• A proposer node suggests a value for a specific data

item. This value is broadcast to a set of acceptor
nodes.
• The acceptor nodes receive and evaluate the pro-
posed value. If they agree that the proposed value is
correct, they send a message to the proposer indicat-
ing that the value has been accepted.
• If the proposed value is accepted by the majority of
acceptor nodes, the proposer sends a message to a
Fig. 4 Flow chart of Paxos algorithm
Table 1 Characteristics of blockchain types
Feature Types
Public Consortium Private
Pros Anyone can join and participate without permission or Transactions are processed faster than public blockchains due Transactions are faster, scalable, efficient, secure and fully
identity verification (Open) to lower network latency and higher throughput (Faster) controlled by a single node
Hussein et al. Cybersecurity
All transactions are publicly visible and verifiable by Transactions can handle higher volumes than public block‑ Lower network latency and higher throughput make
anyone (Transparent) chains due to lower computational complexity and resource transactions faster than public blockchains
consumption (Scalable)
Transactions are validated by a large number of nodes Transactions consume less energy than public blockchains Lower computational complexity and resource consump‑
using a cryptographic consensus mechanism that pre‑ due to simpler consensus mechanisms that do not require tion make transactions scalable to higher volumes than
vents double-spending, fraud, or censorship (Secure) intensive proof-of-work or proof-of-stake algorithms (Efficient) public blockchains
(2023) 6:30
Transactions are permanently recorded on a distributed Transactions are validated by a smaller number of nodes using Simpler consensus mechanisms that do not require inten‑
ledger that cannot be altered or erased by anyone a cryptographic consensus mechanism that prevents double- sive algorithms make transactions efficient and consume
(Immutable) spending, fraud, or censorship (Secure) less energy than public blockchains
Transactions are executed according to predefined rules Transactions are executed according to predefined rules that Cryptographic consensus mechanism that prevents
that cannot be changed by anyone without consensus can be changed by agreement from a subset of nodes (More double-spending, fraud or censorship makes transactions
from all nodes (Censorship-resistant) control over governance and consensus) secure
Predefined rules that can be changed by decision from
a single node give full control over governance and
consensus
Cons High electricity consumption and processing power Less transparent and democratic than public blockchain Highly centralized and dependent on a single authority
required for consensus
Low scalability due to limited throughput and high Prone to collusion or corruption among the controlling enti‑ Lacks immutability and security compared to public
latency ties blockchain
Vulnerable to 51% attacks if a single entity gains majority May face legal or regulatory challenges due to cross-border May not benefit from network effects or innovation due to
control over the network transactions limited participation
Usage Sce‑ Cryptocurrencies: Online payments without intermediar‑ Cross-border payments (e.g., Ripple): A global payment net‑ Enterprise solutions: IBM Blockchain Platform - A plat‑
nario ies or trusted third parties (e.g., Bitcoin) work that enables fast, cheap, and secure transactions across form for developing, deploying, and managing private
different currencies and jurisdictions blockchain networks for business use cases such as trade
finance, asset tracking, digital identity, etc
Smart contracts: Decentralized applications that run on Supply chain management (e.g., Hyperledger Fabric): A Banking and finance: Corda - A distributed ledger platform
self-executing contracts (e.g., Ethereum) framework for building enterprise-grade blockchain solutions designed for financial institutions that enables secure
that enable transparency, traceability, and efficiency across transactions with smart contracts
complex supply chains
Decentralized applications: Applications that run on Identity verification (e.g., Sovrin): A decentralized identity net‑ Healthcare records: MedRec - A prototype for managing
distributed networks without centralized servers or work that enables self-sovereign identity management using electronic medical records using blockchain technology
authorities (e.g., Uniswap) verifiable credentials
Centralization Network and data are not controlled by any single entity Anyone can join and participate without permission or iden‑ Network and data are controlled by a group of entities
Level tity verification
All nodes have equal rights and responsibilities Some nodes have more rights and responsibilities than others Only one node has all rights and responsibilities
Participants Any node can join and do transactions without permis‑ Members only can join and do transactions with permission Individual or company can join and do transactions with
sion or identity verification or identity verification by a group of organizations that jointly permission or identity verification by a single entity that
manage the network owns
Page 7 of 22
(Kim et al. 2021). One key feature of RAFT is that it pro-

vides strong consistency (Hu and Liu 2020), meaning that
all nodes in the system will eventually agree on the same
value or state. It also provides fault tolerance, as the sys-
tem can continue to operate even if some nodes fail or
become unavailable. RAFT is a popular consensus algo-
rithm that is widely used in distributed systems such as
distributed databases and distributed file systems. It is
well-known for its simplicity and ease of implementation,
making it an excellent choice for systems requiring solid
consistency and fault tolerance. RAFT has specific appli-
Fig. 5 VR architecture
cations such as distributed key-value stores, distributed
configuration management systems, and distributed file
systems (Le Brun et al. 2021). It is also used as a foun-
the previous primary node to ensure consistency. VR is dation for other distributed systems such as distributed
able to handle node failures and network partitions with- databases and distributed messaging systems. To sum-
out sacrificing consistency or security, making it suitable marize, this section discussed three non-Byzantine
for use in distributed systems where strong consistency consensus algorithms, namely Paxos, VR, and RAFT.
and fault tolerance are required, such as blockchain net- Non-Byzantine errors are failures that occur in distrib-
works. However, it may be slower and less scalable than uted systems without the presence of malicious nodes.
other algorithms like PoW and PoS. VR was first devel- These algorithms are designed to handle these types
oped in 1988 by Liskov and has since been used in a vari- of errors but cannot guarantee the security of data and
ety of applications, including distributed databases and system stability when malicious nodes are present. Paxos
file systems, as well as blockchain networks (Oki and and VR are fault-tolerant algorithms that are suitable
Liskov 1988). for use in distributed systems where strong consistency
and fault tolerance are required, such as blockchain net-
RAFT algorithm works. Paxos is a complex and inefficient algorithm that
RAFT is a distributed consensus algorithm that allows relies too much on messages exchanged between nodes
a group of nodes (computers) in a distributed system to to reach consensus on a proposed value. We recommend
reach agreement on a single value or state of the system. VR, which allows nodes to agree on the order and integ-
It was developed as an alternative to other consensus rity of transactions in a distributed database without too
algorithms, such as Paxos, and is designed to be easier much overhead. RAFT is another consensus algorithm
to understand and implement. In RAFT, the nodes in the that is designed to be easier to understand and imple-
system are divided into two types: leaders and followers ment than Paxos, but it has some drawbacks. It gives too
(Kim et al. 2021). The leader is responsible for propos- much power to the leader, who is responsible for propos-
ing new values or states for the system and for replicating ing new values or states for the system, and the follow-
those values to the followers. The followers are respon- ers have to accept or reject the proposals based on a set
sible for accepting or rejecting the proposals based on a of rules. If the leader becomes unavailable, the followers
set of rules, and for voting for a new leader if the current have to initiate a leader election process to select a new
leader becomes unavailable. Figure 6 shows the general leader, which can cause delays and conflicts.
architecture of the algorithm (Tian et al. 2021). The pro-
cess of reaching consensus in RAFT works as follows: Byzantine‑based consensus algorithms
the leader proposes a new value or state for the system Byzantine fault-tolerant algorithms (BFTs) are consid-
and broadcasts it to the followers. The followers accept ered the ability of a distributed network to achieve con-
or reject the proposal based on a set of rules. If a major- sensus on a value, even when some nodes in the network
ity of followers accept the proposal, it is considered to be do not respond correctly or at all.
“committed”. The leader then replicates the committed The purpose of a BFT mechanism is to protect against
value to the followers and updates its own log of commit- system failures by using a collective decision-making
ted values. If the leader becomes unavailable, the follow- process that considers the input of both correct and
ers can initiate a leader election process to select a new faulty nodes, in order to minimize the impact of faulty
leader as shown in Fig. 7. The new leader is chosen based nodes. The concept of BFT originated from the Byzantine
on the log of committed values, with the node that has Generals’ Problem (Lamport et al. 2019). Byzantine fault
the most up-to-date log being selected as the new leader tolerance can be achieved when the functioning nodes in
Fig. 6 RAFT architecture
Fig. 8 General phases of pBFT
secondary to primary, typically in the event of primary

node failure. The goal of pBFT is for all honest nodes
to reach a consensus about the system’s state using the
majority rule. A pBFT system can function as long as
the maximum number of malicious nodes is less than
or equal to one-third of all the nodes in the system. As
the number of nodes increases, the system becomes
Fig. 7 Election process
more secure. pBFT consensus rounds are divided into
four phases as shown in Fig. 8: (Wang et al. 2019)
• The client sends a request to the primary (leader)

the network reach a consensus on their values. If a mes- node.
sage is not received within a certain time limit, it can be • The primary (leader) node broadcasts the request to
assumed that the message from that node is faulty, and all the secondaries (backups).
a default vote value can be assigned. Additionally, if the • The nodes (primary and secondaries) perform the
majority of nodes respond with a correct value, a default requested service and then send a reply back to the
response can be assigned (Velliangiri and Karthikeyan client.
2020). • The request is successfully served when the client
receives ’m+1’ replies from different nodes in the
Practical Byzantine fault‑tolerant (pBFT) network with the same result, where m is the maxi-
Practical Byzantine Fault Tolerance (pBFT) is a con- mum number of faulty nodes allowed.
sensus algorithm developed in the late 1990s by Bar-
bara Liskov and Miguel Castro that is optimized for The primary (leader) node is changed during every
asynchronous systems (Zheng and Feng 2021), where view (pBFT consensus round) and can be replaced by a
there is no upper bound on when responses to requests view change protocol if a predefined amount of time has
will be received. It was designed to have low overhead passed without the leading node broadcasting a request
time and address issues with other Byzantine Fault to the backups (secondaries). If necessary, a majority of
Tolerance solutions. pBFT has applications in distrib- the honest nodes can vote on the legitimacy of the cur-
uted computing and blockchain technology. practical rent leading node and replace it with the next leading
Byzantine Fault Tolerance (pBFT) aims to provide a node in line.
practical solution for Byzantine state machine replica-
tion in distributed systems, even when malicious nodes
are present (Wang et al. 2019). In a pBFT-enabled sys- Proof of work (PoW)
tem, nodes are sequentially ordered with one node The process of adding a new block to the blockchain,
designated as the primary (or leader) and the others called “mining,” is performed by nodes named “min-
as secondaries (or backups). Any eligible node in the ers”. Miners compete to solve a complicated math-
system can become the primary by transitioning from ematical problem in a proof-of-work (PoW) consensus
algorithm. The first miner to solve the problem is author- synchronized block in the network. PoW is used in appli-
ized to create a new block of transactions and add it to cations such as Bitcoin and Ethereum, and it takes 10 min
the blockchain which is a decentralized and unchange- on average to generate a block and one hour to confirm
able record of all The process of adding a new block to it (Vilim et al. 2016). Ethereum, in addition to being a
the blockchain, called “mining,” is performed by nodes digital currency, also serves as a platform for developing
named “miners”. Miners compete to solve a complicated applications. The mining procedure is depicted in Fig. 9.
mathematical problem in a proof-of-work (PoW) con-
sensus algorithm. The first miner to solve the problem Proof of stake (PoS)
is authorized to create a new block of transactions and Proof-of-Work (PoW) is a popular method for achieving
add it to the blockchain which is a decentralized and distributed consensus, as seen in the Bitcoin implemen-
unchangeable record of all network transactions (Fullmer tation. However, PoW consumes a significant amount of
and Morse 2018). The solution to the problem is authen- energy, particularly during the Bitcoin mining process
ticated by the other miners, and if it is correct the new (Saad et al. 2021). A PoW system increases an entity’s
block is added to their copy of the blockchain. Mining chances of mining a new block if it has more computa-
requires a substantial amount of computational power, tional resources. Aside from the energy requirement,
and the miner who solves the problem is rewarded with there are several drawbacks to using a PoW-based consen-
a certain number of cryptocurrency units. This incentive sus mechanism as mentioned in table 2. A Proof-of-Stake
encourages miners to participate in the process and con- (PoS) mechanism may be a better alternative. PoS is a type
tributes to the security of the blockchain (Gemeliarana of consensus algorithm in which the next block is chosen
and Sari 2018). PoW consensus algorithms are resistant based on the stake (amount of cryptocurrency held) of
to tampering and fraud because changing a block’s con- the miner (Ganesh et al. 2019), rather than their compu-
tents would require redoing the proof-of-work for that tational power. This can be a more energy-efficient way to
block and all subsequent blocks, making it difficult for achieve distributed consensus. In a Proof-of-Stake (PoS)
a single entity to control or alter the blockchain. Table 2 consensus algorithm, nodes on a network can become
shows the pros and cons of PoW. In a proof-of-work candidates to validate new blocks by staking a certain
(PoW) consensus algorithm, each block of transactions is amount of cryptocurrency. An algorithm then selects one
linked to the previous block using a cryptographic hash of the candidates to validate the new block and earn the
value. The Process of mining is performed by “miners.” transaction fee. The selection algorithm uses a combina-
A miner must select a random value (called a “nonce”) tion of the candidate’s stake (amount of cryptocurrency
and calculate the hash value of the block header, which held) and other factors, such as coin age and randomiza-
includes the nonce and other information such as previ- tion, to ensure fairness among all the nodes on the net-
ous block hash and transaction data. If the hash value is work. One such factor is coin age (Nguyen et al. 2019),
less than a predetermined target value, the block is added which tracks how long a candidate node has been a vali-
to the blockchain. This process is verified by other min- dator. The longer a node has been a validator, the higher
ers in the network. The SHA-256 hash function is used its chances of being selected as the new validator. Another
in Bitcoin (Gayoso Martinez et al. 2020). By setting a tar- factor is random block selection, in which the validator is
get value for every 2,016 blocks, the difficulty of finding chosen based on a combination of the lowest hash value
a valid hash value is maintained. Two miners may some- and the highest stake. The node with the best weighted
times add a block at the same time; a process called “fork- combination of these factors becomes the new validator.
ing”. In this case, all network nodes agree on the most Figure 10 illustrates the PoS Algorithm workflow.
Fig. 9 Mining process in Bitcoin network Fig. 10 Workflow of PoS

Table 2 Pros and Cons of PoW

Pros Cons
Provides a solid mechanism for achieving consensus and preventing Requires a lot of energy consumption and computational power
abuses and misuses
Rewards miners for securing the network and validating transactions May lead to centralization due to mining pools and specialized hardware
Enables trustless transactions without intermediaries or authorities Limits scalability due to low throughput and high latency
Resists attacks such as double-spending, censorship, or denial-of-service May suffer from stagnation due to low incentives for innovation or
improvement
Proof of burn (PoB) as opposed to PoW’s ten minutes (Mohamed and Ibra-
In Proof of Burn (PoB), validators demonstrate their him 2020). PoC increases miners’ chances of winning
commitment to the system by “burning” coins or send- the mining competition by providing more solutions,
ing them to an address from which they can never be or “plots” on a computer. Overall, PoC is intended to
retrieved (Karantias et al. 2020) making it unspendable. address the energy and decentralization issues that
This process is used to determine which validators will be plague PoW mining, making it a potentially appealing
able to mine the next block in the system. Validators may option for blockchain projects. Proof-of-Capacity (PoC)
burn the native currency of the blockchain application or consists of two main components: plotting and mining
the currency of an alternative chain, such as bitcoin, to (Aggarwal and Kumar 2021). Plotting involves creating
increase their chances of being selected for block mining. a series of precomputed hashes and storing them on a
Rather than investing in expensive hardware, PoB allows hard drive using the Shabal hash function, which is used
validators to show their long-term commitment to the by the cryptocurrency Burstcoin (Bamakan et al. 2020).
system through a short-term sacrifice of coins (Karantias This process can take several days or weeks, depending
et al. 2020). The more currency a miner burns, the greater on the size of the hard drive. The hashes are grouped
their chances of being selected to mine the next block on into “scoops”, each of which consists of two neighbor-
the system. The idea behind this is that by destroying ing hashes. Mining entails calculating a scoop number
their currency, the miner is showing a long-term com- and applying it to each nonce stored on the hard drive to
mitment to the system and giving up a short-term gain determine a “deadline” value. If no one else has created a
in exchange for a potential long-term profit (Yusoff et al. new block within that timeframe, the miner chooses the
2022). To prevent early adopters from having an unfair nonce with the shortest deadline and uses it to do so. If
advantage, PoB has a system in place that allows for the the miner creates the block before the deadline, they are
periodic burning of cryptocurrency to maintain min- rewarded with a block reward.
ing capacity. As new blocks are mined, the energy of the
burned coins decreases slightly, resulting in a deflation- Proof of activity (PoA)
ary process in which the overall quantity of currency In Proof-of-Activity (PoA), miners utilize their comput-
decreases over time, potentially increasing its value. In ing power to solve cryptographic problems similar to
contrast, cryptocurrencies that increase in quantity over Proof-of-Work (PoW) while also taking into account the
time tend to lose value. amount of stake (e.g., tokens or cryptocurrency) that a
miner holds, similar to Proof-of-Stake (PoS). This cre-
Proof of capacity (PoC) ates a hybrid system that combines the security of PoW
Proof-of-Capacity (PoC) is a new mining method that with the energy efficiency of PoS (Kaur et al. 2021). By
is currently being used by the cryptocurrency Burstcoin considering a miner’s stake, the network can prioritize
(Mohamed and Ibrahim 2020). This method involves those with a long-term interest in its success rather than
using hard disk space for mining and has the potential to just those with the most powerful computing resources.
be a more energy-efficient alternative to the commonly PoA can be an effective way to balance security and effi-
used Proof-of-Work (PoW) mining method. However, ciency in a blockchain, but it may also be more complex
as the network has grown, mining has become increas- to implement and potentially less secure compared to
ingly difficult and energy-intensive, requiring special- pure PoW or PoS systems. In Proof-of-Activity (PoA), the
ized hardware known as ASICs to be effective. PoC seeks mining process begins like a Proof-of-Work (PoW) pro-
to address these issues by requiring miners to commit cess, with miners using their computing power to solve
processing power and hard disk storage before mining mathematical equations and create new blocks. When a
begins, resulting in a faster system than PoW. PoC also new block is successfully mined, the system transitions
has the advantage of producing blocks in four minutes to a Proof-of-Stake (PoS) phase. A group of validators
is randomly selected to sign the new block (Andola

et al. 2020), which is validated based on the details in
its header. Validators with a larger amount of crypto-
currency have a higher chance of being chosen as sign-
ers. If the required number of validators sign the new
block, it is considered complete and added to the exist-
ing blockchain, with the transactions in the block being
recorded. If the selected signers are not present to sign
the new block, the process moves to the next winning
block, where a new group of validators is chosen based
on their cryptocurrency holdings. If a winning block
does not receive the required number of signatures to
become complete, the process continues. The first miner
and any validators who contributed to the new block are
rewarded. PoA is criticized for its partial use of PoW and
PoS, but it also prevents the risk of a 51% attack (Shrimali Fig. 11 dPoW consensus mechanism
and Patel 2022; Sayeed et al. 2019).
Delayed proof of work (dPoW)

These delegates are chosen by the stakers through a vot-
Delayed Proof of Work (dPoW) is a consensus algo-
ing process, and the number of votes a delegate receives
rithm that aims to improve the security of a blockchain
is determined by the number of tokens they hold. The
network by incorporating elements from a more secure
delegates with the most votes are responsible for creating
blockchain. This is achieved by allowing a secondary
new blocks and validating transactions, while the remain-
blockchain, known as the “notary chain,” to record hashes
ing delegates act as backups (Saad et al. 2020; Bachani
of blocks from the primary blockchain, known as the
and Bhattacharjya 2022). DPoS is intended to be more
“target chain”. In a dPoW system, miners on the notary
efficient and scalable than other consensus algorithms
chain compete to create new blocks just like in a tradi-
such as Proof-of-Work (PoW) or Proof-of-Stake (PoS), as
tional Proof-of-Work (PoW) system. However, instead of
it allows a small group of delegates to handle the major-
including transactions in these blocks, the miners include
ity of the network’s transactions. It also gives token hold-
the hashes of blocks from the target chain. This process
ers more influence over the network, as they can vote for
is known as “notarization”. The notarized blocks are then
or against specific delegates. However, DPoS can also be
added to the notary chain,
more centralized as the power to create new blocks and
providing an additional layer of security for the target
validate transactions is held by a small group of delegates,
chain (Osadchuk and Oliynykov 2019). If an attacker
making it susceptible to collusion and manipulation (Hu
attempted to change the transactions on the target chain,
et al. 2021). Table 3 shows the pros and cons for each
they would also have to change the corresponding nota-
PoX consensus algorithm. Table 3 compares the different
rized block on the notary chain, which would be much
types of PoX consensus algorithm in terms of advantages
more difficult due to the notary chain’s increased secu-
and disadvantages, and we can conclude form Table 3
rity. This makes it much more difficult for an attacker to
that the best consensus algorithm depends on the situa-
successfully alter the target chain, increasing its secu-
tion and the predefined constraints of the system itself.
rity (Sayeed et al. 2020). dPoW can be a good way to
Table 4 shows the advantages and disadvantages of differ-
increase the security of a blockchain network, especially
ent PoX algorithms in detail in terms of energy efficiency,
for smaller or less secure networks that are more vulner-
throughput, scalability, and security. Table 3 and Table 4
able to attacks. It does necessitate the use of a second-
provide an overview of commonly utilized consensus
ary, more secure blockchain, which adds complexity and
algorithms in blockchain networks. Each algorithm has
may not be practical in all situations. Figure 11 depicts
unique advantages and disadvantages that should be
the algorithm.
taken into consideration when selecting an appropriate
consensus algorithm for a particular blockchain project.
Delegate proof of stake (DPoS) Proof of Work (PoW) is a well-known consensus algo-
In Delegated Proof of Stake (DPoS), token holders (stak- rithm that offers substantial protection against attacks
ers) can assign their voting power to delegates or wit- but requires significant energy consumption and com-
nesses to create new blocks and validate transactions on putational resources. On the other hand, Proof of Stake
the blockchain (Yang et al. 2019).
Table 3 Pros and cons for each PoX consensus algorithm
Consensus algorithms References Pros Cons
Proof of Work (PoW) (Yang et al. 2019; Puthal and Protect against DoS attacks and spam Intensive usage of energy and resources required for the system
Mohanty 2018; Dey 2018)
Secure the entire network The risk of a 51% attack
Prevent double spending attacks in distributed systems
Hussein et al. Cybersecurity
Proof of Stake (PoS) (Vashchuck and Shuwar 2018; Less power consumption and less hardware usage compared Coin hoarding
Nguyen et al. 2019; Yang to PoW
et al.2019)
Monopolization
Double spending
(2023) 6:30
Proof of Burn (PoB) (Karntias et al. 2020; Menon Reduced reliance on computational resources The destruction of coins leads to resource waste
et al. 2022)
User commitment over the long term Hoarding of coins
Decentralized structure Attempts to manipulate the system through control of a large number
of coins
Lower energy consumption compared to proof-of-work systems
Proof of Activity (PoA) (Salimitari et al. 2020; Wang This system is more secure against 51% attacks than PoS and Intensive use of resources is required to carry out certain actions
et al. 2020) PoW
It is resistant to DoS attacks The possibility of a monopoly forming increases if there is a high risk of
penalty for attempting to double sign transactions
It promotes decentralization There is a balance to be struck between the energy needed to perform
certain actions and the potential for monopoly power
Proof of Space (PoS) (Park et al. 2018; Benisi et al. Energy efficiency due to the use of low-power hard drives Verification efficiency and storage availability are challenging task
2020) instead of specialized hardware such as ASICs, CPUs, and GPUs
Greater potential for decentralization The potential for a probabilistic monopoly with large amounts of space
Delegated Proof of Stake (Saad et al. 2020; Do et al. Speed: DPoS can facilitate faster transaction processing and Potential for centralization: The use of delegates in DPoS can potentially
(DPoS) 2019) block production compared to other proof of stake algorithms lead to centralization if the same small group of delegates are consist‑
ently elected to represent the network
Energy efficiency: DPoS uses significantly less energy than proof Limited participation: Only those with a significant number of tokens can
of work algorithms, making it more environmentally friendly participate in the delegate selection process, which may exclude some
members of the community
Decentralization: DPoS allows for a more decentralized network Complexity: DPoS is a more complex system than traditional proof of
by allowing token holders to vote for “delegates” who will repre‑ stake algorithms, which may make it more difficult to understand and
sent them in the decision-making process implement
Delayed Proof of Work (Sayeed et al. 2019) Increased security: DPoW uses a secondary blockchain to secure Complexity: DPoW is a more complex system than traditional proof of
(dPoW) the main chain, providing an additional layer of protection work algorithms, which can make it more difficult to understand and
against 51% attacks implement
Decentralization: DPoW can help to decentralize the mining Dependency on secondary chain: DPoW relies on a secondary chain to
process by allowing a wider range of miners to participate in the secure the main chain, which means that if the secondary chain becomes
network compromised, the main chain may also be at risk
Energy efficiency: DPoW uses less energy than traditional proof Compatibility issues: DPoW may not be compatible with certain types of
of work algorithms, making it more environmentally friendly software or hardware, which could limit its use in certain situations
Page 13 of 22
Table 4 Pros and cons for each PoX consensus algorithm in the chain in a particular order and do not require min-
terms of various criteria ers to do expensive proof-of-work calculations. Instead,
Algorithm Energy Throughput Scalability Security transactions are added to the DAG in parallel, result-
efficiency ing in higher throughput. Examples of DAG-based con-
sensus algorithms include IOTA’s Tangle, Nano’s Block
PoW Low Low Low High
Lattice, and Hashgraph. These algorithms have been
PoS High Medium Medium Medium
used in a variety of applications, like distributed ledger
PoB High Medium Medium Medium
technologies, peer-to-peer networks, and decentralized
PoA Medium High High High
applications. DAG-based consensus algorithms are still
PoS (space) High Medium High Low
a relatively new and rapidly evolving field of technology,
DPoS High Very high Very high Low to
medium and there is ongoing debate about their relative advan-
dPoW High High High High
tages and disadvantages compared to traditional block-
chain-based consensus algorithms. IOTA is a distributed
ledger technology with a primary structure based on a
directed acyclic graph (DAG). It was designed to help the
(PoS) presents an alternative approach that consumes
Internet of Things (IoT) (Silvano and Marcelino 2020),
less energy but may be susceptible to coin hoarding and
which is a network of interconnected devices that can
monopolization.
interact and transfer data. One of IOTA’s key character-
Proof of Burn (PoB) is a relatively new consensus algo-
istics is its scalability, which is achieved through the use
rithm that prioritizes user commitment over the long
of the Tangle, a DAG-based consensus algorithm. Unlike
term. However, coin destruction leads to resource waste,
traditional blockchain-based systems that rely on miners
and coin hoarding can manipulate the system. Proof of
to perform proof-of-work calculations to validate trans-
Space (PoS) is energy-efficient and highly decentralized,
actions, IOTA employs a different approach known as
using low-power hard drives, but it may be vulnerable
“proof-of-workless” consensus. In this approach, each
to monopolization with a large amount of space. Proof
transaction in the Tangle must validate two other trans-
of Activity (PoA) is a recently developed consensus algo-
actions before it can be added to the DAG. This allows
rithm that resists DoS attacks and encourages decen-
IOTA to achieve a high level of throughput and low
tralization, but certain actions require resources. Delayed
transaction fees, making it suitable for use in the IoT.
Proof of Work (dPoW) adds an extra layer of protection
IOTA also utilizes a unique form of cryptocurrency
against 51% attacks, and it is more energy efficient than
called MIOTA, which is used to facilitate transactions
traditional PoW algorithms. However, it is more com-
on the IOTA network. In addition to being used as a
plex and has dependency issues. Delegated Proof of Stake
means of exchange, MIOTA can also be used to repre-
(DPoS) is a consensus algorithm that allows token hold-
sent data or store values (Lamberti et al. 2019). IOTA has
ers to vote for delegates who represent them in the deci-
been used in various applications, including supply chain
sion-making process. DPoS is fast and energy-efficient,
management, smart cities, and energy markets. How-
but the use of delegates may lead to centralization and
ever, it’s worth noting that IOTA has faced some contro-
exclude some members of the community from partici-
versy and criticism in the past, including concerns about
pating. In general, there is no one-size-fits-all solution for
the security and centralization of its network. Nano is
selecting a consensus algorithm. The choice of algorithm
a cryptocurrency that utilizes a directed acyclic graph
will depend on the specific needs and goals of the block-
(DAG)-based consensus algorithm called the Block Lat-
chain project.
tice (Morais et al. 2020). In the Block Lattice, each Nano
account has its own blockchain, called an “account-
DAG‑based consensus algorithms
chain,” which is used to track the balance and transaction
DAG is a distributed ledger technology that is built on
history of the account. The Block Lattice’s scalability is
the principles of directed acyclic graphs (DAGs). These
one of its main benefits, as transactions can be processed
algorithms portray transactions as nodes within the
in parallel due to each account having its own blockchain,
DAG, and the edges between nodes display the interde-
allowing for a high level of throughput. The Block Lat-
pendence between the transactions (Chen et al. 2018).
tice also uses a proof-of-workless consensus algorithm,
One of the main advantages of DAG-based consensus
meaning that transactions are validated using a voting
algorithms over conventional blockchain-based consen-
process instead of the costly proof-of-work calculations
sus algorithms is their capacity to process transactions
used by conventional blockchain based systems. Nano
more quickly and flexibly. This is because DAG-based
has several other features that make it appropriate for use
algorithms do not contain blocks that must be added to
as a cryptocurrency, including fast transaction times, low UL‑BlockDAG

transaction fees, and energy efficiency. It has been used UL(Unsupervised Learning)-blockDAG is a distrib-
in various applications such as peer-to-peer payments, uted ledger system that utilizes a directed acyclic graph
online micropayments, and online gaming. It is impor- (DAG) as its primary data structure. It is an extension of
tant to note that Nano has faced some controversy and the blockDAG consensus algorithm, which merges and
criticism in the past, including concerns about the cen- arranges data to reconstruct a blockchain system based
tralization of its network and the security of its consensus on a single chain. The system’s key feature is its UL-scal-
algorithm. Nano team has continued to work on improv- ability, which enables it to handle many transactions per
ing the technology and addressing these issues. second (TPS) with a high throughput rate. In addition, it
uses a proof-of-workless consensus method, certifying
Jointgraph transactions through voting rather than the costly proof-
Jointgraph is a consensus algorithm that is based on of-work calculations used in traditional blockchain sys-
Byzantine fault tolerance. It uses events to pack transac- tems. UL-blockDAG also offers other features such as low
tions and sends them through a gossip protocol, which transaction fees, fast transaction times, and energy effi-
allows anyone to send events to a random node (Xiang ciency, making it ideal for use in distributed ledger tech-
et al. 2021). These events are validated by all members nologies and decentralized applications. UL-blockDAG
of the network, and Jointgraph uses a threshold of 2/3 has already been employed in various areas, including
of all members to reach consensus. To improve consen- supply chain management, smart cities, and energy mar-
sus efficiency, Jointgraph employs a supervisor node that kets. However, it is a relatively new and rapidly evolving
monitors member behavior and collects votes during the technology, and there is ongoing debate about its relative
consensus process to determine the finality of events. advantages and disadvantages compared to other distrib-
Every member of the network has a copy of Jointgraph, uted ledger technologies. (Reddy and Sharma 2020)
so it is possible to know each member would vote in the
consensus process. An example of Jointgraph consensus Dexon
is illustrated in Fig. 12, which shows three ordinary nodes DEXON is a distributed ledger technology that uses a
(A, B, and C) and one supervisory node (D). Red circles consensus algorithm based on proof of participation
represent events that have reached consensus, while light (PoP). In DEXON, every node is equally likely to propose
circles represent events that are not in consensus. Red a block, and the issuer of a block is determined using a
circles are events that are verified by at least three mem- Verifiable Random Function (VRF) (Chen et al. 2018).
bers, including the supervisory node. The confirmation This reduces communication costs and encourages more
time for events depends on the frequency of the gossip nodes to join the protocol. DEXON uses a block lattice
protocol. structure and proposes the use of a fast Byzantine agree-
ment that terminates in 6σ time, where σ is the upper
BlockDAG bound of the network’s gossip period. It generates on-
BlockDAG is a DAG-based consensus algorithm that chain, unpredictable randomness as it achieves consen-
uses sorting and merging to reconstruct a single-chain- sus, and once a DEXON Byzantine agreement confirms
based blockchain system (Gai et al. 2020). It has five a block, a committee of nodes generates a threshold
phases: Block Generation (BG), Sorting Block (SB), Block signature with an unpredictable threshold value. In this
Merging (BM), Consensus Implementation (CI), and consensus, no single block proposer can determine the
Block Splits (BS). In the first phase, BG generates origi- consensus timestamp of a proposed block, and DEXON
nal blocks by adding transactions to the block pool of the achieves second-level latency instead of traditional min-
nearest blockchain node, which then validates the trans- ute-level latency. DEXON is highly decentralized and
actions and bundles them into blocks that are added to robust in practical deployment environments.
the system block pool for further validation. In the sec-
ond phase, SB sorts all unvalidated blocks in the block Spectre
pool using a sorting algorithm to create a sequence in SPECTRE is a protocol for the consensus core of cryp-
the DAG structure. BlockDAG addresses issues of double tocurrencies that offers high throughput and fast con-
spending and consensus conflicts in the BM phase, where firmation times (Kovalchuck et al. 2022). It enables high
the merged blocks are finalized through global consensus block creation rates and uses partial synchronous net-
in the CI phase. In the BS phase, the merged blocks are works. SPECTRE generalizes Nakamoto’s blockchain
split into their original states and placed in the on-prem- into a block DAG, allowing miners to create blocks
ises BlockDAG structure (Gai et al. 2020). concurrently by maintaining a full DAG of blocks. It
conventional blockchain-based consensus algorithms.

DAG-based consensus algorithms, such as IOTA’s Tan-
gle, Nano’s Block Lattice, Jointgraph, BlockDAG, and
UL-BlockDAG, have the ability to process transactions
quickly and flexibly by adding them to the DAG in par-
allel, leading to a higher throughput. Unlike traditional
blockchain-based consensus algorithms, DAG-based
algorithms do not require miners to perform costly
proof-of-work calculations and do not contain blocks
that must be put on the chain in a specific order. The
passage cites examples of applications that have utilized
DAG-based consensus algorithms, including distrib-
uted ledger technologies, peer-to-peer networks, and
decentralized applications. However, there are ongo-
ing debates regarding the advantages and disadvantages
of DAG-based consensus algorithms compared to tra-
ditional blockchain-based consensus algorithms. The
passage provides details on IOTA’s Tangle and Nano’s
Block Lattice as DAG-based consensus algorithms,
highlighting their unique features and criticisms. Addi-
tionally, the passage briefly explains other DAG-based
consensus algorithms such as Jointgraph, BlockDAG,
Fig. 12 Jointgraph consensus algorithm and UL-BlockDAG.
Hybrid‑based consensus algorithms

uses a voting algorithm to determine the order between Hybrid consensus algorithms combine elements from
each pair of blocks in the DAG, with the votes coming various types of consensus mechanisms to achieve spe-
from blocks rather than miners. The votes are algo- cific properties or goals (Wu et al. 2020). These algo-
rithmically determined based on the location of the rithms can adapt to changing conditions or requirements
block within the DAG, and the majority vote becomes by using multiple mechanisms in parallel or sequentially.
irreversible very quickly, providing a consistent set of They can also strike a balance between decentralization
transactions. Tables 5 and 6 Show the analysis of DAG and efficiency, allowing for faster transaction processing
based consensus algorithms in terms of pros and cons, while remaining decentralized. Delegated Proof of Stake
energy efficiency, throughput, scalability and security. (DPoS), Hybrid Proof of Work/Proof of Stake (PoW/PoS),
The table summarizes different consensus algorithms and Byzantine Fault Tolerance are examples of hybrid
along with their references, pros, and cons. The con- consensus algorithms (BFT). These algorithms seek to
sensus algorithms discussed are Jointgraph, Block- combine the advantages of various consensus mecha-
Dag, UL-blockDAG, Dexon, and Spectre. Jointgraph is nisms to create a more flexible, adaptable, and secure
designed to be robust against double spending attacks system. One example of a hybrid consensus algorithm
and has improved throughput compared to other dis- is a proposed algorithm that improves the efficiency and
tributed ledger technologies. BlockDag has high scal- scalability of conversation interactions in multidomain
ability, robustness, and high throughput but may systems, where the number of conversations increases
increase latency when using a merge sort. UL-block- rapidly with the number of domains (Yang et al. 2022).
DAG has robustness, high transaction rate, and a high The proposed algorithm combines reputation-driven vot-
block creation rate but the complexity of the system ing, dynamic construction strategies, and an incentive
increases as the number of nodes increases. Dexon has scheme with both economic and non-economic rewards
low latency and high throughput, reduced cost, scal- to create a decentralized, scalable, and secure system for
ability, and usefulness in resource-constrained envi- experts to communicate across domains. The research-
ronments. Spectre is robust and has a high transaction ers built a master–slave chain model for a multidomain
rate and block creation rate but has a high latency and conversation system that can process multiple interac-
double spending risk. This section examined the use tions concurrently. They tested the proposed approach
of directed acyclic graphs (DAGs) as an alternative to on a local private blockchain network and showed that
Table 5 Pros and cons for DAG based consensus algorithms

Consensus algorithms References Pros Cons
Jointgraph (Xiang et al. 2021) The consensus algorithm is designed to be A central point of control in a network or system
robust against double spending attacks, which can be a target for attack
can help ensure the integrity of the network
and prevent fraud. JointGraph is also known
to have improved throughput compared to
other distributed ledger technologies, such
as hashgraph, which means it can handle
a higher volume of transactions in a given
amount of time
BlockDAG (Gai et al. 2020) High scalability, robustness, and high through‑ Parallel processing requires the assurance of
put transaction records only once, but this may
increase latency when using a merge sort. The
process of splitting transactions for parallel pro‑
cessing can also be vulnerable to attack
UL-BlockDAG (Reddy and Sharma 2020) Robustness, high transaction rate, and block As the number of nodes increases, the complex‑
creation rate is high ity of the system also increases
Dexon (Chen et al. 2018) low latency and high throughput include Under research
reduced cost, transaction ordering fairness,
scalability, unpredictable randomness, secure
transaction finality, and usefulness in resource-
constrained environments
Spectre (Kovalchuck et al. 2022) Robustness, high transaction rate, and block High latency, double spending risk
creation rate is high
Table 6 Pros and cons for DAG based consensus algorithms This section discussed hybrid consensus algorithms,
based on different criteria which combine elements from different types of con-
Algorithm Energy efficiency Throughput Scalability Security sensus mechanisms to achieve specific properties or
goals. These algorithms strike a balance between decen-
Jointgraph High High High High tralization and efficiency, allowing for faster transaction
BlockDAG Medium Medium Medium Medium processing while remaining decentralized. Examples of
UL-BlockDAG Medium Medium Medium High hybrid consensus algorithms include Delegated Proof
Dexon High High High High of Stake (DPoS), Hybrid Proof of Work/Proof of Stake
Spectre Low High High High (PoW/PoS), and Byzantine Fault Tolerance (BFT). The
passage also provides examples of two hybrid consensus
algorithms, one proposed algorithm that improves the
efficiency and scalability of conversation interactions in
it was feasible and effective in offering secure, decentral-
multidomain systems, and another is EOS, a well-known
ized, and scalable multidomain conversation interactions.
blockchain software system that uses a consensus mecha-
Another example is EOS, a well-known blockchain soft-
nism called BFT-DPoS.
ware system with the highest market value aside from
Bitcoin and Ethereum (Rahman and Mohsin 2020). It
Future improvements to consensus algorithms
uses a consensus mechanism called BFT-DPoS, which is
It is uncertain what the future holds for consensus algo-
a hybrid of delegated proof of stake and Byzantine fault
rithms, as the field is constantly changing, and new tech-
tolerance. In the consensus process, nodes are voted on
nologies are being introduced. However, some trends that
to determine decision makers through the DPoS algo-
may impact the future of consensus algorithms include
rithm, and then these decision makers communicate with
an emphasis on increased efficiency to reduce resource
each other to form the block sequence of the system. This
intensity and energy consumption, a focus on enhanc-
results in the continuous generation of six blocks every
ing security to protect against vulnerabilities and attacks,
0.5 s, minimizing the delay in block propagation, increas-
wider adoption in a variety of industries and applications,
ing the speed of block generation, and greatly increasing
and the need for scalability to handle a growing volume
the number of transactions. This allows EOS to support
of transactions.
a customer base of millions using blockchain technology.
Traditional consensus algorithm improvements coin system. Additionally, it offers a compensation sys-
Traditional consensus algorithms frequently have flaws tem to encourage trustworthy nodes to donate efficient
and limitations. Researchers have been working to refine processing power and move each node’s behavior closer
and improve the original algorithms to focus on these to a Nash equilibrium, where shifting a node’s approach
issues and improve them. While maintaining the benefits does not boost their own profitability. ThePoS algorithm’s
of the algorithms. The aim is to address and overcome security may be enhanced by this. The effectiveness of an
their weaknesses in order to expand the development and algorithm is largely determined by its efficiency. A good
evolution of consensus algorithms. The Proof-of-Work consensus algorithm should be efficient, have a low delay,
(PoW) algorithm has several shortcomings, including be secure, and be stable. While traditional consensus
slow consensus formation, low data throughput, and high algorithms can ensure the smooth operation of a block-
computing power utilization. There are numerous ways chain system, they often lack efficiency. For example,
to improve and enhance the PoW algorithm. The low the Proof of Work (PoW) consensus algorithm can take
data throughput of the blockchain can be improved by around ten minutes for each hash calculation and has a
raising the block size and decreasing the block creation confirmation delay of approximately one hour. This level
interval. Key blocks are used for leader elections but do of efficiency is not sufficient for practical use in cases
not contain transaction information, while micro-blocks where frequent computations are needed. The efficiency
are used to hold transaction information according to of an algorithm plays a significant role in the adoption
the Bitcoin-NG protocol (Eyal et al. 2016). As a result, and implementation of blockchain in real-world situa-
block generation can be done more quickly and effec- tions, and improving the efficiency of consensus algo-
tively while using less computational power. The Ethash rithms is an important area of development in the field.
algorithm addresses the issue of high computing power One example of an optimized algorithm is the Matrix
consumption by introducing I/O blocking and a directed Proof of Work (MPoW), which is based on the PoW algo-
acyclic graph to improve the target value solution in PoW. rithm and uses matrix calculations to reduce the block
It uses a small dataset to verify block information and a time (Zeng et al. 2019). Another example is the Proof of
large dataset generated from the small dataset for min- Trust (PoT) algorithm, which introduces a trust-proof
ing, and miners can only save the large dataset to mine mechanism to dynamically assign trust to nodes in the
more efficiently. The Ethash algorithm also uses dynamic blockchain, with higher trust leading to a higher prob-
adjustment to improve the production speed of data ability of accounting. This algorithm reduces network
blocks and reduce transaction times. It is designed to be delay and the time required for consensus, improving the
more suitable for general-purpose computers with large overall efficiency of the system. In the next section, we
memory capacities rather than requiring specific hard- are going to support our review with a table that charac-
ware. The Proof-of-Stake (PoS) algorithm suffers from terizes the different applications of consensus algorithms
centralization issues due to the exclusive accounting and which technologies utilize a certain consensus algo-
rights held by high-stakes nodes. Researchers have tried rithm. Table 7 depicts some examples of the different
to focus on these problems across various approaches. technologies and various domains in which the consen-
One of them involves imitating the Proof-of-Work (PoW) sus algorithms are being used.
algorithm by using virtual mining technology, which
requires only a small amount of computing resources for Conclusion
contributing nodes. This eliminates the competition for This topic has been mentioned many times in other
computing power and ensures randomness in the selec- papers, and many researchers have made a great effort,
tion of new blocks, while also reducing the risk of cen- like (Zheng et al. 2017) who explained the architecture of
tralization, avoiding the waste of computing power, and blockchain. The paper also reviewed some of the exist-
increasing fairness in mining. Another approach includes ing consensus algorithms used on different blockchain
mixing the PoS algorithm with the Byzantine fault-toler- platforms, such as proof-of-work (PoW), proof-of-stake
ant algorithm. By giving varying weights to votes based (PoS), practical Byzantine fault tolerance (PBFT), del-
on stake and requiring a weight of more than two-thirds egated proof-of-stake (DPoS), proof-of-elapsed-time
of the total weight to attain consensus, the Algorand and (PoET), and proof-of-authority (PoA). The paper com-
Ethereum protocols, also known as the BA protocol and pares these algorithms based on their performance, secu-
the Casper Friendly Finality Gadget protocol, employ rity, scalability, and energy efficiency. In addition to some
this approach. An illustration of this kind of develop- future perspectives.
ment is the Ouroboros algorithm employed in the ADA
In (Xu et al. 2019b) the authors provided a systematic Table 7 Different domains and applications with different
review of blockchain literature from various disciplines consensus algorithms
and perspectives. The article analyzes 41 papers pub- Domain Application Consensus algorithm
lished in Web of Science (WOS) from 2016 to 2019 that
cover different aspects of blockchain, such as its defini- DBMS Google Megastore PAXOS
tion, characteristics, classification, applications, chal- Apache cudu RAFT
lenges, and future directions. The article finds that there Cockroach DB RAFT
is no consensus on the definition of blockchain among Cryptocurrency Bitcoin PoW
researchers, but most of them agree that it has some Etherum PoW,PoS
key features such as decentralization, immutability, con- SlimCoin PoB
sensus mechanisms, cryptography, and smart contracts. BurstCoin PoC
The article also proposes a classification scheme for IOTA FPC
blockchain based on its architecture (public vs. private), EOS DPoS
governance (permissionless vs. permissioned), and func- neo DBFT
tionality (generic vs. specific). The article reviewed some Vet PoA
of the existing and potential applications of blockchain in NXT PoS
various domains such as finance, supply chain manage- Platforms for developments Microsoft Azure PoA
ment, healthcare, education, energy, government, and GoChain PoR (Reputation)
social media. The article identifies some of the benefits Hyperledger PBFT
and challenges of blockchain adoption in these domains, E-commerce ALgorand PoS
such as efficiency improvement, cost reduction, trust BitShare DPoS
enhancement, security enhancement, privacy protec- Health Care eHealth Estonia Pow, pBFT
tion, scalability issues, regulatory issues, interoperability Farma Trust Etherum(PoS, PoW)
issues, and user acceptance issues. In our work, unlike Supply Chain EverLEdger PoA,PoS
(Zheng et al. 2017) and (Xu et al. 2019b) we have focused
on the inclusivity of the review, so we have included most
of the aspects of the consensus algorithms, the old school, the chain. The application of a consensus algorithm also
and the new trends in the industry. We have also com- has a substantial impact on the system’s performance
pared the performance of every one of them. Moreover, and security. It is vital to assess the algorithm’s applica-
we have included the applications that are utilizing these bility for a particular use case and take into account a
consensus algorithms. This work is considered inclu- number of variables, including scalability, security and
sive and covers different aspects of the industry without energy consumption. Despite advancements in the field,
skipping the basics of the field. In conclusion, consen- scalability and security issues in decentralized systems
sus algorithms are an essential component of decentral- continue to pose significant challenges. To address these
ized systems and have several applications in distributed challenges continuing research is needed to enhance the
databases, distributed ledgers, and blockchain technol- scalability and security of existing consensus algorithms
ogy. Several well-known consensus algorithms, including and develop new mechanisms that can more effectively
Proof of Work, Proof of Stake, and Practical Byzantine address these issues. Moreover, the environmental
Fault Tolerance, have been featured in our review. These impact of consensus algorithms like energy consumption
algorithms are useful for various use cases since they and sustainability must also be taken into consideration.
each have distinctive characteristics and trade-offs. For In the future, there is a growing interest in researching
instance, Proof of Work is frequently used in blockchain new consensus algorithms for distributed ledger tech-
technology and offers a high level of security and decen- nologies, including those based on sharding, hybrid algo-
tralization, but it also has scalability problems and uses a rithms that combine multiple consensus mechanisms,
lot of energy. In contrast, Proof of Stake uses a different methods for reducing energy consumption in PoW
consensus technique to address the scalability and energy algorithms and addressing security issues in PoS algo-
consumption problems associated with Proof of Work. It rithms. In summary, consensus algorithms are critical for
also has its own set of drawbacks, including the Practi- the functioning of decentralized systems, and ongoing
cal Byzantine Fault Tolerance, while being a more mature research and development in this field is crucial for the
and widely adopted algorithm, is less commonly used in advancement and widespread adoption of decentralized
blockchain technology, due to its requirement of a high technologies. The selection of the appropriate consensus
number of confirmations before a block can be added to algorithm for a given use case is a critical decision that
can greatly impact the performance and security of the Do T, Nguyen T, Pham H (2019) Delegated proof of reputation: a novel
blockchain consensus. In: Proceedings of the 1st international electronics
system. Further research is needed to improve scalabil- communication conference, pp 90–98
ity and security of these algorithms and to develop new El Ioini N, Pahl C (2018) A Review of distributed ledger technologies: con‑
mechanisms that can more effectively address the chal- federated international conferences: CoopIS, C &TC, and ODBASE 2018,
Valletta, Malta, October 22–26, 2018. Proceedings, Part II:277–288. https://
lenges of decentralized systems. The consensus algorithm doi.org/10.1007/978-3-030-02671-4_16
is an active area of research with great potential, and the Eyal I, Gencer AE, Sirer EG, Van Renesse R (2016) Bitcoin-ng: a scalable block‑
future of distributed systems holds promise for new and chain protocol. In: 13th {USENIX} symposium on networked systems
design and implementation ( {NSDI} 16), pp 45–59
innovative consensus mechanisms. Fu X, Wang H, Shi P (2021) A survey of blockchain consensus algorithms:
mechanism, design and applications. Sci China Inf Sci 64:1–15
Acknowledgements
Fullmer D, Morse AS (2018) Analysis of difficulty control in bitcoin and proof-
The authors would like to thank the editor and anonymous referees.
of-work blockchains. In: 2018 IEEE conference on decision and control
(CDC). IEEE, pp 5988–5992
Authors’ contributions
Gai K, Hu Z, Zhu L, Wang R, Zhang Z (2020) Blockchain meets DAG: a BlockDAG
All the authors read and approved the final manuscript.
consensus mechanism. In: Algorithms and architectures for parallel pro‑
cessing: 20th international conference, ICA3PP 2020, New York City, NY,
Funding
USA, October 2–4, 2020, Proceedings, Part III, vol 20. Springer, pp 110–125
Not applicable.
Ganesh C, Orlandi C, Tschudi D (2019) Proof-of-stake protocols for privacy-
aware blockchains. In: Advances in cryptology—EUROCRYPT 2019:
Availability of data and materials
38th annual international conference on the theory and applications
Not applicable.
of cryptographic techniques, Darmstadt, Germany, May 19–23, 2019,
Proceedings, Part I. Springer, vol 38, pp 690–719
Declarations Gayoso Martinez V, Hernández-Álvarez L, Hernandez Encinas L (2020) Analysis
of the cryptographic tools for blockchain and bitcoin. Mathematics
Competing interests 8(1):131
The authors declare no competing interests. Gemeliarana IGAK, Sari RF (2018) Evaluation of proof of work (POW) block‑
chains security network on selfish mining. In: 2018 international seminar
on research of information technology and intelligent systems (ISRITI).
Received: 23 January 2023 Accepted: 31 May 2023 IEEE, pp 126–130
Gervais A, Karame GO, Wüst K, Glykantzis V, Ritzdorf H, Capkun S (2016) On the
security and performance of proof of work blockchains. In: Proceedings
of the 2016 ACM SIGSAC conference on computer and communications
security, pp 3–16
References Guo H, Yu X (2022) A survey on blockchain technology and its security. Block‑
Aggarwal S, Kumar N (2021) Cryptographic consensus mechanisms. In: chain Res Appl 3(2):100067
Advances in computers. Elsevier, vol 121, pp 211–226 Han T, Gao K (2020) Review of blockchain consensus algorithms. Sci J Intell
Ammous S (2016) Blockchain technology: what is it good for? SSRN 2832751 Syst Res 2(12)
Andola N, Venkatesan S, Verma S et al (2020) PoEWAL: a lightweight consensus Hu J, Liu K (2020) Raft consensus mechanism and the applications. J Phys Conf
mechanism for blockchain in IoT. Pervasive Mob Comput 69:101291 Ser 1544:012079
Andrey A, Petr C (2019) Review of existing consensus algorithms blockchain. Hu Q, Yan B, Han Y, Yu J (2021) An improved delegated proof of stake consen‑
In: 2019 international conference “quality management, transport and sus algorithm. Procedia Comput Sci 187:341–346
information security, information technologies” (IT &QM &IS). IEEE, pp Ismail L, Materwala H (2019) A review of blockchain architecture and consen‑
124–127 sus protocols: use cases, challenges, and solutions. Symmetry 11(10):1198
Bachani V, Bhattacharjya A (2022) Preferential delegated proof of stake Karantias K, Kiayias A, Zindros D (2020) Proof-of-burn. In: Financial cryptog‑
(PDPoS)-modified DPoS with two layers towards scalability and higher raphy and data security: 24th international conference, FC 2020, Kota
TPS. Symmetry 15(1):4 Kinabalu, Malaysia, February 10–14, 2020 revised selected papers 24.
Baldominos A, Saez Y (2019) Coin.AI: a proof-of-useful-work scheme for Springer, pp 523–540
blockchain-based distributed deep learning. Entropy 21(8):723 Kaur M, Khan MZ, Gupta S, Noorwali A, Chakraborty C, Pani SK (2021) MBCP:
Bamakan SMH, Motavali A, Bondarti AB (2020) A survey of blockchain performance analysis of large scale mainstream blockchain consensus
consensus algorithms performance evaluation criteria. Expert Syst Appl protocols. IEEE Access 9:80931–80944
154:113385 Kim D, Doh I, Chae K (2021) Improved raft algorithm exploiting federated
Benisi NZ, Aminian M, Javadi B (2020) Blockchain-based decentralized storage learning for private blockchain performance enhancement. In: 2021
networks: a survey. J Netw Comput Appl 162:102656 international conference on information networking (ICOIN). IEEE, pp
Bentov I, Lee C, Mizrahi A, Rosenfeld M (2014) Proof of activity: extending 828–832
bitcoin’s proof of work via proof of stake [extended abstract]. ACM SIG‑ Kovalchuk L, Oliynykov R, Bespalov Y, Rodinko M (2022) Comparative analysis
METRICS Perform Eval Rev 42(3):34–37 of consensus algorithms using a directed acyclic graph instead of a
Chen T-Y, Huang W-N, Kuo P-C, Chung H, Chao T-W (2018) DEXON: a highly blockchain, and the construction of security estimates of spectre proto‑
scalable, decentralized DAG-based consensus algorithm. arXiv preprint col against double spend attack. In: Information security technologies in
arXiv:1811.07525 the decentralized distributed networks. Springer, pp 203–224
De Prisco R, Lampson B, Lynch N (2000) Revisiting the Paxos algorithm. Theo‑ Lamberti R, Fries C, Lücking M, Manke R, Kannengießer N, Sturm B, Komarov
ret Comput Sci 243(1–2):35–91 MM, Stork W, Sunyaev A (2019) An open multimodal mobility platform
Denisova V (2019) Blockchain infrastructure and growth of global power based on distributed ledger technology. In: Internet of things, smart
consumption. Int J Energy Econ Policy spaces, and next generation networks and systems: 19th international
Dey S (2018) Securing majority-attack in blockchain using machine learning conference, NEW2AN 2019, and 12th conference, ruSMART 2019, St.
and algorithmic game theory: a proof of work. In: 2018 10th computer Petersburg, Russia, August 26–28, 2019, Proceedings 19. Springer, pp
science and electronic engineering (CEEC). IEEE, pp 7–10 41–52
Lamport L (2001) Paxos made simple. ACM SIGACT News (Distributed Com‑ Sayeed S, Marco-Gisbert H (2019) Assessing blockchain consensus and secu‑
puting Column) 32, 4 (Whole Number 121, December 2001), 51–58 rity mechanisms against the 51% attack. Appl Sci 9(9):1788
Lamport L, Shostak R, Pease M (2019) The byzantine generals problem. In: Sayeed S, Marco-Gisbert H (2020) Proof of adjourn (PoAj): a novel approach to
Concurrency: the works of Leslie Lamport, pp 203–226 mitigate blockchain attacks. Appl Sci 10(18):6607
Le Brun MA, Attard DP, Francalanza A (2021) Graft: general purpose raft con‑ Sheth H, Dattani J (2019) Overview of blockchain technology. Asian J Conver‑
sensus in elixir. In: Proceedings of the 20th ACM SIGPLAN international gence Technol AJCT. ISSN 2350-1146
workshop on Erlang, pp 2–14 Shrimali B, Patel HB (2022) Blockchain state-of-the-art: architecture, use cases,
Li J, Wu J, Chen L (2018) Block-secure: blockchain based scheme for secure P2P consensus, challenges and opportunities. J King Saud Univ Comput Inf
cloud storage. Inf Sci 465:219–231 Sci 34(9):6793–6807
Liskov B, Cowling J (2012) Viewstamped replication revisited Silvano WF, Marcelino R (2020) Iota tangle: a cryptocurrency to communicate
Liu H, Luo X, Liu H, Xia X (2021) Merkle tree: A fundamental component of internet-of-things data. Futur Gener Comput Syst 112:307–319
blockchains. In: 2021 international conference on electronic information Singhal B, Dhameja G, Panda PS, Singhal B, Dhameja G, Panda PS (2018) How
engineering and computer science (EIECS). IEEE, pp 556–561 blockchain works. Beginning blockchain: a beginner’s guide to building
Menon AA, Saranya T, Sureshbabu S, Mahesh A (2022) A comparatıve analysis blockchain solutions, pp 31–148
on three consensus algorithms: proof of burn, proof of elapsed time, Sunny J, Undralla N, Pillai VM (2020) Supply chain transparency through
proof of authority. In: Computer networks and inventive communication blockchain-based traceability: an overview with demonstration. Comput
technologies: proceedings of fourth ICCNCT 2021. Springer, pp 369–383 Ind Eng 150:106895
Merkle RC (1988) A digital signature based on a conventional encryption Tian S, Liu Y, Zhang Y, Zhao Y (2021) A byzantine fault-tolerant raft algorithm
function. In: Advances in cryptology—CRYPTO’87: proceedings. Springer, combined with Schnorr signature. In: 2021 15th international conference
vol 7, pp 369–378 on ubiquitous information management and communication (IMCOM).
Mohamed AA, Ibrahim AO (2020) Blockchain consensuses algorithms based IEEE, pp 1–5
on proof of work: a comparative analysis. Int J Comput Commun Netw Vashchuk O, Shuwar R (2018) Pros and cons of consensus algorithm proof
2(1):12–20 of stake. Difference in the network safety in proof of work and proof of
Morais R, Crocker P, de Sousa SM (2020) A tool for implementing privacy in stake. Electron Inf Technol 9(9):106–112
nano. In: 2020 IEEE international conference on decentralized applica‑ Velliangiri S, Karthikeyan P (2020) Blockchain technology: challenges and
tions and infrastructures (DAPPS). IEEE, pp 159–163 security issues in consensus algorithm. In: 2020 International conference
Nakamoto S (2008) Bitcoin: a peer-to-peer electronic cash system. Decentral‑ on computer communication and informatics (ICCCI). IEEE, pp 1–8
ized business review, 21260 Vilim M, Duwe H, Kumar R (2016) Approximate bitcoin mining. In: 2016 53nd
Nguyen CT, Hoang DT, Nguyen DN, Niyato D, Nguyen HT, Dutkiewicz E ACM/EDAC/IEEE design automation conference (DAC). IEEE, pp 1–6
(2019) Proof-of-stake consensus mechanisms for future blockchain Wang S, Tang X, Zhang Y, Chen J (2019) Auditable protocols for fair pay‑
networks: fundamentals, applications and opportunities. IEEE Access ment and physical asset delivery based on smart contracts. IEEE Access
7:85727–85745 7:109439–109453
Oki BM, Liskov BH (1988) Viewstamped replication: a new primary copy Wang H, Guo K (2019) Byzantine fault tolerant algorithm based on vote. In:
method to support highly-available distributed systems. In: Proceed‑ 2019 international conference on cyber-enabled distributed computing
ings of the seventh annual ACM symposium on principles of distributed and knowledge discovery (CyberC). IEEE, pp 190–196
computing, pp 8–17 Wang D, Jin C, Li H, Perkowski M (2020) Proof of activity consensus algorithm
Osadchuk M, Oliynykov R (2019) Method of proof of work consensus algo‑ based on credit reward mechanism. In: Web information systems and
rithms comparison. Radiotekhnika 198:105–112 applications: 17th international conference, WISA 2020, Guangzhou,
Park S, Kwon A, Fuchsbauer G, Gaži P, Alwen J, Pietrzak K (2018) Spacemint: China, September 23–25, 2020, Proceedings. Springer, vol 17, pp 618–628
a cryptocurrency based on proofs of space. In: Financial cryptography Wright A, De Filippi P (2015) Decentralized blockchain technology and the rise
and data security: 22nd international conference, FC 2018, Nieuwpoort, of lex cryptographia. SSRN 2580664
Curaçao, February 26–March 2, 2018, Revised Selected Papers 22. Wu Y, Song P, Wang F (2020) Hybrid consensus algorithm optimization: a
Springer, pp 480–499 mathematical method based on POS and PBFT and its application in
Perard D, Lacan J, Bachy Y, Detchart J (2018) Erasure code-based low storage blockchain. Math Probl Eng 2020
blockchain node. In: 2018 IEEE international conference on internet of Xiang F, Huaimin W, Peichang S, Xue O, Xunhui Z (2021) Jointgraph: a DAG-
things (iThings) and IEEE green computing and communications (Green‑ based efficient consensus algorithm for consortium blockchains. Softw
Com) and IEEE cyber, physical and social computing (CPSCom) and IEEE Pract Exp 51(10):1987–1999
smart data (SmartData). IEEE, pp 1622–1627 Xiao Y, Zhang N, Li J, Lou W, Hou YT (2019) Distributed consensus protocols
Puthal D, Mohanty SP (2018) Proof of authentication: IoT-friendly blockchains. and algorithms. Blockchain Distrib Syst Secur 25:40
IEEE Potentials 38(1):26–29 Xiong H, Chen M, Wu C, Zhao Y, Yi W (2022) Research on progress of block‑
Puthal D, Malik N, Mohanty SP, Kougianos E, Das G (2018) Everything you chain consensus algorithm: a review on recent progress of blockchain
wanted to know about the blockchain: its promise, components, pro‑ consensus algorithms. Future Internet 14(2):47
cesses, and problems. IEEE Consum Electron Mag 7(4):6–14 Xu G, Liu Y, Khan PW (2019) Improvement of the DPoS consensus mechanism
Rahman MU (2020) Scalable role-based access control using the EOS block‑ in blockchain based on vague sets. IEEE Trans Ind Inf 16(6):4252–4259
chain. arXiv preprint arXiv:2007.02163 Xu M, Chen X, Kou G (2019) A systematic review of blockchain. Financ Innov
Reddy S, Sharma G (2020) Ul-blockDAG: unsupervised learning based consen‑ 5(1):1–14
sus protocol for blockchain. In: 2020 IEEE 40th international conference Yang F, Zhou W, Wu Q, Long R, Xiong NN, Zhou M (2019) Delegated proof
on distributed computing systems (ICDCS). IEEE, pp 1243–1248 of stake with downgrade: a secure and efficient blockchain consensus
Saad SMS, Radzi RZRM (2020) Comparative review of the blockchain consen‑ algorithm with downgrade mechanism. IEEE Access 7:118541–118555
sus algorithm between proof of stake (POS) and delegated proof of stake Yang R, Wakefield R, Lyu S, Jayasuriya S, Han F, Yi X, Yang X, Amarasinghe G,
(dPOS). Int J Innov Comput 10(2) Chen S (2020) Public and private blockchain in construction business
Saad M, Qin Z, Ren K, Nyang D, Mohaisen D (2021) e-PoS: making process and information integration. Autom Constr 118:103276
proof-of-stake decentralized and fair. IEEE Trans Parallel Distrib Syst Yang W, Garg S, Huang Z, Kang B (2022) A hybrid consensus algorithm for
32(8):1961–1973 master-slave blockchain in a multidomain conversation system. Expert
Saez Y (2019) Coin.Ai: a proof-of-useful-work scheme for blockchain-based Syst Appl 204:117300
distributed deep learning. Entropy 21(8):723 Yang X, Chen Y, Chen X (2019) Effective scheme against 51% attack on proof-
Salimitari M, Chatterjee M, Fallah YP (2020) A survey on consensus methods of-work blockchain with history weighted information. In: 2019 IEEE inter‑
in blockchain for resource-constrained IoT networks. Internet Things national conference on blockchain (blockchain). IEEE, pp 261–265
11:100212
Yusoff J, Mohamad Z, Anuar M (2022) A review: consensus algorithms on

blockchain. J Comput Commun 10(9):37–50
Zeng L, Xin S, Xu A, Pang T, Yang T, Zheng M (2019) Seele’s new anti-Asic con‑
sensus algorithm with emphasis on matrix computation. arXiv preprint
arXiv:1905.04565
Zhang R, Xue R, Liu L (2019) Security and privacy on blockchain. ACM Comput
Surv CSUR 52(3):1–34
Zheng X, Feng W (2021) Research on practical byzantine fault tolerant consen‑
sus algorithm based on blockchain. J Phys Conf Ser 1802:032022
Zheng Z, Xie S, Dai H, Chen X, Wang H (2017) An overview of blockchain
technology: architecture, consensus, and future trends. In: 2017 IEEE
international congress on big data (BigData congress). IEEE, pp 557–564
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub‑
lished maps and institutional affiliations.

Evolution of Blockchain Consensus Algorithms: A Review On The Latest Milestones of Blockchain Consensus Algorithms

Uploaded by

Copyright:

Available Formats

Evolution of Blockchain Consensus Algorithms: A Review On The Latest Milestones of Blockchain Consensus Algorithms

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Evolution of Blockchain Consensus Algorithms: A Review On The Latest Milestones of Blockchain Consensus Algorithms

Uploaded by

Copyright:

Available Formats

Hussein et al.

Cybersecurity (2023) 6:30 Cybersecurity

REVIEW Open Access

Evolution of blockchain consensus

Introduction blockchain technology, signaling the emergence of a new

Working theory of blockchain

Classes and structure of blockchain

Data blocks, a linked list, Merkle trees, and other data

Paxos’ algorithm VR consensus algorithm

• A proposer node suggests a value for a specific data

(Kim et al. 2021). One key feature of RAFT is that it pro-

Fig. 6 RAFT architecture

Fig. 8 General phases of pBFT

secondary to primary, typically in the event of primary

• The client sends a request to the primary (leader)

Fig. 9 Mining process in Bitcoin network Fig. 10 Workflow of PoS

Table 2 Pros and Cons of PoW

is randomly selected to sign the new block (Andola

Delayed proof of work (dPoW)

as a cryptocurrency, including fast transaction times, low UL‑BlockDAG

conventional blockchain-based consensus algorithms.

Hybrid‑based consensus algorithms

Table 5 Pros and cons for DAG based consensus algorithms

Yusoff J, Mohamad Z, Anuar M (2022) A review: consensus algorithms on

You might also like