A Comprehensive Review of Blockchain Consensus Mechanisms

Received February 1, 2021, accepted March 9, 2021, date of publication March 12, 2021, date of current version March
24, 2021.
Digital Object Identifier 10.1109/ACCESS.2021.3065880
A Comprehensive Review of Blockchain

Consensus Mechanisms
BAHAREH LASHKARI 1, (Graduate Student Member, IEEE),
AND PETR MUSILEK 1,2 , (Senior Member, IEEE)
1 Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB T6G 2R3, Canada
2 Department of Applied Cybernetics, University of Hradec Králové, 500 03 Hradec Králové, Czech Republic
Corresponding author: Bahareh Lashkari (bahareh1@ualberta.ca)
The research is supported by the Government of Alberta under the Major Innovation Fund, project RCP-19-001-MIF.
ABSTRACT Since the advent of distributed ledger technologies, they have provided diverse opportunities
in a wide range of application domains. This article brings a comprehensive review of the fundamentals
of distributed ledger and its variants. Analyzing 185 publications, ranging from academic journals to
industry websites, it provides a comparative analysis of 130 consensus algorithms using a novel architectural
classification. The distribution of the reviewed algorithms is analyzed in terms of the proposed classification
and different application domains, along with the applicability of each class among the top 10 platforms in
the most prominent blockchain application domains. Additional conclusions are drawn from the evolution of
consensus mechanisms, and the analysis concludes envisaging future prospects for consensus as an important
part of distributed ledger technology.
INDEX TERMS Blockchain, distributed ledger technology, consensus mechanisms, cryptocurrency.
I. INTRODUCTION consensus mechanisms that not only allows the examination

The emergence of ledgers can be tracked back more than of the existing consensuses but also provides a structure that
thousands of years. It was followed by a conventional banking subsequent algorithms can be related to. The current studies
system where data records have been authenticated by a cen- in this area have either analyzed a confined number of algo-
tral authority. With the advent of computers, ledgers became rithms without any concerns regarding the features that are in
digitized and evolved the preceding centralized ledger bank- common among consensus algorithms, or they have studied
ing system, mirroring what was initially carried out on paper. the algorithms in particular categories.
A few years later, the distributed ledger technology has been Previously, there have been several other attempts to anal-
proposed by Satoshi Nakamoto with the intention of exclud- yse consensus mechanisms with the goal of establishing
ing the former authoritative environments towards a verifi- their taxonomy. Hattab [1] classifies the consensus mech-
able structure. Distributed ledger technology (DLT) enabled anisms into 3 groups based on hardware, stake, and vote.
a novel form of recording transactions using cryptography, In another study, performed by Chaudhry and Yousaf [2],
advanced algorithms, and massive computing capacity. As a the consensus mechanisms are categorized with respect to
digital database instance, DLT is shared between individuals their scalability, communication model, category, and failure
with certain characteristics that not only preserve particular models. Alsunaidi and Alhaidari [3] also classifies the con-
communication protocols but also go through an agreement sensus algorithms into proof-based and vote-based. Another
procedure that leads to a shared decision exclusive to the investigation, by Xiao et al. [4], classifies the algorithms
group of individuals that operate the DLT. into Namakato-based and variations, PoS-based, and other
This paper reviews the fundamentals of DLT that led emerging consensus protocols.
to the advent of blockchain and provides a comparative Unfortunately, none of the existing studies provides a
analysis of the 130 most recently introduced consensus comprehensive classification that incorporates all classes of
mechanisms. We propose an architectural classification of consensus mechanisms. This would require identification of
prevalent factors so that diversity of the individual approaches
The associate editor coordinating the review of this manuscript and can be clearly discerned. The proposed classification not
approving it for publication was Guangjie Han . only supports detailed analysis of the building blocks of each
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
43620 VOLUME 9, 2021
B. Lashkari, P. Musilek: Comprehensive Review of Blockchain Consensus Mechanisms
FIGURE 1. Publication outlets from industrial and academic perspective.
algorithm but also provides an extensive guide on the com- demonstrated their potency. Their dynamic nature can accel-
munication model and performance parameters important for erate transactions and reduce associated expenses by elimi-
algorithm evaluation. To facilitate the consensus selection nating the requirements for a central authority. DLT is also
procedure for future experiments, this study also examines referred to as a reliant record handling mechanism that tones
the applicability and reputation of each class of algorithms in down cyberattack vulnerabilities through the transparency of
different areas of blockchain applications. transactions. It is a digital database that is constructed, shared,
The contributions of this survey and the added value it validated, updated, and synchronized by participants to elim-
brings in comparison to other reviews in this area can be inate the single central authority. A distributed ledger is either
summarized as follows: public (open to all users) or private (shared among particular
• It provides a comparative review of blockchain as one of participants that are more likely to adopt specific protocols
the precedent forms of distributed ledger. during communication). Every participant in a distributed
• It develops a comprehensive classification of consensus ledger network is required to traverse an agreement phase
algorithms. that differs from one ledger to another and leads to a single
• It reviews an extensive set of 130 consensus algorithms decision. A distributed ledger is a secure approach to keep
and discerns the classes they are associated with. permanent records as it cannot be tampered retroactively.
• The algorithms in the same class are compared regarding This allows the contributors to communicate through DLT
their scalability, finality, adversary tolerance, accessibil- more confidently. The emergence of smart contracts in the
ity, agreement, incentives, centralization, and cost. context of distributed ledgers as a digital means for verifica-
• Finally, it analyses the distribution of each class in dif- tion and execution of contracts has significantly changed the
ferent blockchain application domains. prospect of DLTs. The integration of smart contracts into the
To provide a comprehensive analysis of consensus mecha- distributed ledger has enhanced the reliability, accountability,
nisms, we investigated a wide range of resources from aca- and transparency of transactional applications. Other than
demic journals, industrial websites/blogs, conferences and financial transactions, DLTs are applicable to various other
workshops, to white papers. A total of 185 publications have use cases.
been selected for consensus analysis. As illustrated in Fig.1, The Focus Group on Application of Distributed Ledger
there are two leading publication sources: academic journals Technology (FG-DLT) [5], classifies the DLT applications
(39%) that represent the data gathered from the academic in a horizontal and vertical domain. Vertical domain depicts
domain, and industrial websites (17%) that denote websites different sectors of economy and the horizontal domain rep-
and blogs from the blockchain industry. As further shown resents their corresponding use cases [6]. The following sub-
in Fig. 3, academic journals are also classified according to sections provide an overview of the characteristics of the
their publication date to help identify potential literature gaps. different types of DLT categorized in Fig. 2.
This article is organized as follows. Section II introduces
the distributed ledger technology and its variants. The body A. BLOCKCHAIN
of the survey, dedicated to the consensus mechanisms, is con- The notion behind Blockchain as a digital, distributed, and
tained in Section III and analyzed in Section IV. Final decentralized data structure is the development of transaction
Section V summarizes the main results and conclusions. blocks that store digital transactions without the need for a
central authority. Information concerning new transactions
II. DISTRIBUTED LEDGERS is appended to the chain after it has been encrypted and
The widespread reputation that distributed ledgers have now confirmed by the majority of the participating agents. Each
attained started with the advent of bitcoin cryptocurrency that block is then timestamped and cryptographically linked to
VOLUME 9, 2021 43621

FIGURE 2. Variants of distributed ledger technologies.
keys associated with transacting agents ensure the integrity

on top of validation procedures that hinder manipulation.
The building blocks of blockchain security are cryptographic
hashing functions which generate unique identifiers with reg-
ulated length independent of the input. Each hash is asso-
ciated as an identifier for a block and correlates with the
hash value of the former block. The hash function is further
utilized in a consensus mechanism for verification of ongoing
transactions.
3) SPEED
FIGURE 3. Academic journals annual distribution. Distributed ledger has overcome the decelerated transactions
associated with the classic banking system. Blockchain’s
the former blocks as a demonstration for the sequence of transaction speed depends on block size, transaction fee,
recorded transactions. As a distributed database, Blockchain and network congestion. Blockchain facilitates global trans-
comprises an expanding record of transactions accompanied actions by decreasing the block time, which represents the
by the chronological order of their occurrence. It keeps the required interval for appendance of a novel block. Moreover,
identity of the contributors anonymous by employing digital the transmission time decreases with the increase of block
signatures [7]. size, which improves the transaction speed.
1) IMMUTABILITY 4) CONSENSUS
One of the outstanding features of blockchain is immutability, Consensus mechanisms have been incorporated in
in which no one can modify the distributed ledger by any blockchains as a fault-tolerant mechanism for transaction
means. The blockchain remains irreversible since any trans- verification. The consensus is utilized to preserve agreement
action cannot be altered, deleted, or reversed unless more among the nodes in the network. When the network expands,
than 51% of the nodes agree with the modification. This the number of nodes increases and it is quite challenging
would require the attacker to gain control over more than half to achieve agreement. Public blockchain requires the par-
of the nodes which is highly improbable. However, although ticipation of users for verification and authentication of the
breaching the immutability of the blockchain is considered transactions. Since blockchain is a dynamic, self-regulating
improbable and complicated, it is possible in practice if a system, it requires the incorporation of a secure mechanism to
substantial amount of resources is available. The immutabil- ensure the authenticity of the transactions, having participants
ity concept is correlated with both data and the code of the reach agreement on a consensus. Various types of consensus
distributed ledger. Blockchain considers the immutability of mechanisms have been proposed that differ in terms of their
the data records uncontroversial, while data can be tampered underlying principles and applications.
and erroneous prior to its appendance to the chain. Although
consensus mechanisms are incorporated for verification of B. DAG/TANGLE
the data inputs, they are confined by what participants can Directed Acyclic Graph (DAG) is a variant of DLT that
faithfully consent to. On the other hand, the immutability of has been proposed as an alternative to blockchain. DAG’s
the code is also questioned by pointing out the fact that no co-operating nodes are capable of cross verification due
code is developed in an impeccable manner integrating all to their arrangement in a directed graph. Implementa-
operative requirements. This concern is supported by the fact tion of DAG enhances the scalability of the network and
that, in many cases, the blockchain code has been constantly reduces the transaction fees as it supports fee-less nano-
adapted. transactions. DAG significantly improves the transaction val-
idation speed without incentivizing the participants. Since
2) SECURITY DAG reaches consensus without the implementation of the
Distributed ledgers are recognized for their exceptional classic hash-protected PoW (PoW), it requires neither min-
security measures. Participating agents utilize cryptography ers nor the underlying energy intensive infrastructure. How-
encryption to compose transactions. The public and private ever, DAG cannot reach a secure decentralized consensus to
43622 VOLUME 9, 2021

preserve the security level provided by its counterparts like incorporates a gossip protocol to transmit transaction infor-
blockchain. Significant features provided by DAG are listed mation through the network. Each node in the hashgraph
below: selects a random neighbor for information transmission,
aggregates all acquired information and passes it to another
1) SCALABILITY randomly chosen node. In a short time, all nodes know about
DAG is known for its virtually infinite scalability. Unlike the transactions and, using a virtual voting mechanism, each
other distributed ledgers, DAG enhances the scalability with node can validate and append them to the ledger [11].
the expansion of the network. It requires each node to verify
at least two former transactions to proceed with the con- 1) FAIRNESS
firmation of their corresponding transactions. Correspond- Hashgraph allows all contributing nodes to develop signed
ingly, the hashing power required for the validation procedure transactions that are later shared across the network. Since
decreases. hashgraph employs an enhanced level of fairness, nodes are
less likely to be affected by influencers once they agree on a
2) COMPATIBILITY transaction.
Employing microtransactions is avoided in blockchain as
2) DATA STRUCTURE
it increases the transaction fees. On the contrary, DAG,
as a decentralized channel, enables the participants to make Hashgraph can be considered a distributed database due to
instant micro- or even nano-transactions by incorporating its atomicity, consistency, isolation, and durability properties.
transaction fee-free schemes. This feature makes DAG more This is often referred to as ACID compliance. Once the
compatible with microtransactions. distributed ledger reaches a consensus on the sequence that
each transaction has occurred at, the consensus order will be
3) RESILIENCE
shared among the nodes’ local copy of the database. As the
database preserves ACID properties, each node of the ledger
One of the most significant features of DAG is quantum resis-
as the community can also retain the same features.
tance that makes the underlying distributed ledger less sus-
ceptible to quantum computers with higher-level computing 3) INFORMATION SHARING
properties using Winternitz one-time signature scheme [8].
As mentioned earlier, hashgraph incorporates a gossip proto-
col to broadcast transaction information across the network.
4) VALIDATION
To ensure all nodes are informed about the changes in the net-
The quantum resistance of DAG results in masked authen- work, the hasgraph ledger keeps a record of each gossip from
ticated messaging and parallelly lined transactions, which is initiation to termination. This assists hashgraph in inspecting
an excellent approach for information transformation through if all nodes have been engaged in the process and are aware
encryption and authentication. of the transactions.
C. DAG-BASED BLOCKCHAIN 4) VOTING

Ever since blockchain has been proposed, many investiga- Hashgraph incorporates virtual voting for validation of the
tions have been performed to address the deficiencies of transaction by relying on the agreement of at least 2/3 of
the blockchain, such as its limited throughout. As the next the network on a particular transaction. It also considers the
generation of the blockchain, DAG-based blockchain has number of famous witnesses that have contributed to the
been proposed. It inherits the significant features of both voting procedure. Famous witness stands for a transaction
DAG and blockchain, as discussed earlier. The proposed that has occurred in sequential order and receives the wit-
distributed ledger envelopes transactions in the form of blocks ness’s vote. The name comes from the fact that most nodes
that are structured as DAG. It adopts a verification proce- are aware of its occurrence comparatively faster through the
dure that requires every novel transaction to be validated gossip protocol. This accelerates reaching consensus in the
by at least two earlier transactions to be appended to the same order as the transactions within the network.
blockchain. This procedure is inherited from DAG as it elim-
inates the participation of miners for the authentication of E. TEMPO
transactions, which speeds up the whole process. Moreover, Tempo is another variation of distributed ledger that incorpo-
it employs the gossip protocol for communication among rates the same principle of retaining the order of information.
nodes, which enhances the linear structure of traditional In addition, it associates timestamps and subset of the ledger
blockchain [9], [10]. to the users. Once a node begins to verify a transaction,
it is not capable of maintaining a consensus using traditional
D. HASHGRAPH timestamps as the timestamp changes from one sub-ledger
Hashgraph emerged as a consensus-oriented scheme using to another. Hence, it proceeds by using the logical clock,
the PoS algorithm. It is capable of storing several transactions which compares whether the former transaction matches its
in parallel by associating uniform timestamps. Hashgraph recorded sequence.
VOLUME 9, 2021 43623

1) STRUCTURE generated by the sidechain network and its underlying plat-

Tempo is a variation of a distributed ledger that scales hori- form, Monax [13].
zontally. The data structure and storing mechanism of tempo
rely on sub-ledgers perceived as shards. Diverse fragments G. HOLOCHAIN
of shards with distinctive identification keys are the building Holochain is another variation of distributed ledger, advanced
blocks of a universal distributed ledger. Tempo ensures that in terms of incorporating an agent-centric structure rather
all cooperating shards are storing the transaction information than a data-centric scheme. This type of distributed ledger
in the right order. associates a distinctive forking approach with each agent
to eliminate the use of a global consensus mechanism. The
2) COMMUNICATION motivation behind this approach is to significantly enhance
the scalability of the underlying network [11].
As mentioned earlier, tempo is required to ensure that coop-
erating shards in the distributed ledger are able to store the
1) VALIDATION STRUCTURE
transaction information in the right sequence. Accordingly,
One of the significant properties of holochain is that individ-
it adopts the hashgraphs random gossip communication pro-
ual modules constitute the distributed ledger, which results in
tocol that allows all agents to communicate and propagate
the agent-centric structure. Accordingly, each agent is only
shared pieces of information. Hence, all shards on the ledger
in charge of its personal ledger and is not required to execute
are able to synchronize their information.
consensus for validation of each transaction. This not only
accelerates the verification procedure but also enhances the
3) VERIFICATION energy efficiency in consecutive runs.
One of the most significant features of tempo is associat-
ing integer values to agents that increase with the number 2) EXPLICIT PARTICIPATION
of unprecedented transactions that the agent observes. This Holochain, as its name suggests, implements a holographic
integer number represents the overall number of unique trans- methodology that allows developers to build decentralized
actions that each agent stores and is referred to as a logical applications that inherit distributed ledger features, such as
clock. Once the information regarding a transaction is stored scalability. One of its significant attributes is the explicit
by the agent, the subsequent logical clock is stored alongside participation of the user in the distributed ledger. This endows
to facilitate the verification of upcoming transactions. each user with explicit access to and control over data.
F. SIDECHAIN 3) SECURITY AND DECENTRALIZATION

Sidechain is another variation of the disturbed ledger that Stemming from the properties mentioned above, holochain
aims to address the shortcomings of traditional blockchain provides a very high level of security using the
concerning security, privacy, and performance. The structure DNA scheme [14]. Once a malicious action takes place in
of sidechain comprises a combination of two blockchains terms of broadcasting invalid information, the attacker is
that control access requests through a central consortium required to provide the DNA of the system. Each agent in
and manage local transactions as a permissioned blockchain. the distributed ledger is required to validate the DNA of the
Sidechain fragments the network, which allows each subnet sender with its own. Holochain is also capable of handling
to verify its corresponding transactions and eliminates the unlimited transactions due to its scalability that results from
scalability challenges caused by acquiring the consensus of the agent-centric procedure employed for storing the ledger.
all network nodes [12].
III. CONSENSUS
The notion behind the consensus mechanism is performing
1) PRIVACY AND SECURITY
frequently secure updates on the distributed ledger. One of the
In addition to the significant features of blockchain that have essential techniques is state machine replication that ensures
been employed by sidechain, it is capable of preserving the the presence and execution of shared states concerning the
privacy of transactions by keeping specific data confidential predefined state transition regulations. Since the state is
from competitor agents in the network. Moreover, it asso- shared among several replicas within the network, the exe-
ciates each sidechain with a validator node that communicates cution of the state will eventually result in identical outputs.
with the consortium network and verifies local transactions. Hence, replicas are required to interact and reach agreement
on potential modifications of the state using a consensus
2) PERFORMANCE scheme. Consensus helps replicas to decide on the finality
The structure of sidechain comprises several linked lists that of each state. However, the implementation of consensus in
follow the principle of the blockchain in block generation and a distributed system is complicated as it requires a consensus
reaching consensus. As a private ledger, it has been able to mechanism to maintain adversity tolerance, failure resilience,
decrease the adversity of PoW and enhance the performance partitioning throughout the network, delay perseverance, and
of the network. However, sidechain suffers from overhead other important properties. Moreover, it involves security
43624 VOLUME 9, 2021

FIGURE 4. Classification of consensus mechanisms.
FIGURE 5. Venn diagram for different classes of consensus.
measures such as the management of malicious nodes by consensus mechanisms guided by the proposed classification
adopting regulations like synchrony or message broadcast. and executed using the procedure depicted in Fig. 6.
The significance of consensus in distributed ledgers, includ- Accordingly, the first class belongs to the traditional con-
ing blockchain, is the preservation of three critical properties sensus approach paxos that took primary steps towards a fault
that ensure the efficiency of the underlying network. Other tolerance mechanism in the presence of unreliable system
than maintaining an agreement on a consistent global state for provisioning. The notion behind this classification can be
a distributed ledger, consensus ensures the safety, liveness, described as follows. If a consensus mechanism is not a primi-
and fault tolerance of the network. Accordingly, consensus tive consensus such as paxos, then it is either proof compliant,
protocols can be evaluated based on these properties. The Byzantine fault tolerance (BFT) compliant, primitive com-
consensus mechanism can preserve safety if it ensures all pliant, or cross-complaint. Primitive compliant mechanisms
nodes will contribute to an identical, consistent, and valid represent all variations of paxos that have been proposed as
output. Liveness is referred to as the ability of a consen- an extension to this consensus adopting the protocol’s main
sus mechanism to direct the contribution of nonfaulty nodes characteristics. Proof protocols and blockchains have been
towards the production of value. The consensus mechanism intertwined since the advent of distributed ledgers as they
needs to be capable of recovering from potential failures have demonstrated to be effective protocols that improve the
of contributing nodes to maintain a certain level of fault audibility and accountability of decentralized networks and
tolerance [15]. preserve their privacy. Moreover, they have been incorporated
Over the last two decades, there have been introduced in the configuration of enhanced proof protocols (proof com-
various consensus mechanisms. In this contribution, we intro- pliant) or other cryptographic primitives – these are referred
duce a novel classification of the most important consensus to as cross-compliant in the proposed taxonomy. Proof com-
mechanisms and their variations. The proposed taxonomy is pliances are either extensions to the proof protocols that
illustrated in Fig. 4 and a more detailed view of the categoriza- inherit the main characteristics of the protocol and attempt
tion is shown in Fig. 5. We present a comprehensive review to enhance the algorithm, or alternatives that have been intro-
of the functionalities, shortcomings, and advantages of the duced as novel proof mechanisms. These algorithms address
VOLUME 9, 2021 43625

FIGURE 6. Flow chart of Consensus classification procedure.
43626 VOLUME 9, 2021

the deficiencies of a particular proof mechanism by incor- This process eliminates the proportional incentive mecha-
porating some outstanding features of other proof consensus nism of the former PoS. PoC encourages the participants to
mechanisms leading to a hybrid proof compliant consensus. actively execute their CLAM, which results in improvement
One of the most common faults recognized in distributed of network security. The notion behind the distribution of
ledgers is caused by the erratic behavior of participating CLAMS is to engage as many contributors as possible to not
nodes, which is known as Byzantine fault. This was first only facilitate the authenticity verification process but also to
pointed out by Lamport as the Byzantine general prob- make the network widely spread and difficult to track [17].
lem, which happens as a consequence of a compromised
node where a Byzantine node contributes to ambiguous 2) PROOF OF STAKE TIME
responses or misleads other agents. In response, the BFT Proof of stake time (PoST) is a time-accepted nonlinear con-
algorithm and its variants have been proposed. These are sensus mechanism that has been proposed as an alternative to
categorized as BFT-compliant algorithms. An algorithm from address the deficiencies of PoS. PoST incorporates a periodic
this category is either proposed as an extension to existing time acceptance function that correlates with retained coins
BFT approaches, or as an alternative that integrates the char- and strives to enhance the security and distribution of the
acteristics of BFT protocols and builds upon them to address network. The contribution of volunteers is determined using
the corresponding deficiencies. Otherwise, it is considered an interest rate that maintains inverse proportion to network
an alternative that integrates differing BFT protocols and strength. PoST defines a quantified idle-time attribute to rep-
contributes to a hybrid BFT-compliant alternative. resent a fraction of age that does not reinforce the distribution
Pure alternatives are associated with protocols that present of the consensus anymore. This parameter affects the fraction
a novel consensus without preserving the features of previ- of acquired interest and eliminates the probability of meeting
ously proposed mechanisms trying to address the existing the proof. Hence, to enhance the interest rate, the participating
shortcomings. node is required to stake constantly to pass all corresponding
nodes through stake-time window [18], [28].
A. PRIMITVE CONSENSUS MECHANISM
3) PROOF OF WORK TIME
1) PAXOS
Discovering the target nonce in PoW, a waste of computa-
As the very first proposed consensus algorithm, paxos facili-
tional power seems inevitable as a consequence of regulated
tates the selection of a single value beneath the crash or faulty
intervals for block creation. As a PoW alternative, PoWT con-
circumstances of the network. Paxos classifies the nodes
sensus incorporates a block time attribute to enhance not only
into proposers, acceptors, and learners. Proposers provide a
the mining power but also the transactions of the blockchain.
message indicating a proposal number and forward it to the
It conforms the speed of the transactions to mining power
acceptor. The proposal number is considered as a time line
and facilitates auto-adjusting towards profitable mining. This
throughout the process, in which the proposal with a higher
scheme proposes a variable block creation rate that correlates
number is the most recent update. Acceptor compares the
with increments of mining power that simply eliminates the
acquired proposal number with the current known value, and
waste of required computation power [29].
only accepts the proposal if it is more recent. Afterward,
the acceptor forwards a response message indicating whether 4) PROOF OF SPACE TIME
the proposal has been accepted or rejected, corresponding
Proof of space-time (PoST) is another implementation of
proposal number, and all accepted value. The proposer is
the proof of storage in which the server can publish a
required to investigate whether the majority of acceptors have
proof of storage where verifiers can refer to and investi-
rejected the proposal or not. In case of rejection, the proposer
gate whether the transmitted data was being stored during
updates the proposal number with the most recent value.
a particular period of time. Proof of space time elimi-
Otherwise, the acceptor broadcasts the accepted value to
nates the submission of proofs to the blockchain that corre-
all learners on the network. To reach consensus in paxos,
spondingly prevents frequent interactions of the verifier and
the proposer should receive at least N /2 − 1 acceptances
prover. Hence, miners are entrusted to store the user’s data
(N is the number of proposals) from the acceptors [16].
in exchange for a collateral deposit. Afterward, the miner
will store data within the duration indicated in the settlement
B. PROOF COMPLIANT HYBRID ALTERNATIVE and submits the corresponding PoST to the network as an
CONSENSUS MECHANISM evidence [20], [30], [31].
1) PROOF OF WORKING STAKE/PROOF OF CHAIN
Proof of chain has been initially employed by CLAIM Coin 5) PROOF OF DEVOTION
as an alternative to PoS. PoC is a derivative of PoS that Proof of devotion (PoD) is an integration of PoS and PoI
enhances network security with distribution and transparency. in which nodes with the greatest impact on the network’s
This approach incentivizes staking users by choosing an ecology are given permission for block generation. Block
active client within 1-minute intervals. The client verifies all proposer will be chosen using participants from a divi-
pending transactions associated with the CLAMS network. sion of validator sets, which are referred to as dynasties
VOLUME 9, 2021 43627

TABLE 1. Primitve consensus mechanism.
TABLE 2. Proof-Compliant hybrid consensus mechanism.
in a round of BFT voting. This process facilitates deter- employs a rotating committee that selects miners with respect
mining the legitimacy of the proposed block. Moreover, to their PoW and PoS power and capabilities. The pro-
in order to eliminate the titled probability that may cause a posed fork-free mechanism is a generalized PoW variant that
monopoly, PoD grants bookkeeping titles to the designated enhances the evaluation procedure of the hash function and
nodes [21], [24], [32]. contributes to the exclusion of selfish mining [25].
6) PROOF OF ACTIVITY 10) PROOF OF STAKE VELOCITY

Proof of activity has been proposed as an alternative for Considering the major drawbacks of PoW and PoS, proof of
PoS that integrates both PoS and PoW to incentivize the Stake Velocity (PoSV) has been proposed as an underlying
participating agents instead of penalizing the passive agents. consensus algorithm for Reddcoin to facilitate secure peer-
Proof of activity exploits the hash of the most recent block to to-peer (P2P) transactions. PoSV integrates both stake and
select pseudorandom stakeholders for validation of a recently velocity to decrease the mining waste and prevent multipool
mined block template. Each block will be appended to the threats by incorporating novel coinage functions. In compar-
blockchain after being validated and having its hash signed ison with PoS protocols that consider coin age linear, PoSV
by the stakeholders [22], [33]. exploits a monotonic decay function for coinage estimation
that significantly changes the incentives and results in the
7) SNOW WHITE exceeding rate of coinage accumulation for the most recent
Snow white proposes a consensus mechanism known as coins. The exponential decay function imposes an asymptotic
sleepy. It incorporates the same procedure as the proof of restriction on coinage for security means that makes the
activity for the election of the committee nodes that are execution of 51% attacks extremely difficult [26].
in charge of voting on the block generation leader. Snow
White ensures to select a qualified leader concerning the hash 11) PROOF OF DDoS
function. Snow White differs from PoW in terms of feeding Proof of DDos is the underlying consensus mechanism for
the hash function with timestamps, which is an alternative for DDosCoin and operates by incentivizing the participants that
the arbitrary nonce. Finally, Snow white can bear successive attack victimized servers by transmitting a large amount of
committee reconfigurations and fraud nodes [23], [34]. network traffic. This approach is a PoW alternative, also
known as malicious PoW. It requires numerous TLS con-
8) HYBRID PROOF OF WORK nections between the miners and the target server. The target
Hybrid proof of work, which was initially introduced by server is fairly selected running a PoS consensus that ensures
Lynx, is a PoW alternative that maximizes the contribution the participants hold a substantial stake from mining activ-
of miners with limited computational resources by elimi- ities. Proof of DDos can also be incorporated to measure
nating profit incentives. HPoW does not necessarily grant the bandwidth utilization or resource consumption of the
the reward to the fastest node since it randomly selects the target [27], [37].
candidate blocks without imposing any requirements on the
hashing power or speed. Moreover, it does not allow a sin- C. BFT COMPLIANT
gle miner to successively win a block in 30 minute time 1) LOOP FAULT TOLERANT
intervals [24], [35], [36]. ICON is an interchain proposing an underlying ecosystem
to connect a wide range of blockchains preserving their
9) FLEXIBLE PROOF OF ACTIVITY consensus algorithms. This platform enables data sharing
Flexible PoA is a hybrid consensus mechanism that employs mechanisms among different associations such as universi-
PBFT to address forking and provide integrity. Flexible PoA ties, banks, or any other private blockchains without imposing
43628 VOLUME 9, 2021

TABLE 3. BFT-Compliant hybrid consensus mechanism.
any requirements for intermediaries. ICON is a variation of different stages of the consensus mechanism. To decide on the
tendermint [41] that reaches consensus using an enhanced contributing agents, DPoS is applied to the network. Selected
BFT known as loop fault tolerance. LFT speeds up the nodes contribute to the transaction verification procedure
consensus procedure and eliminates forking using a group and append a block that comprises a sequence of authentic
of trusted nodes that are allowed to regulate the number operations into the blockchain. BFT is in charge of updating
of required votes. LFT incorporated spinning to facilitate the ledger and confronting potential security threats. The
the complications of selecting the primary node. Moreover, participation of agents in each round requires transmission
it reaches consensus based on message relay and allows a of the hash of the most recent block to its corresponding
limited number of nodes to generate a block while the remain- generator within a particular interval. Failing to do so would
ing nodes participate in voting. This way, the communication eliminate the node from participation in the corresponding
overhead is eliminated due to the integration of messages round. Moreover, nodes that provide the correct hash value
from the network. LFT has successfully decreased the BFT’s are stored as authenticate nodes within a chronologically
traditional 3 stage execution procedure to 2.5 stages in which ordered list [43], [50].
a certain number of nodes are associated with a block gener-
ator broadcast at the same time that the rest of the network 2) CASPER
contributes to the voting procedure [38], [42]. Casper is an alternative to PoS consensus algorithm that
2) PRACTICAL BYZANTINE FAULT TOLERANCE
integrates the BFT mechanism. This approach incorporates
dynamic validator sets along with a correct-by-construction
PBFT has been proposed as a promising solution for Byzan-
forking scheme in which validators are required to vote and
tine faults. Several approaches have accordingly adopted
broadcast their signed votes throughout the network. Casper
PBFT as their underlying consensus mechanism. PBFT pro-
typically adds PoS on top of PoW as a supplementary layer
ceeds by determining a novel block during each round to clas-
to ensure finality. This also enhances the network’s modular
sify transactions concerning their sequence. Each node will
overlay. Casper aims to improve the security issues associated
be passed through 3 different phases known as pre-prepared,
with PoS like long-range revision attacks. However, the eval-
prepare and commit, if it has been verified by at least 2/3 of
uation results show that Casper is not able to tackle 51% of
the co-operating nodes in the blockchain network. Other than
attacks [44].
blockchain, hyperledger is one of the DLT solutions that
incorporates PBFT since it can address over 1/3 malicious
3) BFT-RAFT
replicas. As a permissioned and network-intensive consensus
mechanism, PBFT ensures the security of the ongoing trans- BFT-Raft has been proposed as an alternative to the classical
actions among acknowledged participants, but it is not able to raft. It integrates the features of both BFT and raft, includ-
scale to large networks. This makes it a perfect fit for private ing security and fault tolerance characteristics. To ensure
blockchains [22], [39]. the authenticity of the messages, BFT-raft exploits digital
signatures to preserve the integrity of messages and elimi-
3) hBFT nate forging. Messages that convey invalid signatures can be
hBFT is a hybrid, leaderless BFT variation that incorporates easily acknowledged and outcasted since they are signed by
PBFT’s checkpoint mechanism which enables the detection both nodes and users. BFT-raft elects leader nodes through
of potential inconsistencies in replicas during the message the voting procedure and is capable of sustaining networks
exchange phase. hBFT narrows down the required number functionality along f Byzantine failures in the presence of at
of cryptographic procedures and implements speculation to least 3f + 1 nodes in the network [45], [51].
increase performance and resilience while reducing the oper-
ation costs. Although hBFT is capable of tolerating any num- 4) PeerCensus
ber of faulty clients, it cannot refrain from its significant effect PeerCensus is a hybrid consensus mechanism that aims to
on performance [40]. dissociate the block creation procedure from transaction ver-
ification. It inherits the functionalities of both PoW and BFT
D. CROSS COMPLIANT HYBRID ALTERNATIVE to grant voting and block generation rights. PoW mechanism
CONSENSUS MECHANISM is employed for the election of the block leader while block
1) COMBINED DELEGATED PROOF OF STAKE AND validation proceeds through BFT’s pre-prepare, prepare, and
BYZANTINE FAULT TOLERANCE commit stages. Miners committee is arranged concerning the
The integration of DPoS and BFT, which was initially intro- miners whose preceding mined blocks reach a certain level
duced by Credits [49], incorporates both algorithms for of depth in the chain. Hence, the transaction that has been
VOLUME 9, 2021 43629

TABLE 4. Cross-Compliant hybrid consensus mechanism.
approved through BFT does not require any further mining comprises a nonce that is frequently modified by miners to
process [34], [46]. obtain different hash values. Hence, the consensus necessi-
tates the earned value to remain within a particular bound.
5) VERIFIABLE BYZENTINE FAULT TOLERANCE To maintain an agreement across the network concerning
VBFT is an alternative consensus mechanism for BFT that the propagation of new blocks, PoW imposes a complicated
has been initially adopted by the Ontology Consensus Engine puzzle that needs to be resolved by cooperating nodes. Miners
(OCG). It integrates PoS, BFT, and the Verifiable Random that overcome the puzzle will be granted access to append a
Function(VRF). The OCG’s consensus network comprises new block. The puzzle preserves an adjusted difficulty and
consensus nodes in charge, preserving the blockchains bal- is confronted by estimating the nonce’s value. This value is
ance. Incorporation of VRF facilitates the consensus popula- incorporated with the block’s header information to feed the
tion generation, as it provides randomness in the selection of SHA-256 hash function. The hash function will then convert
nodes, whether they are proposers or verifiers. This not only all inputs to generate the hash value. If the output of the
enhances the resilience but also accelerates the finality of the hash function holds a value beneath an appointed threshold,
consensus mechanism [47], [52], [53]. the estimated nonce will be accepted, and the miner is allowed
to append a block into the blockchain. Hence, whenever the
6) DELEGATED PROOF OF STAKE WITH DOWNGRADES objective value is acquired by a miner, it will accordingly
Delegated proof of stake with downgrades (DDPoS) is a con- broadcast the block throughout the network, and every single
sensus mechanism inspired by PoW and DPoS. DDPoS min- node within the entire network will be asked to confirm the
imizes the implications of stakes and computing resources authenticity of the hash value and append the corresponding
on generating blocks. Using the downgrading mechanism, block to their blockchain [55].
DDPoS is able to replace malicious nodes and enhance the
security of the system. The performance evaluation results of 3) PROOF OF LUCK
DDPoS indicate that it outperforms PoW and PoS in terms Proof of luck has been originally built upon TEE [96] and
of efficiency. However, its performance is still slightly lower XGS [97]. With the proof of luck consensus algorithm,
than DPoS [48]. the procedure of appending a new block to the miner’s chain
requires allocating a luck assess to each block. The luck
E. PURE ALTERNATIVE CONSENSUS MECHANISM assess is a random number that ranges from 0 to 1 within
1) ZAB a uniform distribution, and the chain that retains the overall
Zab has been proposed as an atomic broadcast protocol that largest assess is consented to be considered as the main chain.
incorporates primary ordering, which is significant in the The chain that maintains the highest luck is preferred by
context of primary back-up systems. Primary ordering is the blockchain. These characteristics make the proof of luck
responsible to ensure the validation of state alteration orders resistant to the double-spending attack. However, the proof of
over time and course of transactions. The validation proce- luck suffers from power deficiency since it decides on the luck
dure confirms the incremental order of each state change access after several examinations. Moreover, unsynchronized
with respect to the preceding one. This approach maintains an timepiece between the node and the network may eliminate
implicit interdependence on the sequence of state alterations. the chances of the node for being lucky, which marks the sig-
For the identification of misplaced or missing alterations, zab nificance of executing the proof of luck after synchronization
exploits a transaction identification method on state alter- of miners [56].
ations to conduct an initial determination on which order
of transactions can be employed to retrieve the application 4) PROOF OF BURN
state. Zab was originally implemented on the ZooKeeper [95] Proof of burn has been proposed as an energy-efficient, sus-
and has proven to be adequate for web-scale applications. tainable alternative to PoW where miners use an irretrievable
It outperforms paxos, performing thousands of broadcasts address to convey coins and burn them. The irretrievable
per second [54]. address is referred to as eater address and incorporates a pub-
lic key that is not correlated with any private key to prevent
2) PROOF OF WORK coin retrieval. Once a coin is sent to the eater address, it is
The proof-of-work mechanism commences with the calcula- permanently eliminated from the network. In PoB, the miners
tion of the hash value of the block header. The block header do not invest in physical currency as the cryptocurrencies
43630 VOLUME 9, 2021

TABLE 5. Pure alternative consensus mechanism.
are burned intentionally to denote the investment in 6) PROOF OF TIME

blockchain. Burning cryptocurrency generates virtual min- Considering the computational deficiency of PoW, this
ing power; hence, the more coins a user burns in favor of approach cannot be adopted in various range of applications,
the system, the more mining power it acquires. In addition, including electric vehicles that maintain more confined pro-
the miner, as mentioned above, is more likely to be appointed cessing capability. Hence, the proof of time is an alterna-
as the validator of the block. The notion behind PoB is similar tive that prevents spanning and Sybil attacks by exploiting
to PoW as getting the title of mining a block in PoB bears a the time difference between the two transactions. Moreover,
resemblance to purchasing computing resources in PoW. All clients are obliged to collect random tokens which make the
transactions that indicate transferring coins to eater addresses potential attacks costly as the intruder needs to outpace the
are recorded, and SHA-256 is employed for the calculation throughput of each transaction, along with their associated
of the burn hash concerning each transaction in the network. timestamps [56], [59], [99].
Eventually, the miner who holds the least assess of the burn
hash attains the mining right [58], [98]. 7) PROOF OF SPACE/PROOF OF CAPACITY/PROOF
OF STORAGE
5) PROOF OF EXCELLENCE Proof of space is a consensus algorithm that is cheaper than
Proof of excellence is an unexplored conceptual solution PoW in terms of required computing infrastructure, as it
to the distribution problem that has been originally intro- requires the use of hard disks or cloud storage systems.
duced in the PoS white paper [57]. This approach incorpo- It operates based on large stored data sets, known as parcels.
rates intermittent tournaments and the performance of the Their multiplicity enhances the probability of mining a new
contributors in the tournament, which imitates the real-life block for the corresponding node. The proof of space is
tournament rewards. A game is employed to select the node executed in two stages. The first stage is plotting, where the
that keeps the consensus of the blockchain and establishes hard disk capacity that miner has devoted is evaluated by
an inequitable pattern that allows competent players to write incorporating Shabal [100] hash function and plotting the
blocks frequently. As a consequence, the blockchain will hard disk. The hash function is then seeded using miner’s
turn into a partially centralized platform controlled by good ID and nonce [60]. Mining is performed during the second
players [57], [68]. stage. It refers to the most recent block on the chain to
VOLUME 9, 2021 43631

calculate the generation hash. The total number of scoops is Lazooz Dapp for the social transportation web. It is a decen-
then calculated by incorporating the hash module to generate tralized autonomous platform that makes official decisions
the target value that also uses the outputs of the plotting using the collective mechanism. As such, it can be used for
stage. Afterward, the network re-calculates the scoop for each decision-making procedures that require the elimination of
hash to validate the deadlines for each miner. The miner that human interventions [62], [108].
correlates the shortest published deadline generates the next
block and receives a reward for the transaction. The advantage 10) PROOF OF AUTHORITY
of PoSpace is its energy efficiency, as it does not impose Proof of Authority has been proposed as an underlying con-
high requirements on hardware. However, it is known to be sensus algorithm for permissioned blockchains. This algo-
susceptible to malware attacks, since its hashes are stored on rithm substitutes a lighter message transmission scheme in
a hard disk. This enhances the vulnerability of the data and comparison with BFT algorithm, which has led to the supe-
risks of being tampered with. Spacecoin [101], Chia [102], riority of this approach concerning its performance. There
and Burstcoin [103] are a few of the many approaches that are two implementations of the PoA algorithm known as
have incorporated PoSpace [43], [104]. Other names used for Aura and Clique [109], which was primitively implemented
this consensus mechanism are Proof of Capacity and Proof of on ethereum for private networks. Both Aura and Clique
Storage. exploit a similar block proposal scheme in which the trusted
authority such as the current mining leader proposes a new
8) PROOF OF EXISTENCE block. Afterward, Aura performs the block acceptance pro-
Proof of existence as an online service exploits a decentral- cedure, which is not required in Clique implementation.
ized certification SHA256. PoE permanently validates the PoA is executed in several time divisions and during each
existence of data by storing its cryptographic digest and interval; the authorities alternate using round-robin to pro-
the corresponding submission date using blockchain. This pose blocks. Each proposed block is accepted once its signed
service can publicly prove the ownership of data without off by the majority of authorized entities. Moreover, the pro-
revealing the data itself. It also eliminates the requirements cedure of discerning authorities results in the centralized
for trusting any central authority. This approach provides configuration of PoA, which makes this approach appropriate
anonymity, privacy, and decentralized proof that does not for private consortiums [63], [110].
rely on a single centralized entity. The application of PoE
ranges from ensuring the integrity of documents, document 11) PROOF OF APPROVAL
time stamping, and denoting the ownership of data without Proof of approval is acknowledged as a permissionless con-
disclosing the content. PoE, as a blockchain notary service, sensus that intermittently publishes blocks within predefined
provides instant and secure validation of the existence of any intervals. Each node can propose a new block; however, nodes
document, agreement, or contract. Moreover, it implies rules that do not indicate valid transactions are eliminated, and
to allow updates in the documents and keeping track of the stakeholders with a minimum stake are authorized to compete
updates [61], [105]. Factom [106] is an example of approach in the block creation procedure. Once the block generator has
that incorporates proof of existence. been selected, it broadcasts its corresponding approval block
containing the acquired confirmations and is rewarded with
9) PROOF OF MOVEMENT the transaction fees of the proposed block and coins [64].
Proof of movement has been proposed as an innovative
consensus that incentivizes road miners to run Lazooz 12) PROOF OF KNOW HOW
Dapp [107] on their smart devices. Lazooz encourages the Since the adaption of blockchain in the context of standard-
miners to participate in sharing their transportation data and ization, communities can significantly enhance the develop-
to assist Lazooz to eliminate meandering by weaving the ment procedure of standards. Proof of know how (PoKH) is
social transportation web. Participants receive incentives in an underlying consensus algorithm for blockchain-oriented
the form of tokens that are acknowledged as ‘‘zooz’’ and standard drafting that incorporates KH now as tokens. This
can be exploited for transportation and ride-sharing services. consensus will ensure the each proposed piece of guidance
The number of collected tokens correlates with the traveled will be implemented at least once prior to being appended
distance. Moreover, Lazooz integrates several algorithms to to the standard. Each participant will be incentivized to
facilitate decision-making procedures in the absence of user use PoKH concerning the quality and quantity of proved
intervention, monitoring the utilization of particular districts tests. Moreover, PoKH will make sure to close each block
to enhance services in accordance with the region’s active after appending the implementations that exemplify guid-
contributors. Lazooz is a decentralized scheme that allo- ance. This approach will affect market intake and standard
cates weight to miners conforming to frequent crowdsourc- adoption through practical exemplifications [65].
ing since conventional decisions are made by participants.
Proof of movement has been introduced as an alternative 13) PROOF OF INTELLIGENCE
to commonly used consensus algorithms. This algorithm Proof of intelligence has been introduced as an underly-
incentivizes users to share their transportation information on ing consensus mechanism for high-level blockchain-oriented
43632 VOLUME 9, 2021

smart networks beyond cryptocurrencies and ordinary trans- mining procedure. This will facilitate ensuring the sequence
actions. It can be incorporated as an economic assess to pre- of block generation rights. In order to meet the consensus and
vent denial of service attacks. Moreover, it can be considered authorize a block, FC defines a consensus height parameter
as a reputation qualifier in the context of machine learning or that is reached when more than 50% of the miners have
neural network training [66]. appended a block to any given block on the chain [70], [114].
14) PROOF OF CREDIT 18) PROOF OF COOPERATION

NULS project has proposed an opensource blockchain infras- Proof of cooperation has been implemented by Faircoin [115]
tructure that incorporates cross-chain technologies to facil- in which every single node follows the same regulations to
itate the development of applications. NULS provides the preserve the integrity and security of the network. In order
required building blocks and allows developers to employ the to maintain the integrity of the cooperatively validated nodes
consensus of their choice. However, the main chain executes (CVNs), each node is authorized using P2P proof of cooper-
the Proof of Credit consensus, which has been proposed as ation. It facilitates the collection and transmission of assets.
an alternative to PoS as it requires to lock a certain number Moreover, PoC validation mechanism has proven to be effi-
of tokens prior to the execution of any node on the network. cient since no matter how many CVN’s have authenticated
Moreover, the proof of credit incorporates the master node the proof, only one mutable signature is required for the
approach in which the agent node that is considered the appendance of the block into the blockchain [116].
owner is more likely to be incentivized from transaction
fees. It ensures the stability and integrity of cooperating 19) OBELISK
nodes [67], [111], [112]. Obelisk consensus mechanism has been initially employed
in SkyCoin [72] that proposes a fully scalable protocol for
15) PROOF OF PLAY decentralized services. To prevent decentralization and its
In the context of blockchain P2P games, coping with trans- corresponding issues, SkyCoin leverages nodes that are sub-
action costs and latency is a significant challenge. Previously scribed to a list of trusted nodes. Nodes that maintain greater
proposed consensus algorithms such as PoW have become the subscriptions are given more credit in terms of influencing
bottleneck when applied to P2P blockchain games. Hence, the network. Since obelisk is not dependent on the mining
Proof of Play has been proposed as a decentralized approach incentives, it does not suffer from the vulnerabilities of the
to address the compatibility issues using a seamlessly inte- commonly employed consensus mechanisms [117].
grated blockchain in P2P games. This consensus overcomes
the data storage issues by simply creating a consensus using 20) PROOF OF VALUE
the blockchain itself, which prevents modification of the Proof of value has been introduced as an alternative to
game for the use of blockchain. The notion behind PoP is proof of replication. It regulates the discerned values of
to incorporate the user’s cognitive cooperation in the mining different participants exploiting a P2P evaluation mecha-
procedure as both miner and user. This incentivizes the pro- nism. To maintain consensus, it assigns influence weights
cedure by changing the users’ perception from running the concerning the participating value and the acquired align-
blockchain to playing the game [68]. ments with the overall perception of the value. PoV con-
centrates on the human contribution and incentivizes the
16) PROOF OF FOUNDATION active participation with respect to community values. This
Nexty [113] is an implemention of a blockchain platform approach alters perspectives from algorithms to human
that performs immediate transfers at zero cost. Nexty exploits intervention [73].
a Dual Cryptocurrency Confirmation System (DCCS) along
with Proof of Foundation (PoF) consensus to eliminate 21) PROOF OF DISINTEGRATION
transaction delays and spamming. Proof of foundation is a B3Coin [118] implements proof of disintegration as an under-
consensus algorithm that is incorporated within a transaction lying consensus mechanism to address the drawbacks of
confirmation protocol and generates the mining reward con- Proof of burn. Using PoB, coins are burned once sent to
cerning the computing power that has been utilized during an irretrievable address. To eliminate the circulation caused
transaction confirmation. Proof of foundation outperforms by the proof of burn, PoD destroys the corresponding coins
the commonly used consensus algorithms in terms of asso- to make them irrecoverable and eliminates the additional
ciated transaction fees and throughput [69]. circulation and supply. Proof of disintegration is applied on
the nodes known as fundamental nodes that are more likely
17) FLASH CONSENSUS to be incentivized for staking [24], [74].
Flash consensus (FC) exploits representatives to maintain
consensus. This allows participants to use FLASH coins 22) PROOF OF LEARNING
to vote for representatives concerning the transaction fees. Proof of learning is introduced as an underlying consen-
Instead of incorporating a common PoW or PoS, FC allocates sus mechanism for WekaCoin. To decrease the computa-
a time slot to each miner that allows them to perform the tional loss affiliated with hashing puzzles, proof of learning
VOLUME 9, 2021 43633

incorporates a distribution of classified machine learning 26) PROOF OF PARTICIPATION AND FEES
schemes as a verifiable database. Hence it implements Proof of participation and fees consensus algorithm has been
machine learning contest for verification of the transactions proposed as an alternative to PoW, initially employed on
within the blockchain. The term ‘‘machine learning compe- JCLedger. PoPf performs mining procedure exploiting the
tition’’ refers to the crowdsourcing methods where participa- contribution of candidates that are selected in accordance
tors are incentivized for performing published tasks. Proof of with two factors: the fees that the participant has paid and the
learning was initially inspired from reCAPTCHA [119]. participation intervals. PoPf ensures that users with a constant
It proceeds to line up transaction validation with classifica- contribution to JCLedger are given the chance of being an
tion and storage of machine learning approaches towards the accountant. Performance evaluation of PoPf indicates the
development of a public distributed database [75]. efficiency of this approach when compared with PoE in terms
of ensuring the computing power efficiency without imposing
23) PROOF OF ELIGIBILITY security threats [79].
BFCV (Byzantine fault tolerance connected vehicles) incor-
porates an underlying consensus mechanism based on proof 27) PROOF OF LOCATION
of eligibility. This approach reaches Byzantine agreement Proof of location has been introduced as a distributed and
exploiting the unique features of each node (vehicle) and decentralized consensus mechanism to localize incorporating
eliminates the attempts of irrelevant nodes for cooperation agents in a timely manner. PoL reaches consensus when
in a consensus mechanism. Eligibility of cooperating nodes verification regarding the presence of an agent at a certain
is evaluated based on qualities such as the presence of a point is attained in due course. After the agent’s location
node (vehicle) within the vicinity of the information source. has been broadcasted throughout the blockchain network,
Proof of eligibility accelerates the consensus procedure in other agents can be confident with the received information
a distributed manner. The performance evaluation results concerning location coordinates without trusting the broad-
indicate the superiority of this algorithm among information casting agent itself. Dynamic proof of location also provides
dissemination schemes [76]. a permissionless and autonomous network of radio beacons
that exploits decentralized time synchronization to provide
24) PROOF OF REPUTATION conserved location verification services [86], [121], [122].
Proof of reputation is the underlying consensus mechanism
for the blockchain network that generates a reputation for 28) PROOF OF CREDIBILITY
each node regarding its assets, transactions, and contribu- Proof of credibility has been proposed as an underlying con-
tions in the consensus procedure. Proof of reputation consists sensus mechanism for the detection and prohibition of invalid
of 3 main stages, including leader selection and block gener- news within social networks. Performance evaluation of this
ation, reputation-oriented consensus, and finally updating the approach indicates 89% precision in the detection of fake and
reputation values. Once a leader node proposes a new block, tampered news. Proof of credibility considers each user in the
it is evaluated through reputation-oriented voting. The node social network as a peer that contributes to a distributed ledger
retaining the highest reputation value is validated, and its that indicates immutable and cryptographically secured logs
proposed block is verified accordingly. Nodes that maintain of discovered rumors. Each block comprises several invalid
higher reputation values are involved in the voting process, or tampered news. It is appended to the blockchain network
and each node is incentivized regarding its preserved rep- after complying with a pre-defined number of rumors that
utation value. The competency of the proposed scheme is should be incorporated in each block. Finally, the detection
highly dependent on the leader selection since the voting procedure performed by proof of credibility is shared among
consensus of a highly reputed node enhances the security of all peers within the social network platform [87].
the protocol [77], [120].
29) PROOF OF HISTORY
25) PROOF OF VOTE Proof of history has been introduced to tackle issues asso-
Proof of vote has been proposed as a PoW alternative for ciated with intensive computation. This approach executes
consortium blockchain. The consensus is reached through a the SHA-256 hashing algorithm in a consecutive manner to
decentralized voting arbitration among consortium partici- exploit the output of each round as the corresponding input to
pants. Four security identities are designated for participants the subsequent round. Leaders are in charge of confirmation
to pursue the voting mechanism. Using the proof of vote, and integration of each transaction with the prevalent hash.
submission or validation of the generated blocks does not PoH is known to be energy efficient as it does not perform
require the intermediary of third parties. In comparison with intensive mining procedures in comparison with traditional
PoW as a fully decentralized approach, this scheme follows a PoW. However, it leans towards wealthier leaders which
discrete voting and executive principle to decrease the trans- results in a more centralized and deterministic process and
action verification time, enhance the convergence, reliability requires more capacity due to consecutive execution of the
and security [78]. hashing function [43], [88], [123].
43634 VOLUME 9, 2021

30) PROOF OF INDIVIDUALITY 35) RIPPLE/PROOF OF CORRECTNESS

Proof of individuality proceeds through individual-to- Ripple is a consensus mechanism that incorporates validat-
individual verification where participants are arranged into ing nodes to preserve a set of trusted nodes acknowledged
groups of 5 and each group is in charge of a virtual (video) as Unique Node List (UNL). To append transactions into
pseudonym party hangout within the same periods. Partici- the ledger, UNL is required to maintain agreement among
pants of each group are required to ensure their counterparts 80% of the nodes. UNL nodes verify the transactions and
are involved in just one hangout at each interval. Hence, broadcast their corresponding votes to the network. Unver-
each participant validates POI of the others and, with the ified transactions are discarded and retained in the open
termination of hangouts, every contributor is aware of the ledger until meeting the validation criteria. As long as the
POI which belongs to each user. However, this consensus number of faulty nodes remains under 20%, the ledger is
mechanism cannot ensure security as physical pseudonym authentic [85], [127], [128].
parties and it is difficult to implement due to its reliance on
Ethereum, specifically in terms of security [81], [124]. 36) PROOF OF ENERGY
31) PROOF OF PERSONHOOD

Proof of Energy is a consensus mechanism for administra-
tion of the P2P energy trading using DLTs. PoE uses smart
Proof of personhood has been proposed as an underlying con-
contracts for regulating energy transactions without excessive
sensus mechanism that maintains anonymity by exploiting
energy consumption. After the validation of each smart con-
collective ring signatures. Ring signatures do not disclose the
tract, the next block generator needs to be elected to decide
keys that have been employed for the computation of a des-
on the next offer. The block generator is elected using the
ignated signature. This property allows PoP to acknowledge
proof of energy that incorporates a consumption-production
each agent using a cryptographic identity that correlates their
function to calculate the self-consumption proportion of each
physical and virtual identities. PoP allows agents to become
prosumer. The user that retains equal consumption and gener-
miners after being substantiated and configuring a mining
ation is chosen as the block proposer and incentivized accord-
pool. Hence, the prospective block proposer will be elected
ingly. This approach empowers the prosumers to enhance the
by applying a RandHound method on the ledger [81].
operation of both distribution and transmission systems [89].
32) SIEVE
SIEVE was initially employed by Hyperledger as an under- 37) PROOF OF WITNESS PRESENCE
lying consensus mechanism that tolerates non-determinism. Proof of Witness Presence has been introduced as a con-
Once performed by distinct replicas, it results in contrasting sensus mechanism for location verification to facilitate sit-
outputs. SIEVE considers the blockchain as a block box uation awareness for crowdsourcing applications in smart
that compares the results from different replicas to sieve out cities. It aims to securely raise location awareness without
the sequence of diverging outputs. If the diverging values disclosing privacy-sensitive information. Proof of location
within a procedure reach a certain threshold, the procedure is the vital core of the proof of witness that enables secure
is eliminated [15], [82], [125]. verification of the user’s location. Collective measurements
are required to ensure the presence of users in particular loca-
33) PROOF OF STAKE tions, which is then verified in a private or public network.
Proof of stake (PoS), which was initially proposed for Peer- The notion behind proof of witness presence is to affirm the
coin, has been used as an alternative to PoW to eliminate the decisions made in the physical space [90].
excessive power consumption of nodes. Since the election
of the block proposer based on the account balance seems 38) PROOF OF OWNERSHIP
unfair, many proposed solutions incorporate the stake size.
Proof of ownership has been proposed to ensure a trusted exe-
Although PoS is energy efficient in comparison with PoW,
cution environment for participants. This procedure can be
it is not resilient to attacks. Accordingly, several blockchain
employed to certify the integrity and ownership of contracts.
solutions initially employ PoW and gradually transform to
The proof is established using a block header and pseudonym.
PoS [83], [126].
The consensus is met when a proposed block generated by a
34) PROOF OF ELAPSED TIME particular trusted execution environment retains most proofs
Proof of elapsed time is an underlying consensus mecha- with unique pseudonyms [56].
nism which has been initially proposed by Intel to improve
energy efficiency and eliminate the waste of resources. 39) BITCOIN-NG
PoET is highly dependent on dedicated hardware to restrict Bitcoin-NG is a consensus mechanism that employs BFT
cooperation and decentralization. PoET reaches consensus and Bitcoin’s trust pattern. As blockchain consensus, Bitcoin-
by the random election of block leaders where the win- NG’s performance has been significantly enhanced as it dis-
ning odds are spread out evenly throughout the network, sociates the leader selection and transaction arrangements.
and each node holds the same chance of becoming the Accordingly, the robustness of the blockchain improves as
winner [84]. it reaches more bandwidth. This mechanism increases the
VOLUME 9, 2021 43635

permissive latency and bounds it to the network diameter, 2) FaB PAXOS

which actualizes the execution of trustless consensus at a FaB paxos has been proposed as the very first Byzantine
global scale [34], [91]. approach that reaches consensus within a minimum num-
ber of communication arrangements while eliminating the
40) PROOF OF ASSET/PROOF OF PROVENANCE expenses associated with digital signatures. FaB paxos is
Proof of asset was initially employed by DigixDAO, generalized by dissociating fault tolerance proliferation from
an Ethereum inspired platform that eliminates the middleman speed replication. To enhance the processing speed, FaB
from real gold transactions. It has been incorporated to track paxos relies on gracious execution that, in the presence of
down the gold during different stages of the transaction using authentic leaders, can tolerate process failures. The outstand-
the asset card. The asset card contains the transaction infor- ing difference between the previously discussed fast paxos
mation, such as gold bar’s serial number and purchase receipt, and FaB paxos is their failure models: fast Paxos tends to fail
that is uploaded into a smart contract. Moreover, the tracking by a crash while FaB Paxos fails more arbitrarily, [131].
process is performed using private keys, which enhances the
security and transparency of DigixDAO [92], [129]. 3) RAFT
Raft in an underlying consensus algorithm that was initially
41) PROOF OF BELIEVABILITY adapted by Quorum. This algorithm is a simplified extension
Proof of believability has been proposed as an extension to the for the paxos algorithm, which is too theoretical and compli-
PoS consensus mechanism. PoB incentivizes the participants cated to be well received. Raft incorporates a state replication
using untradeable tokens known as Servi. Servi serves as a model in which all transactions are propagated among all
measure of the believability value associated with each user participating nodes. It allows a leader node to generate the
that indicates positive reviews regarding former behaviors. next block and eliminates the generation of inessential vacant
Higher believability value increases the chances for block blocks. To endure f faulty nodes in the blockchain network,
creation. PoB employs two types of validators: believable val- raft requires the deployment of at least 2f + 1 nodes within
idators that are elected algorithmically, and normal validators the network [132].
that are chosen randomly. In case of fraud behavior detection,
the node will be penalized by losing all of its stakes [24], [93]. 4) XPaxos
Xpaxos was introduced as a state-machine replication con-
42) PROOF OF TRUST sensus mechanism for the XFT protocol and initially
Proof of trust is a reliable consensus algorithm proposed employed by Apache Zookeeper. This consensus is crash
for open public service networks such as crowdsourc- fault-tolerant and capable of detecting noncrash faults that
ing. It exploits the participant’s trust value and RAFT’s result in a contradictory state of the framework. To eval-
leader election mechanism as a reference to determine the uate the performance of xpaxos, it has been employed on
transaction validators. Incorporation of trust components and Amazon EC2 data centers. Results indicate the significant
incentive measures within PoT prevents the resource pitfalls performance of this scheme as it outperforms the existing
associated with traditional PoW. Proof of trust can tackle BFT protocols in terms of throughput and latency in the
the deceptive behaviour correlated with crowdsourcing net- presence of geo-replicated settings. There is a special case
work and the scalability challenges of BFT-based consensus of message patterns in xpaxos where, in the case of t = 1,
algorithms. Proof of trust enhances fairness and security by it can tolerate a single fault with only two active replicas in
reaching consensus through four phases as it distributes par- each span [133].
ticipants power into different stages to increase scalability
and ensure consistency of the procedure [94]. 5) MPaxos
Multipaxos is another variation of paxos that designates a
F. PRIMITIVE COMPLIANT EXTENSION CONSENSUS leader to order the received commands and preserves only
MECHANISM one leader at a time. The single leader at a time principle,
1) FAST PAXOS preoccupied with the computational resources, breaks down
Fast Paxos is proposed as an extension to the classic Paxos, the performance, and prohibits it from scaling alongside the
where nodes directly transmit their proposed block to the val- deployment dimensions. Moreover, multipaxos is neither sus-
idator and bypass the leader. This approach narrows down the ceptible to locality or complicated command patterns, nor it
traditional three-message delay to only two-message delay. incorporates dependency relations [16].
Saving one message delay accelerates the learning procedure
in the absence of collision and attains any required extent 6) RPaxos
of fault tolerance with the least feasible procedures. In case Ringpaxos is an optimized protocol that has been derived
of collision, only a single message delay will be appended, from paxos for clustered environments. Unlike paxos, ring
which is still the least number of required message delays for paxos arranges the acceptors in a logically conducted ring and
a common consensus mechanism [130], [141], [142]. allows coordinators to stabilize the communication amongst
43636 VOLUME 9, 2021

TABLE 6. Primitive-compliant extensions.
acceptors. Ringpaxos preserves the reliability of the clas- EPaxos can achieve high throughput through uniform load
sical paxos, which can be employed efficiently to ensure balance and slight performance reduction in the advent of
safety under asynchronous circumstances. Similar to paxos, crashed replicas. This consensus has been evaluated while
Ringpaxos can ensure safety if several coordinators execute implemented on Amazon EC2. In addition, epaxos incor-
concurrently, but may not be able to ensure liveness [134]. porates fast quorums to convey nonconflicting commands.
However, the linear graph-based scheme that it has employed
7) MRPaxos to present the sequence for the execution of the commands
Multi-ring paxos has been introduced to scale the group com- might result in confronting complex dependencies. The inves-
munication protocols to a large number of nodes. This con- tigations indicate that the decentralized and uncoordinated
sensus scheme proceeds by parallel orchestrating a boundless nature of epaxos has resulted in the availability and perfor-
number of ring paxos instances. Multi-ring paxos employs an mance consistency of this approach in the presence of local
atomic multicast process that allows multicasting messages and wide-area replications [137].
to groups of receivers and ensures delivery by evaluating
the receivers that convey identical messages. The complexity 10) WIDE AREA NETWORK FLEXIBLE PAXOS
of multi-ring paxos is due to the deterministic incorporation WPaxos has been introduced as an extension to paxos with
scheme, which results in dynamic load and deviation among improved throughput and decreased latency. As a multileader
engaged paxos rings. In the two-ring execution of multi-ring approach for WAN deployment, it allows multiple simul-
paxos in the presence of a single message, two learners are taneous leaders to obtain the ownership of objects from
associated with two groups. After learner2 receives the m one another. This enables the algorithm to conform to the
message, it is not able to deliver it as it needs to ensure transformations of access locality and maintain object space
the execution order by delivering one from group2. Hence, partitioning. WPaxos has been evaluated by implementation
it begins to buffer message m until the coordinator of ring- across 5 AWS zones that resulted the superiority of this algo-
paxos2 realizes its current rate is below the expected rate and rithm amongst other partitioned and leaderless implemen-
puts forward a skip message order to allow learner2 deliver tations of paxos. WPaxos implements the FPaxos’s flexible
message m [135]. quorum approach which allows the existence of multiple
simultaneous leaders in the object space that is associated
8) M2 Paxos with each leader through object space partitioning [138].
M 2 Paxos is a variation of paxos that leverages the quorums
that compound a great number of nodes to accelerate the deci- 11) FPaxos
sions. It is a scalable and high-performance implementation As a variation of paxos, fpaxos is fast and capable of imple-
of paxos that consists of the coordination phase, accept phase, menting flexible quorums. Unlike paxos, it only relies on
decision phase, and acquisition phase. This allows the con- disjoint sets of participants for affirmation of the proposals,
sensus mechanism to determine the sequence of commands which significantly decreases the latency of the network.
with the optimal cost of two communication delays in the By representing the quorum size to developers during the
advent of dispensable workloads. M 2 Paxos has been able to replication phase, fpaxos simply allows them to maintain a
address the shortcomings of other variations of paxos, includ- customized balance among adversary tolerance and latency.
ing the single leader layout of paxos and multipaxos that This enables absolute scalability that comes at the price of
prevents performance scaling and the performance deficiency ineffective fault tolerance [139].
of epaxos when the number of nodes overtakes seven [136].
12) CHEAP PAXOS
9) EPaxos Most consensus algorithms with asynchronous communica-
Egalitarian paxos is another variation of paxos that can toler- tion mode need at least 2f + 1 processors to be able to
ate up to two failures and preserve optimal commit latency. tolerate the failure of f processors. Accordingly, cheap paxos
VOLUME 9, 2021 43637

represents a dynamic paxos variation that proposes the coop- employed on dynamic procedures as it incorporates authenti-
eration of f + 1 processors that are actually capable of cated structures like merkle tree. Within dynamic procedures,
proceeding within the network and keeping the remaining DPos eliminates the requirement of regenerating tags for all
processors as auxiliaries that provide recovery in the failure existing blocks. Instead, DPos incorporates homomorphic
of leading processors [140]. message authentication codes and homomorphic signatures
that allow regenerating tags only for the updated blocks dur-
G. PROOF COMPLIANT EXTENSION CONSENSUS ing any dynamic procedure that comprises inspection of data
MECHANISM integrity [145].
1) INTERACTIVE PROOF OF STAKE
4) PROOF OF STAKE BOO
Interactive proof of stake is able to entail communication in
the block generation procedure. IPoS minimizes the number PoS Boo is an extension to Etheurums PoS casper that has
of variables that a single miner can go over, which results been implemented as SHEILDS’s underlying consensus. This
in augmenting the network’s resistance to grinding attacks. approach maintains a progressive quality to enhance the resis-
Moreover, this algorithm maintains a static balance resistance against a wide range of malicious attacks, including the
tance against drifting attacks since it does not incorporate 51% attack in which the fraudster should possess the majority
timestamps or delays. During block generation, to follow the of minted coins and risk them on the execution of an attack.
ticket generation rules, instead of initiating the process with In PoS Boo, the reward will be calculated as a multiplication
one genesis block, the blockchain starts with one genesis of the network weight and fixed block reward. This method
block per participant. Tickets are generated using the seed eliminates transaction censoring of PoW by choosing the
and hold a distinctive value known to miners. Afterward, block creators randomly and the validators globally. PoS Boo
each block is evaluated in accordance with the ticket score, penalizes the miners that contribute to a fork by voting for
and the blockchain that has acquired the highest score is more than one block with the same height or voting for a false
incentivized [143]. block greater than a preset threshold [24], [146], [176].
5) PROOF OF EXERCISE
2) DELEGATED PROOF OF CAPACITY
Proof of exercise is a sustainable extension to PoW that
Fii is a user-friendly crypto platform implementation that allocates a matrix-based logical challenge acknowledged as
alters the perception of participants about cryptocurrencies. an eXercise to each miner in the blockchain network. The
This platform, available for all users regardless of their procedure initiates by allocating the miner’s valid transac-
motives, incorporates a delegated proof of capacity consensus tions to a random exercise matrix to find the corresponding
that makes this platform compatible with a wide range of end solution to each matrix. The solutions are published to ver-
devices. DPoC works based on a set of precalculated hashes ifiers for endorsement until they satisfy the minimum num-
that are registered in the mining pool and FiiPOS, which is ber of required validations. Afterward, the entire transaction
incorporated as an extra payment accepting feature, initializes procedure (including the exercise and verification details) is
the nonce. This is then be followed by submitting the nonce appended as a new block to the blockchain network [147].
and the corresponding hashes to the mining pool. Finally,
incentives are assigned concerning the number of hashes 6) PROOF OF USEFUL WORK
that have been involved throughout the block generation Proof of useful work is an alternative for PoW that oper-
procedure [174], [175]. ates based on the delegation of low-degree polynomial prob-
lems. Within this approach, miners are expected to approach
3) DEDUPLICATED DYNAMIC PROOF OF STORAGE the problems posted by delegators. Contributors can choose
Deduplicated Dynamic proof of storage(DeyPoS) is a proof among the problems that have been posted to mine the block.
of storage extension that aims to effectively update the After termination of the mining procedure, the miner appends
records that have been outsourced to the cloud server. It facil- the proof of use details into the block that enables verifiers to
itates the integrity verification process for the corresponding inspect whether the problem has been solved using the hash
users. DeyPoS attains a secure cross user deduplication in of the block. Proof of useful work can be applied to a wide
multiuser cloud storage system using Homomorphic Authen- range of practical problems by preserving their PoW qualities
ticated Tree (HAT). This approach simply authorizes clients as it is also acknowledged as a delegation for computation
to acquire the ownership of the data they previously uploaded, methods [24], [148].
without imposing an extra upload procedure. HAT is a binary
tree that allows DeyPoS to significantly decrease the com- 7) FAIR PROOF OF STAKE
munication expenses in both proof of storage and deduplica- Unfairness is one of the most significant drawbacks of PoS
tion stages. Each node of HAT is in correlation with a data since miners with the greatest share of stake are more likely to
block, and it does not impose any restriction on the number generate blocks, which results in collecting more incentives.
of data blocks. Moreover, it facilitates integrity verification Moreover, PoS is vulnerable to multibranching attack and this
and dynamic operations. Dynamic Proof of storage is mainly problem needs to be addressed in PoS extensions. Hence,
43638 VOLUME 9, 2021

TABLE 7. Proof compliant extensions.
the fair proof of stake has been proposed as a consensus 10) ALT-PoW
algorithm that incorporates exponential distribution in the The adaption of PoW attains consensus in a slow and
selection of block originators, which was formerly performed energy-efficient manner since this approach does not allow
by uniform distribution. It also successfully reduces the num- participants to gain any perception of progression as it
ber of forks and their corresponding length that minimizes the does not provide any information on how far they are
attacker’s acquisition [149]. from the solution. Accordingly, Alt-PoW suggests that
providing the aforementioned information for participants
8) TRUSTLESS PROOF OF STAKE simplifies the miner’s decision on whether it is worth devot-
TPoS has been proposed as an underlying consensus ing subsequent resources to keep solving a certain problem.
mechanism for Stakenet (XSN) that intends to establish Hence, Alt-PoW aims to provide progress information by
an integrated decentralized structure towards an operative fragmenting a problem into a sequence of problems that
investment implementation. TPoS ensures security of the requires the problem to be solved in several rounds. This
blockchain by incorporating merchant nodes that can be allows miners to evaluate their chances for mining the block
either the participant itself or the representative hired by as they receive information regarding all miners and their
participants in exchange for an agreed commission on the par- corresponding rounds. This approach allows miners to make
ticipant’s desired amount of XSN for staking. The merchant a viable decision in terms of identifying the right time to
nodes have been authorized to validate the transactions and commit to a block or suspend the mining procedure [152].
are not involved in block creation. However, the stakeholders
that comply with the minimum collateral requirements can 11) PROOF OF PROOF
run master nodes that are authorized to vote, verify transac-
Proof of proof consensus provides scalability by allowing a
tions, and generate blocks as well [177], [178]
recently developed blockchain also referred to as Security
Inheriting (SI) blockchain, to derive safety measures from
9) LIMITED CONFIDENCE PROOF OF ACTIVITY other Security Providing (SP) blockchains. The inheritance
LCPoA is a consensus algorithm proposed for IZZZIO net- procedure of PoP is independent of SP blockchain miners
work. This mechanism leverages proof of activity and limited and does not require permission from SP blockchain or
confidence that imposes a restriction on rewriting blocks. authorization from a centralized network or federated entities.
This feature enhances the resistance of LPCoA to 51% attack Inheritance does not impose any nontrivial or technologi-
by confining the target nodes so that the attack can be per- cal restrictions on SI blockchain that employs this protocol,
formed only on a limited number of blocks, which does other than eliminating the interaction of SI non-mining users
not affect the network. To track the rewriting block, LCPoA with SP networks since they are required to contribute to
generates automatic checkpoints to reduce the chances of retaining the blockchains native tokens. The SI blockchain
rewriting the blocks [24], [179]–[181]. is incentivized based on the state that has been published
VOLUME 9, 2021 43639

in the SP blockchain network. These publications are fur- 16) DELEGATED PROOF OF STAKE
ther referenced once a substitute fork is proposed to the DPoS has been proposed as an underlying consensus mecha-
SI network [24], [153], [182]. nism that outperforms its counterparts, such as PoW and PoS
using a block generation procedure that leads to faster trans-
12) MAGI’s PROOF OF WORK actions. DPoS reduces energy consumption by incorporating
MPOW restrains the blockchain network’s hash rate a one vote per share mechanism that enhances the number of
by frequently regulating the incentives based on an process coins. Since stakeholders vote for randomly selected
attraction-repulsion scheme. In order to hinder mining pools witnesses to preserve consensus, they are incentivized and
and incorporate low end devices in the mining procedure, penalized concerning their generated blocks and accomplish-
MPoW increases the incentives during passive mining inter- ments, respectively. However, DPoS suffers from a lack of
vals to prompt network operations in opposition to the decentralization as it incorporates an extensive number of
dynamic mining intervals so that incentives are decreased validators to reach consensus. Accordingly, it becomes sus-
to diminish redundant mining resources. However, this ceptible to primary attacks such as 51% attack, long-range
approach makes the network vulnerable to 51% attack as attack, balanced attack, etc., [48], [185]–[187].
it facilitates the procedure to conquer the network hashing
power for an adversary [24], [154]. 17) DELAYED PROOF OF WORK
DPoW, which was initially adopted by Komodo, is a promis-
13) MAGI’s PROOF OF STAKE ing solution for the double-spending issues. Other than
After the implementation of Magi’s PoW that maintains fea- blockchain, several other cryptocurrencies have adapted
sible solo-mining and cannot establish a correlated reward DPoW with multiple features in common, including insuf-
mechanism concerning hash rate, MPoS was introduced. This ficient staking power and susceptibility to potential attack-
approach incorporates the same attraction-repulsion scheme; ers. DPoW enhances the security of ongoing transactions
however, the stake weight is relatively proportional to the by exploiting the established blockchain with an enhanced
stake’s time span and number of coins. In MPoS imple- hash rate. In the case of 51% attack on Komodo, DPoW
mentation, there is a staking time threshold that restricts appends a security layer and integrates existing notaries to
the accumulation of offline coin age to one week and spec- ensure the security of the hashes. It also allows notaries to
ifies that increment in the number of coins in a stake switch between PoW networks if a significant hashing power
does not necessarily guarantee the augmentation of stake’s is provided [24], [158], [188].
weight [24], [155].
18) TRANSACTION PROOF OF STAKE
14) PROOF OF REPLICATION TaPoS has been proposed as another extension to PoS in
Proof of replication is a novel implementation of proof of which all nodes are required to participate in the introduced
storage that enables servers to persuade users about the security framework. In order to eliminate the reply attacks on
replication of certain data in a dedicated storage. Proof of forks, TaPoS requires every transaction to contain the hash
replication provides an interactive protocol to ensure the of the most recent block’s header as a proof of validation.
storage of unique physical copies, prevent deduplication of This approach reduces the fraud attempts for the generation of
several copies, and finally assure verifiers that the chal- alternative chains since stakeholders are constantly validating
lenge/response protocol has been incorporated for the accu- the blockchain after each transaction [189].
mulation of each replica [31], [156].
19) PRIME NUMBER PROOF OF WORK
15) PROOF OF CONSENSUS Prime number proof of work was originally adopted by
Proof of consensus maintains consensus, allowing servers Ppcoin and then Primecoin as a non-hashcash PoW mech-
on the network to authenticate transactions. PoC adapts the anism, also referred to as pure PoW. Unlike bitcoin, the lack
transaction verification procedure of PoS that does not require of currency within the network is adjusted using Moore’s law,
mining new cryptocurrencies. However, PoC reaches consen- and the mint rate is established in accordance with complexity
sus at 80% validation when employed by Casinocoin, which of the hashcash. Hence, the attacker is able to take over the
represents the fact that this approach maintains consensus network by manipulating the complexity model as it only
at an augmented validation rate. Each time a new ledger is requires obtaining under 50% of the network’s power. Since
generated, it is shared across the network. Once its verified, network security relies on the accuracy of the complexity
it is referred to as Last Closed Ledger (LCL) and used as a estimation, a fixed ratio is appointed for the approximation
precise reference for future transactions. Once the network of difficulty among prime chains [160], [190].
reaches a consensus on a valid transaction, it is followed
by the generation of a new LCL. However, using proof of 20) PROOF OF RETRIEVABILITY
consensus, the validation of transactions is mainly confined Proof of Retrievability is proposed as an extension to Proof of
by the number of servers [183], [184]. Capacity. It is incorporated as a distributed cryptographic
43640 VOLUME 9, 2021

cloud storage that verifies the integration of stored files with- 25) PROOF OF BLOCK AND TRADE
out requiring to preserve a copy or retrieve the original file. One of the challenges for implementing business
The verification procedure is performed using authentication blockchains (BBC) in the context of IoT is their lack of
data where challenges can be addressed without actually feasibility. Increasing the scalability of the BBC to meet IoT
possessing the response value [161]. criteria, the consensus mechanism is required to be moderate.
Accordingly, PoBT has been proposed to make BBC com-
21) PROOF OF ENERGY GENERATION/PROOF OF ENERGY patible with IoT applications by decreasing the computation
CONSUMPTION time and improving the storage of IoT nodes. The proposed
Proof of energy generation/consumption are proposed as mechanism proceeds with a two-step mechanism, including
extensions to the proof of energy. The difference is that PoEG trade verification and consensus formation, to improve the
prioritizes the prosumers that preserve higher generation val- performance of the network. As the increments in consensus
ues with reference to consumption values. Accordingly, these formation time result in decrements in transaction rates,
proposers are chosen as validators that verify the energy PoBT reduces the number of cooperating nodes and performs
transactions and append them to the blockchain. They are verification just for trade. Hence, the network reaches con-
incentivized by the amount of energy. Proof of energy consensus based on the number of participating nodes, which not
sumption is used to eliminate the peak hour consumption only increase the transaction rates but also reduce the required
and facilitate the evaluation procedure of transactions for bandwidth [166].
PoEG [163].
26) PROOF OF WEIGHT
22) MULTICHAIN CONSENSUS MECHANISM
Proof of weight is an extension to Proof of Stake and has
Multichain is an extension to PoW that employs round-robin been employed by Algorand. When using PoS, the number
to choose the validator nodes and attempts to address forking of tokens held by each participant determines their chance
by choosing the longest chain. Multichain grants the adminis- of discovering the subsequent block. However, the proof of
trative dominance to the miners of the genesis block. To pre- weight allocates weight values to participants in accordance
serve the mining diversity in the node election procedure, with the asset that each user holds in its account. This consen-
it relies on the rotations of the round-robin. Accordingly, each sus mechanism makes network resistance to double-spending
node gets the chance to append its proposed block to the chain attacks as far as at least two-thirds of the overall weighted
and broadcast that to the rest of the network [34], [162], [191]. fraction of participants are truthful. Despite the eminence of
proof of weight consensus, it is very difficult to incentivize
23) LEASED PROOF OF STAKE
users of such networks as PoW is not developed for genera-
LPoS is an extension to PoS. It is a promising solution to tion of passive revenue streams [167].
address the uncertainty issues associated with PoS. It incor-
porates a leasing option to allow nodes with a lower balance to
27) THRESHOLD RELAY
co-operate in the block verification procedure. This scheme
creates a flow within the network in which wealthier nodes DEFINITY [194] incorporates a threshold relay mechanism
can lease their funds to nodes on demand. This flow sig- as a consensus that includes beacons for leader selection
nificantly enhances the chances of inferior nodes in solving and consists of four layers. The layers provide, respectively,
the blocks. Accordingly, the acquired rewards are shared registered client information, distributed random beacons,
with the leaseholders. Nodes that preserve a higher amount probabilistic leader ranking protocol, and time-stamping.
of leased balance are more likely to be selected for block Beacons assign priority ranks to each node, and the blocks
creation. This scheme makes the network more decentral- that have been proposed by the nodes with higher ranks
ized and does not allow the network to be ruled by certain are more likely to be authorized. Finally, the block that
members [164], [192], [193]. holds the highest rank is sent throughout the network and
nodes are allowed to append the corresponding copies to the
24) PROOF OF TEE-STAKE (PoTS) blockchain [195], [196].
Proof of TEE Stake is an extension to PoS and aims to
address the deficiencies of this protocol in terms of secu- 28) DELEGATED PROOF OF REPUTATION
rity. To prevent long span attacks caused by Nothing-at- DPoR is an extension that addresses the deficiencies of Dele-
Stake phenomena, PoTS employs Trusted Execution Envi- gated Proof of Stake. This consensus is a semi-decentralized
ronments (TEEs) to ensure individuals will generate at least mechanism. Unlike DPoS, it does not rely on staking as a pre-
one block per height. TEE imposes all verifiers to sign the eminent factor resulting in a more constant coin circulation
blocks for exclusively growing heights to protect the net- within the network. DPoR employs the reputation factor as a
work from nothing at stake and malicious verifiers. It can representation of the node’s staked value, resource consump-
ensure the security without sacrificing the performance of tion, and contribution to transactions. Accordingly the voting
the network by preventing potential grinding or posterior procedure is performed considering the vote weight of each
corruptions using cryptographic schemes [165]. node, which is defined by the reputation value it holds [168].
VOLUME 9, 2021 43641

29) PROOF AUTHENTICATION the security of the Proof of Stake against adaptive attacks.
Proof of Authentication (PoAh) has been proposed as an Also known as Crypsinous, it incorporates a noninterac-
alternative consensus mechanism that preserves a lightweight tive zero-knowledge (NIZKs) proof and key privacy to
procedure by withdrawing the hash function from PoW. establish zero-like transactions and retain their autonomy.
Every node is involved in the ledger updating procedure Accordingly, a cheap key erasure is employed by NIZK for
where nodes are penalized by losing trust values in case of leadership proof to prevent revealing the coin value that
proving invalid authentication. Like PoW, PoAh performs was formerly performed by ouroboros genesis. This allows
two forms of authentication, including the authentication of explicit construction of the transaction system and stake shifts
each block concerning its source and incrementing the trust since the stake distribution is not communicated throughout
value associated with the corresponding validator. The per- the network [173].
formance evaluation of proof of authentication has shown that
this approach outperforms traditional consensus mechanisms H. BFT COMPLIANT
such as PoS, PoW, and PoA in terms of latency, computing 1) HYDRACHAIN CONSENSUS
procedure, and energy consumption [169], [197]. Hydrachain consensus mechanism is proposed as an exten-
sion to the BFT, which is highly dependent on a set of
30) OUROBOROS validators that confirm the sequence of transactions with
Ouroboros is a PoS extension that has been initially adopted low overhead. Round robin is used to select the proposer of
by Cardano [198]. It is a synchronous and permissionless the block from a set of validators in each round, and each
protocol that operates by dividing chains into epochs. Each round is initiated only after receiving more than 2/3 votes
epoch is associated with a slot leader selected from qualifying in the former round. Implementing this approach, normal
stakeholders. Hence, the chances of becoming a block pro- operations maintain low overhead since the proposed blocks
poser are proportional to the stake of a node. For adversary are presented with the quorum of signatures on the block of
tolerance considerations, ouroboros operates on a settlement the recent height [127], [221]–[223].
delay that ensures the security of the ledger when transferred
among participants. Participants are also incentivized con- 2) MODIFIED FEDERATED BYZENTINE ALGORITHM
cerning their honest contribution, evaluated through game mFBA is implemented on BOScoin [224] as an extension
theory using the participant’s collective interest [170]. to Federated Byzentine Agreement (FBA) that incorporates
Proof of Stake to preserve the governance framework. BOS
31) OUROBOROS PAROS allows users to freeze coins if, across all nodes, their total
Ouroboros Paros is a variation of ouroboros that provides amount of frozen assets is within a specific range of the
security in the presence of fully adaptive fraudulent. Paros number of coins. The frozen node ensures the security and
operates in a partially synchronous environment and informs integrity of the blockchain and it can be used to incentivize
the stakeholders about their leading slots in advance. Like the operating nodes. If a node is detected to act maliciously
Algornad [167], ouroboros paros incorporates the verifiable and forging the blockchain, the corresponding frozen coins
random function (VRF) for the generation procedure. The will be surrounded as a common budget [200], [201].
VRF is fed by a private key and a nonce that all participants
have agreed upon to generate a random number that deter- 3) HONEY BADGER BFT
mines the slot leader [171]. HB-BFT is introduced as another extension to BFT. Unlike
other alternatives, it is not concerned about the synchronicity
32) OUROBOROS GENESIS of underlying network or the timing suppositions. HB-BFT
Ouroboros genesis is the third variation of ouroboros, geared is an asynchronous BFT extension that tackles the network’s
towards security in a partially synchronous environment. deficient bandwidth with adequate computation assets. Nodes
Using a novel chain selection mechanism, it overcomes agree on the sequence of transactions that have been formerly
the deficiencies associated with former ouroboros variations stored in their buffers. This approach can be considered a
concerning the long-range attack. Having the genesis block transaction processing scheme based on an asynchronous
information, ouroboros genesis allows individuals to enter protocol. Investigation performed by Miller showed more
protocol execution for robust and dynamic operation. As the than 20,000 transactions per second for networks with less
performance evaluation of ouroboros genesis suggests, this than 40 nodes using HB-BFT [202].
approach retains dependability of the network against a fully
adaptive attacker in the dominance of under half of the stakes 4) SUMERAGI
and preserves the security [172]. SUMERAGI is an underlying consensus algorithm for Hyper
ledger Iroha that was inspired by the BChain algorithm [225].
33) OUROBOROS CRYPSINOUS SUMERAGI adapts BFT features in tolerating faulty nodes
Ouroboros Crypsinous is yet another ouroboros variation. and exploits a global sequence that considers two sets of
It integrates ouroboros genesis with zerocash [199] to ensure nodes in which 2f + 1 nodes are allocated to the first set,
43642 VOLUME 9, 2021

TABLE 8. BFT-compliant extensions.
and the other set is composed of the remaining nodes. Hence, 7) LEADER FREE BYZANTINE
to validate transactions, 2f + 1 signatures are required to All previously discussed consensus algorithms in the con-
reach a consensus on every transaction, which makes only text of blockchain consensus are either deterministic or
the first set of nodes capable of contributing to the consensus leader-based algorithms derived from BFT. In the BFT con-
procedure. Subsequently, the transaction is verified by the sensus, once a leader transmits a message at a slow pace with-
first set of nodes, and then evaluated by the other set in terms out activating the corresponding timeout protocol, it results
of the authenticity of signatures and contents. Accordingly, in an unstable performance. Hence, leader-free BFT has been
the ledger is updated, and the transactions corresponding to proposed as a deterministic approach that develops an asyn-
the hash are sent through the network [203]. chronous consensus from a partially synchronous one. It not
only preserves the security of a asynchronous consensus,but
5) TENDERMINT also integrates the liveness of the synchronous protocol [205].
Tendermint is a Byzantine algorithm that relies on DLS [226]
protocol that communicates through round leaders using 8) YAC
a star network topology. Tendermint reaches consensus YAC has been proposed as a decentralized BFT consen-
through three stages. Within the pre-vote stage, right after a sus algorithm to address the deficient message transmis-
new block has been proposed using round-robin, the valida- sion and leader associated with the classical BFT consen-
tors are required to decide whether to broadcast a pre-vote sus mechanism. This algorithm was initially deployed in
for the aforementioned block. To be a validator, Tendermint Hyperledger Iroha to ensure security and liveness of the
requires each node to lock its coins, which will be further used underlying transactions by tolerating the utmost f faulty
to incentivize or penalize the contributing validators. To pro- validators among 3f + 1 participating peers. Experimental
ceed from one phase to another (pre-vote, pre-commit,and results evaluating the performance of YAC in Hyperledger
commit) the block needs to obtain at least 2/3 of the votes Iroha illustrate eminent scalability of this scheme. How-
for transmission which in the commit phase corresponds to ever, to decrease peers exposed faults, the vote step delay
appendance of the proposed block to the network [41], [227]. needs to be modified concerning the number of participating
validators [206].
6) ISTANBUL BFT
Istanbul BFT is a replication-based consensus mechanism 9) FEDERATED BYZANTINE FAULT TOLERANCE
originally employed by Quorum Chain and incorporates Federated BFT has been incorporated by Ripple and Stel-
PBFT’s 3 stage procedure (PRE-PREPARE, PREPARE, and lar as their underlying consensus mechanism. Since both
COMMIT). During each round, every node participates in a these cryptocurrencies perform work on decree currencies,
random selection of the block proposer through the round- they require incorporating a highly fault-resistant consen-
robin. Within the execution pattern of Istanbul BFT, each sus to ensure higher transaction rates along with scalability.
proposed block needs to acquire 2f + 1 state messages from In FBFT, each node generates a unique node list (UNL) indi-
the validators in order to be verified from one stage to another. cating committed nodes that will be then used for validating
After obtaining 2f + 1 state messages, the commit message transactions. Receiving at least 80% of the UNL votes is
is sent throughout the network to finalize appending the required to ensure the verification of the transaction and, thus,
proposed block into the blockchain [204]. it is appended to the blockchain [210], [228].
VOLUME 9, 2021 43643

10) HASHGRAPH/ABFT enables dynamic ownership transformations (as opposed to

Hashgraph is an asynchronous BFT that retains consensus the former procedure of assigning an exclusive leader to
by imposing restrictions that indicate only messages from all commands inconsiderate of their corresponding fluctuant
genuine nodes will be transmitted. Moreover, the consensus characteristics). Eplis adapts a linear mechanism that results
will be preserved unless the attacker is in control of more in 2-fold and 3.5-fold acceleration versus XPaxos and BFT,
than 1/3 of the votes. Hashgraph does not rely on votes as respectively [212].
it executes a gossip protocol in which each participant is
allowed to reach out to its counterparts and perform synchro- 15) ZYZZYVA
nization. Hashgraph employs an event-based data structure to Zyzzyva has been introduced to address the shortcomings of
store transactions that are accompanied by timestamp asso- BFT by employing a speculation mechanism that eliminates
ciated with the generation time of each event. ABFT makes the execution of the three-phase commit protocol prior to
the network resilient since each event carries not only the replying to the client’s request. Accordingly, the order pro-
transactions but also the hashes of the former events signed posed by the primary server is employed to allow the client
to ensure cryptographic security [208], [229]. to discover potential inconsistencies. Zyzzyva relies on three
stages of agreement, view change, and checkpoint. It also
11) SCALABLE BYZENTINE FAULT TOLERANCE consists of a fast, two-way case communication pattern. The
Scalable byzentine fault tolerance is proposed as an exten- fast pattern follows a simple regimen in which the primary
sion to PBFT. SBFT allows block generators to validate node forwards the acquired request to the replicas for an
transactions within 24-hour intervals and reaches consensus speculative implementation. Once the request has been exe-
concerning the approval from at least 2f + 1 nodes out of cuted by replica, the corresponding response is returned to
3f + 1 cooperating nodes. SBFT enhances the scalability the client. The two-way pattern is adapted so that the client
of the network and preserves decentralization by providing obtains coherent responses ranging from 2f + 1 to 3f instead
replication. Agents perform replication as a pillar that is of 3f +1. Since Zyzzyva is a state machine replication (SMR)
accountable for consensus. Performance evaluation of SBFT that operates on 3f + 1 replicas, it allows a client to collect
has demonstrated its resilience to the propagation of active 2f +1 coherent responses within a commit certificate that has
replicas [209], [230]. been spread out among replicas [213].
16) DELEGATED BYZENTINE FAULT TOLERANCE

12) STELLAR CONSENSUS PROTOCOL
DBFT has been introduced as an underlying consensus for
SCP is a global consensus that has been proposed as an
NEO. It is an extension to PBFT that does not impose any
extension to FBA. SCP employs the same concept where a
restrictions on the contribution of all nodes for appending
quorum is considered as a set of nodes that work towards
new blocks. It follows the same principles as DPoS where a
reaching consensus using quorum slices as a subset to procure
particular set of nodes are in charge of validating the trans-
agreement. Hence, SCP does not need to trust the whole
actions and generating new blocks. DBFT integrates a set
network and reaches consensus using the validator nodes.
of ordinary nodes along with bookkeepers: randomly chosen
However, to obtain a comprehensive consensus, it mainly
bookkeepers propose the next block and ordinary nodes vote
relies on quorum intersections. SCP is composed of a ballot
for them. The new block will be appended after preserving at
protocol to verify the agreement of nodes on a quorum and
least 66% agreement among bookkeepers [214], [231].
the nomination protocol that feeds the ballot [22], [210].
17) PROOF OF PREVIOUS TRANSACTION
13) PROOF OF QoS Proof of Previous Transaction has been proposed as a
PoQ is a consensus mechanism that relies on the quality of BFT-Compliant consensus mechanism for parallel account-
service. It considers a network as a collection of subregions, ing. It incorporates PBFT for the candidate selection proce-
where each region nominates a representative node based on dure. Using a consistent hash function, it can provide equity
its quality of service. Accordingly, a BFT algorithm is applied in the computing power of the participants resulting in impar-
to nominees to facilitate the election procedure. Evaluation of tial accounting opportunities and avoiding unbalanced incen-
PoQ indicates a significant enhancement of throughput while tive assignments. Parallel accounting allows more than one
preserving a fair environment for participants [211]. accountant to administer transactions, which results in a pos-
sibility of multiple blocks to be appended to the blockchain
14) EPLIS simultaneously. PoPT increases the scalability of JCledger
The motivation behind the advent of Eplis was moving using parallel accountants that confront large transactions
towards an accelerated decision within three communica- more efficiently [215].
tion threads by incorporating the locality of the geo-scale
implementations. Not only it can stabilize the geo-replication, 18) DEMOCRATIC BYZANTINE FAULT TOLERANCE
but it also ensures the linearizability of multishared pro- DBFT has been introduced as a deterministic, leader-
ceedings. Elpis associates ownership with individuals and less consensus mechanism to attenuate the drawbacks of
43644 VOLUME 9, 2021

FIGURE 7. Distribution plots.
classical leader-based algorithms and improve their resilience. and rectifying defective replicas. One of the most important
It enables cooperating nodes to agree upon an intrinsically features of BFT-Smart is its evolution over time since its
democratic decision in terms of execution of the consensus. introduction in 2007, based on feedback on the applicability
DBFT can operate in a partial synchronous manner, provide of this mechanism in transaction processing engines and
scalability, and ensure termination of the algorithm even in application-level firewalls [219].
the presence of a defective coordinator [216].
22) OUROBOROS-BFT
19) DBFT Ouroboros-BFT is a BFT compliant consensus mechanism
DBFT has been proposed as a leader-based BFT exten- proposed as an extension to the classic Ouroboros proto-
sion to address the limitation of formerly proposed consen- col. It is a deterministic protocol that incorporates a predeter-
suses using a novel double-response technique. This allows mined round-robin to broadcast transaction blocks. It shares
the replica nodes to reply simultaneously and eliminate the common characteristics with other BFT variants, such as
requirements for detection of any instability that may cause PBFT, in terms of incorporating passive clients. However,
a performance decrease during speculative implementation. unlike the other variants, Ouroboros-BFT provides instant
Performance evaluation of DBFT indicates the superiority of transaction verification and full network speed transaction
this algorithm among similar BFT solutions such as zyzzyva processing in the absence of faults [220].
and PBFT, specifically in the presence of Byzantine faults,
by preserving load balance, security, and liveness of the IV. ANALYSIS
network [217]. As discussed earlier, it is crucial for all networks within a
decentralized ledger platform to collectively agree upon the
20) DIVERSITY OF OPINION BYZANTINE FAULT TOLERANCE consensus regulations. This ensures the authenticity of the
DBFT has been implemented on a randomized mesh ongoing transactions by validating the contributions made to
blockchain (RMBC) to diversify the number of participants the blockchain. Since the advent of blockchain, there have
resembling PoW while preserving the resilience and reliabil- been a range of consensus algorithms proposed. Their diver-
ity of the network. Accordingly, DBFT employs a two-stage sity, in terms of communication model, adversity tolerance,
consensus procedure. During the first phase, a general BFT and several other factors, makes them applicable to a variety
is applied, considering a high possibility of the presence of scenarios, as discussed in Section III. The classification
of malicious nodes. The second agreement phase is applied introduced in this survey allows an effective analysis of the
to classified verifiers randomly chosen through RMBC. 130 reviewed consensus algorithms within the proposed tax-
To decrease the probability of collision and ensure the onomy and among the common blockchain applications.
integrity of the network, each transaction is executed only if
both consensus agreement stages coincide [218]. A. CONSENSUS DISTRIBUTION
As expected, most of the examined consensus mechanisms
21) BFT-SMaRt (42%) have been actively used in the context of cryptocurren-
BFT-SMaRt is a BFT variant, inspired by PBFT, providing cies. However, 26% of the examined consensus algorithms
reconfiguration support and modularity. It outperforms the are not associated with any particular application domain,
former BFT compliant algorithms, such as PBFT [39] and as shown in Fig. 7. Notably, in several studies that pro-
upright [232], in terms of performance, fault-free execution, pose alternatives or extensions to improve some deficiencies
VOLUME 9, 2021 43645

TABLE 9. Consensus in different blockchain platforms [234].
FIGURE 8. Distribution of proposed classification in BC applications.
of existing consensus approaches, no application domain PE (Primitive Compliant Extension), PCE (Proof Compliant
is specified. This either means that the proposed protocol Extension) and BCE (BFT Compliant Extension).
applies to the same application area as the precedent con- In this regard, BCE, PCE, and PA are more commonly used
sensus, or that it can be applied to any domain as long as to preserve consensus regardless of the application domain.
it satisfies its specific application requirements. As a con- This aligns with the results of consensuses distribution in the
sequence, Dapps, IoT, and cloud computing are the appli- proposed classification. In the top 10 blockchain platforms,
cation domains that incorporate consensus algorithms the BCE and PA are used more frequently than other consen-
most. On the contrary, at the time of writing, only 1% of suses. PCE have been incorporated in the top cryptocurrency
the reviewed consensuses have been used in the smart grid platforms to the same extent as PA. However, PA appears
and localization applications, making them the least common to be the dominant consensus amongst the top 10 supply
blockchain application domains. chain and health-care platforms. In addition, the top 10 supply
The analysis also reveals the distribution of the reviewed chain platforms do not employ BCE consensuses. This is an
consensus algorithms among the different classes of the pro- indication of scarcity of this class of consensus among supply
posed classification. As also shown in Fig. 7, 38% of the chain applications.
130 reviewed consensus algorithms belong to PA (pure alter-
natives). The next three classes with the largest representation B. ANALYSIS OF THE CONSENSUS DISTRIBUTION
are PCE (proof compliant extensions), BCE (BFT-compliant The reason behind the dominant adoption of certain classes
extensions), and PE (primitive complaint extensions). of consensus mechanisms (e.g. PCE, BCE, PA) in particular
Another interesting perspective is provided by analysing application areas can be further viewed from the perspective
the distribution of the proposed consensus classes among of the underlying application architecture. Robust consensus
the top 10 widely used BC platforms in general (regardless and block finality are significant requirements preceding the
of their application), cryptocurrencies (based on blockchain initiation of communication across chains.
activity matrix [233]), supply chain BC platforms, and health- BCE consensus mechanisms employ an exceptional final-
care BC platforms. The results of this analysis are summa- ity that executes high-valued chained transactions very fast
rized in Table 9 and illustrated in Fig. 8. This illustration not and protects blockchain from being forked. BCE algorithms
only shows the distribution of consensus mechanisms among are built within regulatory considerations that make them
a diverse range of platforms, but also demonstrates how cer- compatible with business use cases. For instance, health care
tain classes of algorithms have been designated for the devel- applications conduct altruistic considerations and interact
opment of specific applications in the context of distributed with genuine identities, unlike exceedingly anonymous and
ledgers. The category that each consensus may fall into is unregulated structures such as Bitcoin. Accordingly, avoiding
referred to as Primitive, PHA (Proof Compliant Hybrid Alter- PCE consensus mechanisms for this type of applications
native), BHA (BFT Compliant Hybrid Alternative), CHA outweighs the risks associated with BCE algorithms. Using
(Cross Compliant Hybrid Alternative), Pure Alternative (PA), BCE algorithms, the consensus decisions are determined
43646 VOLUME 9, 2021

conforming to all submitted decisions while eliminating the ments has not yet been fulfilled. Hence, a significant number
energy expenditures associated with PCE. Although BCE of consensus mechanisms are expected to emerge in the
consensus mechanisms can be applied exclusively to per- CHA class.
missioned blockchains due to lack of anonymity, their pre-
dominant superiority over other consensus mechanisms is V. CONCLUSION
the transaction finality that does not require the confirmation This paper provides a thorough review on precedent forms
procurement employed by PCE mechanisms. of distributed ledger focusing on blockchain and its con-
On the other hand, PCE consensus mechanisms are known sensus mechanisms. A total of 130 consensuses have been
to provide significant decentralization by refraining from reviewed analyzing 185 publications ranging from academic
Application-Specific Integrated Circuits (ASIC). This allows journals, industrial websites, conferences, and workshops
PCE variants to avoid the re-centralization subsequent to to technical white papers. To provide a suitable analyti-
decentralization, caused by rising barriers to obtaining min- cal framework, we propose a comprehensive classification
ing permissions. of consensus mechanisms based on their building blocks.
Pure Alternatives are also used where transaction speed Accordingly, a consensus mechanism that is not primitive like
and energy consumption are the priorities of the use case. This paxos, is either proof compliant, BFT compliant, primitive
class of consensus mechanisms can significantly reduce the compliant, or cross compliant. Consensus mechanisms in
power consumption and increase the transaction throughput the same category tend to share certain characteristics that
by restraining the computing power of the network. However, are discussed in terms of functionality, shortcomings, and
the consensus mechanisms of this class are still in their advantages.
infancy, promising to improve their resilience and mitigate The proposed classification not only facilitates the anal-
reliance on specialized hardware. ysis of existing consensus mechanisms, but also provides
a framework that subsequent algorithms can be related to.
C. EVOLUTION OF CONSENSUS MECHANISMS Unlike previous attempts for classification of the consensus
With the advent of distributed ledger technology in various mechanism, this approach relies on identifying the prevalent
application areas, the requirements for consensus protocols features that help to discern the building blocks and commu-
significantly raised, especially for protocols that are reliable nication model of each algorithm. Accordingly, a consensus
for both financial institutions and frameworks. This led to algorithm may belong to one of the following 8 classes:
the emergence of consensus mechanisms that do not rely Primitive, PHA (Proof Compliant Hybrid Altenrative), BHA
on bitcoins’ traditional proof of work. Resulting substantial (Bft Compliant Hybrid Alternative), CHA (Cross Compli-
breakthrough brought consensus protocols such as Ripple. ant hybrid Alternative),Pure Alternative (PA), PE (Primitive
It also triggered the emergence of consensus mechanisms Compliant Extention), PCE (Proof Compliant Extention) and
that migrated from permission-less systems to token-less BCE (Bft Compliant Extension).
permissioned blockchains not allowing anonymous nodes to This article also provides analysis on the distribution of the
participate in the verification of transactions. reviewed algorithms in the proposed categories. As expected,
The main concern of any distributed ledger technology is to pure alternatives (PA) and proof compliant extensions (PCE)
ensure the security of network transactions. Consensus pro- are the dominant categories. Most proposed consensuses are
tocol verifies all transactions have authorized source through either used in the context of cryptocurrencies or are not asso-
agreement on the state of the ledger. Therefore, several con- ciated with any specific application domain. The latter occurs
sensus protocols have been proposed with different levels when an alternative or an extension consensus is proposed to
of self-enforcing regulations and incentive mechanisms to improve the deficiencies of their predecessor and its applica-
ensure that participants act legitimately. In addition, the secu- tion area is not further discussed. This either means that the
rity system of the pioneer consensus protocols such as proof consensus is applicable to the same application domain as the
of work does not function effectively for use cases with strict precedent consensus or that it can be applied to any domain
financial regulations. Hence, based on the preferences of an as long as the requirements are satisfied.
organization at a given time span, different consensus mech- Finally, this survey evaluates the distribution of each pro-
anisms can be employed without enforcing a rigid consensus posed consensus group in the top 10 general blockchain,
layer. cryptocurrency, supply-chain, and health care platforms.
The results indicate that BCE (BFT Compliant Exten-
D. FUTURE PROSPECTS sion), PCE (Proof Compliant Extension), and Pure Alterna-
Since the advent of Hyperledger, the attention has been tives (PA) are most commonly used to reach consensus within
drawn towards cross-industry and open source distributed a network.
ledger solutions improving the cross-compliant hybrid alter-
native (CHA) solutions introduced in consensus section. REFERENCES
Although many providers prefer developing consensus solu-
[1] S. Hattab and I. F. T. Alyaseen, ‘‘Consensus algorithms blockchain:
tions based on specific use case requirements, the need for A comparative study,’’ Int. J. Perceptive Cognit. Comput., vol. 5, no. 2,
consensus mechanisms capable of addressing diverse require- pp. 66–71, Dec. 2019.
VOLUME 9, 2021 43647

[2] N. Chaudhry and M. M. Yousaf, ‘‘Consensus algorithms in blockchain: [33] I. Bentov, C. Lee, A. Mizrahi, and M. Rosenfeld, ‘‘Proof of activity:
Comparative analysis, challenges and opportunities,’’ in Proc. 12th Int. Extending bitcoin’s proof of work via proof of stake [extended abstract]
Conf. Open Source Syst. Technol. (ICOSST), Dec. 2018, pp. 54–63. Y,’’ ACM SIGMETRICS Perform. Eval. Rev., vol. 42, no. 3, pp. 34–37,
[3] S. J. Alsunaidi and F. A. Alhaidari, ‘‘A survey of consensus algorithms 2014.
for blockchain technology,’’ in Proc. Int. Conf. Comput. Inf. Sci. (ICCIS), [34] W. Wang, D. T. Hoang, P. Hu, Z. Xiong, D. Niyato, P. Wang, Y. Wen,
Apr. 2019, pp. 1–6. and D. I. Kim, ‘‘A survey on consensus mechanisms and mining
[4] Y. Xiao, N. Zhang, W. Lou, and Y. T. Hou, ‘‘A survey of distributed strategy management in blockchain networks,’’ IEEE Access, vol. 7,
consensus protocols for blockchain networks,’’ IEEE Commun. Surveys pp. 22328–22370, 2019.
Tuts., vol. 22, no. 2, pp. 1432–1465, 2nd Quart., 2020. [35] (2020). Lynx. [Online]. Available: http://cdn.getlynx.io/2019-03-
[5] (2020). ITU-T. [Online]. Available: https://www.itu.int/en/ITU- 17_Lynx_Whitepaper_v1.1.pdf
T/focusgroups/dlt/Pages/default.aspx [36] (2020). Lynx. [Online]. Available: https://getlynx.io/
[6] (2020). ITU-T. [Online]. Available: https://www.itu.int/en/ITU- [37] H. Jennath and S. Asharaf, ‘‘Survey on blockchain consensus strategies,’’
T/focusgroups/dlt/Documents/d21.pdf in ICDSMLA 2019. Singapore: Springer, 2020, pp. 637–654.
[7] M. Andoni, V. Robu, D. Flynn, S. Abram, D. Geach, D. Jenkins, [38] (2020). LFT. [Online]. Available: http://docs.icon.foundation/en/white
P. McCallum, and A. Peacock, ‘‘Blockchain technology in the energy paper/_static/LFT.pdf
sector: A systematic review of challenges and opportunities,’’ Renew. [39] M. Castro and B. Liskov, ‘‘Practical Byzantine fault tolerance,’’ in Proc.
Sustain. Energy Rev., vol. 100, pp. 143–174, Feb. 2019. OSDI, vol. 99, 1999, pp. 173–186.
[8] R. C. Merkle, ‘‘A certified digital signature,’’ in Proc. Conf. Theory Appl. [40] S. Duan, S. Peisert, and K. N. Levitt, ‘‘HBFT: Speculative Byzantine fault
Cryptol. New York, NY, USA: Springer, 1989, pp. 218–238. tolerance with minimum cost,’’ IEEE Trans. Dependable Secure Comput.,
[9] H. Pervez, M. Muneeb, M. U. Irfan, and I. U. Haq, ‘‘A comparative vol. 12, no. 1, pp. 58–70, Jan. 2015.
analysis of DAG-based blockchain architectures,’’ in Proc. 12th Int. Conf. [41] E. Buchman, ‘‘Tendermint: Byzantine fault tolerance in the age of
Open Source Syst. Technol. (ICOSST), Dec. 2018, pp. 27–34. blockchains,’’ Ph.D. dissertation, School Eng., Univ. Guelph, Guelph,
[10] S. Yang, Z. Chen, L. Cui, M. Xu, Z. Ming, and K. Xu, ‘‘CoDAG: ON, Canada, 2016.
An efficient and compacted DAG-based blockchain protocol,’’ in Proc. [42] (2020). LFT. [Online]. Available: https://medium.com/@2infiniti/a-
IEEE Int. Conf. Blockchain (Blockchain), Jul. 2019, pp. 314–318. primer-to-lft-loop-fault-tolerance-consensus-algorithm-d692bdece85a
[11] H. Anwar. (2020). Distributed Ledger Technology: Where Technologi- [43] L. Ismail and H. Materwala, ‘‘A review of blockchain architecture and
cal Revolution Starts. [Online]. Available: https://101blockchains.com/ consensus protocols: Use cases, challenges, and solutions,’’ Symmetry,
distributed-ledger-technology-dlt/#3 vol. 11, no. 10, p. 1198, Sep. 2019.
[12] N. El Ioini and C. Pahl, ‘‘A review of distributed ledger technologies,’’ [44] V. Buterin and V. Griffith, ‘‘Casper the friendly finality gadget,’’ 2017,
in Proc. OTM Confederated Int. Conf. Move Meaningful Internet Syst. arXiv:1710.09437. [Online]. Available: http://arxiv.org/abs/1710.09437
Valletta, Malta: Springer, 2018, pp. 277–288.
[45] C. Copeland and H. Zhong, ‘‘Tangaroa: A Byzantine fault tolerant raft,’’
[13] (2020). Monax. [Online]. Available: https://monax.io/
Stanford Secure Comput. Syst. Group, Stanford, CA, USA, Tech. Rep.,
[14] R. T. Frahat, M. M. Monowar, and S. M. Buhari, ‘‘Secure and scalable
2016.
trust management model for IoT P2P network,’’ in Proc. 2nd Int. Conf.
[46] C. Decker, J. Seidel, and R. Wattenhofer, ‘‘Bitcoin meets strong con-
Comput. Appl. Inf. Secur. (ICCAIS), May 2019, pp. 1–6.
sistency,’’ in Proc. 17th Int. Conf. Distrib. Comput. Netw., Jan. 2016,
[15] A. Baliga, ‘‘Understanding blockchain consensus models,’’ Persistent,
pp. 1–10.
vol. 2017, no. 4, pp. 1–14, 2017.
[47] (2020). VBFT. [Online]. Available: https://dev-docs.ont.io/#/docs-en/
[16] L. Lamport, ‘‘Paxos made simple,’’ ACM SIGACT News, vol. 32, no. 4,
DeveloperGuide/02-VBFT-introduction
pp. 18–25, 2001.
[48] F. Yang, W. Zhou, Q. Wu, R. Long, N. N. Xiong, and M. Zhou, ‘‘Dele-
[17] (2020). Clamcoin. [Online]. Available: https://clamcoin.org/#
gated proof of stake with downgrade: A secure and efficient blockchain
[18] (2020). VeriCoin. [Online]. Available: https://vericonomy.ams3.cdn.
consensus algorithm with downgrade mechanism,’’ IEEE Access, vol. 7,
digitaloceanspaces.com/documents/VeriCoin-Proof-of-Stake-Time-
pp. 118541–118555, 2019.
Whitepaper.pdf
[49] (2020). Credits. [Online]. Available: https://credits.com/
[19] (2020). VeriCoin. [Online]. Available: https://wiki.vericoin.info/
index.php?title=Proof-of-Work-Time [50] (2020). DPoS. [Online]. Available: https://medium.com/eosio/dpos-bft-
[20] J. Benet and N. Greco, ‘‘Filecoin: A decentralized storage network,’’ pipelined-byzantine-fault-tolerance-8a0634a270ba
Protoc. Labs, pp. 1–36, Jul. 2017. [51] B. Podgorelec, V. Kersic, and M. Turkanovic, ‘‘Analysis of fault tolerance
[21] (2020). Nebulas. [Online]. Available: https://nebulas.io/docs/ in permissioned blockchain networks,’’ in Proc. 27th Int. Conf. Inf.,
NebulasTechnicalWhitepaper.pdf Commun. Autom. Technol. (ICAT), Oct. 2019, pp. 1–6.
[22] L. S. Sankar, M. Sindhu, and M. Sethumadhavan, ‘‘Survey of consensus [52] (2020). VBFT. [Online]. Available: https://ontio.github.io/documentation/
protocols on blockchain applications,’’ in Proc. 4th Int. Conf. Adv. Com- vbft_intro_en.html
put. Commun. Syst. (ICACCS), Jan. 2017, pp. 1–5. [53] Y. Hassanzadeh-Nazarabadi, A. Küpçü, and Ö. Özkasap, ‘‘LightChain:
[23] I. Bentov, R. Pass, and E. Shi, ‘‘Snow white: Provably secure proofs of A DHT-based blockchain for resource constrained environments,’’ 2019,
stake,’’ IACR Cryptol. ePrint Arch., vol. 2016, no. 919, 2016. arXiv:1904.00375. [Online]. Available: http://arxiv.org/abs/1904.00375
[24] A. Shahaab, B. Lidgey, C. Hewage, and I. Khan, ‘‘Applicability and [54] F. P. Junqueira, B. C. Reed, and M. Serafini, ‘‘Zab: High-performance
appropriateness of distributed ledgers consensus protocols in pub- broadcast for primary-backup systems,’’ in Proc. IEEE/IFIP 41st Int.
lic and private sectors: A systematic review,’’ IEEE Access, vol. 7, Conf. Dependable Syst. Netw. (DSN), Jun. 2011, pp. 245–256.
pp. 43622–43636, 2019. [55] S. Nakamoto. (2017). Bitcoin: A Peer-to-Peer Electronic Cash System,
[25] Z. Liu, S. Tang, S. S. M. Chow, Z. Liu, and Y. Long, ‘‘Fork-free hybrid Oct. 2008. [Online]. Available: http://www.bitcoin.org/bitcoin.pdf
consensus with flexible Proof-of-Activity,’’ Future Gener. Comput. Syst., [56] M. Milutinovic, W. He, H. Wu, and M. Kanwal, ‘‘Proof of luck:
vol. 96, pp. 515–524, Jul. 2019. An efficient blockchain consensus protocol,’’ in Proc. 1st Workshop Syst.
[26] L. Ren, ‘‘Proof of stake velocity: Building the social currency of the Softw. Trusted Execution, Dec. 2016, pp. 1–6.
digital age,’’ Reddcoin, New York, NY, USA, White Paper, 2014. [57] S. King and S. Nadal, ‘‘Ppcoin: Peer-to-peer crypto-currency with proof-
[27] E. Wustrow and B. VanderSloot, ‘‘DDoSCoin: Cryptocurrency with a of-stake,’’ Chainwhy Group, Tech. Rep., 2012.
malicious proof-of-work,’’ in Proc. 10th USENIX Workshop Offensive [58] F. Tschorsch and B. Scheuermann, ‘‘Bitcoin and beyond: A technical sur-
Technol., 2016, pp. 1–10. vey on decentralized digital currencies,’’ IEEE Commun. Surveys Tuts.,
[28] (2020). VeroCoin. [Online]. Available: https://vericoin.info/vericoin- vol. 18, no. 3, pp. 2084–2123, 3rd Quart., 2016.
digital-currency/ [59] (2020). PoB. [Online]. Available: https://tokens-economy.gitbook.
[29] (2020). VeriCoin. [Online]. Available: https://vericoin.info/ io/consensus/chain-based-proof-of-burn/proof-of-time
[30] T. Moran and I. Orlov, ‘‘Simple proofs of space-time and rational proofs [60] G. Ateniese, I. Bonacina, A. Faonio, and N. Galesi, ‘‘Proofs of space:
of storage,’’ in Proc. Annu. Int. Cryptol. Conf. Santa Barbara, CA, USA: When space is of the essence,’’ in Proc. Int. Conf. Secur. Cryptogr. Netw.
Springer, 2019, pp. 381–409. Amalfi, Italy: Springer, 2014, pp. 538–557.
[31] (2020). Filecoin. [Online]. Available: https://filecoin.io/ [61] (2020). PoE. [Online]. Available: http://docs.proofofexistence.com/#/?
[32] (2020). Nebulas. [Online]. Available: https://nebulas.io/ id=proof-of-existence
43648 VOLUME 9, 2021

[62] Y. Yuan and F.-Y. Wang, ‘‘Towards blockchain-based intelligent trans- [87] M. Torky, E. Nabil, and W. Said, ‘‘Proof of credibility: A blockchain
portation systems,’’ in Proc. IEEE 19th Int. Conf. Intell. Transp. Syst. approach for detecting and blocking fake news in social networks,’’ Int.
(ITSC), Nov. 2016, pp. 2663–2668. J. Adv. Comput. Sci. Appl., vol. 10, no. 12, pp. 321–327, 2019.
[63] O. Samuel, N. Javaid, M. Awais, Z. Ahmed, M. Imran, and M. Guizani, [88] A. Yakovenko, ‘‘Solana: A new architecture for a high performance
‘‘A blockchain model for fair data sharing in deregulated smart grids,’’ in blockchain,’’ Solana Labs, Tech. Rep., 2018.
Proc. IEEE Global Commun. Conf. (GLOBECOM), Dec. 2019, pp. 1–7. [89] P. Siano, G. De Marco, A. Rolan, and V. Loia, ‘‘A survey and evaluation of
[64] S. Takahashi, ‘‘Proof-of-approval: A distributed consensus protocol for the potentials of distributed ledger technology for peer-to-peer transactive
blockchains,’’ Tech Rep Group, Tech. Rep., 2018. energy exchanges in local energy markets,’’ IEEE Syst. J., vol. 13, no. 3,
[65] M.-L. Marsal-Llacuna and M. Oliver-Riera, ‘‘The standards revolution: pp. 3454–3466, Sep. 2019.
Who will first put this new kid on the blockchain?’’ in Proc. ITU Kalei- [90] E. Pournaras, ‘‘Proof of witness presence: Blockchain consensus for aug-
doscope, Challenges Data-Driven Soc. (ITU K), Nov. 2017, pp. 1–7. mented democracy in smart cities,’’ 2019, arXiv:1907.00498. [Online].
[66] M. Swan, ‘‘Blockchain thinking: The brain as a decentralized autonomous Available: http://arxiv.org/abs/1907.00498
corporation [commentary],’’ IEEE Technol. Soc. Mag., vol. 34, no. 4, [91] I. Eyal, A. E. Gencer, E. G. Sirer, and R. Van Renesse, ‘‘Bitcoin-Ng:
pp. 41–52, Dec. 2015. A scalable blockchain protocol,’’ in Proc. 13th USENIX Symp. Netw. Syst.
[67] X. Han, Y. Yuan, and F.-Y. Wang, ‘‘A fair blockchain based on proof of Design Implement., 2016, pp. 45–59.
credit,’’ IEEE Trans. Comput. Social Syst., vol. 6, no. 5, pp. 922–931, [92] (2020). DigixGlobal. [Online]. Available: https://github.com/
Oct. 2019. DigixGlobal/digix-press-kit/blob/master/digix-whitepaper.pdf
[68] H. Y. Yuen, F. Wu, W. Cai, H. C. B. Chan, Q. Yan, and V. C. M. Leung, [93] (2017). Internet of Services: The Next-Generation, Secure, Highly
‘‘Proof-of-play: A novel consensus model for blockchain-based peer-to- Scalable Ecosystem for Online Services. [Online]. Available:
peer gaming system,’’ in Proc. ACM Int. Symp. Blockchain Secure Crit. https://tokeninsight.com/api/upload/whitePaper/InternetofServices_en.
Infrastruct. (BSCI), 2019, pp. 19–28. pdf
[69] (2020). Nexty. [Online]. Available: https://nexty.io/nexty-whitepaper.pdf [94] J. Zou, B. Ye, L. Qu, Y. Wang, M. A. Orgun, and L. Li, ‘‘A proof-of-
[70] Flash White Paper. Accessed: 2016. [Online]. Available: https://www. trust consensus protocol for enhancing accountability in crowdsourcing
flashcoin.io/docs/FLASHWhitepaper.pdf services,’’ IEEE Trans. Services Comput., vol. 12, no. 3, pp. 429–445,
[71] (Jul. 2018). The Proof-of-Cooperation Blockchain Faircoin. May 2019.
[Online]. Available: https://fair-coin.org/sites/default/files/FairCoin2_ [95] (2020). Zookeeper. [Online]. Available: https://zookeeper.apache.org/
whitepaper_V1.2.pdf [96] M. Sabt, M. Achemlal, and A. Bouabdallah, ‘‘Trusted execution envi-
[72] Skycoin. Accessed: 2018. [Online]. Available: https://downloads. ronment: What it is, and what it is not,’’ in Proc. IEEE Trust-
skycoin.com/whitepapers/Skycoin-Whitepaper-v1.2.pdf com/BigDataSE/ISPA, vol. 1, Aug. 2015, pp. 57–64.
[73] A. Pazaitis, P. D. Filippi, and V. Kostakis, ‘‘Blockchain and value systems [97] (2020). Genesis. [Online]. Available: https://genesisx.network/GenesisX-
in the sharing economy: The illustrative case of backfeed,’’ Technolog. WP-v2.0.pdf
Forecasting Social Change, vol. 125, pp. 105–115, Dec. 2017. [98] K. Karantias, A. Kiayias, and D. Zindros, ‘‘Proof-of-burn,’’ in Proc. Int.
[74] (2020). PoD. [Online]. Available: https://b3coin.io/faq/proof- Conf. Financial Cryptogr. Data Secur., 2019, pp. 523–540.
disintegration-pod-explained/ [99] (2020). Chronologic. [Online]. Available: https://chronologic.network/
[75] F. Bravo-Marquez, S. Reeves, and M. Ugarte, ‘‘Proof-of-learning: [100] E. Bresson, A. Canteaut, B. Chevallier-Mames, C. Clavier, T. Fuhr,
A blockchain consensus mechanism based on machine learning com- A. Gouget, I. Icart, J. F. Misarsky, M. Naya-Plasencia, P. Paillier, and
petitions,’’ in Proc. IEEE Int. Conf. Decentralized Appl. Infrastruct. T. Pornin, ‘‘Shabal, a submission to NIST’s cryptographic hash algo-
(DAPPCON), Apr. 2019, pp. 119–124. rithm competition,’’ NIST, SAPHIR Project, SHABAL Group, France,
[76] H. Liu, C.-W. Lin, E. Kang, S. Shiraishi, and D. M. Blough, Tech. Rep., 2008.
‘‘A Byzantine-tolerant distributed consensus algorithm for connected [101] (2020). CoinSpace. [Online]. Available: https://coinspace.com/
vehicles using proof-of-eligibility,’’ in Proc. 22nd Int. ACM Conf. Model., [102] (2020). Chia. [Online]. Available: https://www.chia.net/
Anal. Simulation Wireless Mobile Syst. (MSWIM), 2019, pp. 225–234. [103] (2020). BurstCoin. [Online]. Available: https://www.burst-coin.org/
[77] F. Gai, B. Wang, W. Deng, and W. Peng, ‘‘Proof of reputation: [104] R. Gennaro and M. Robshaw, Advances in Cryptology–CRYPTO 2015,
A reputation-based consensus protocol for peer-to-peer network,’’ in vol. 9216. Santa Barbara, CA, USA: Springer, 2015.
Proc. Int. Conf. Database Syst. Adv. Appl. Gold Coast, QLD, Australia: [105] Y. Zhang, S. Wu, B. Jin, and J. Du, ‘‘A blockchain-based process prove-
Springer, 2018, pp. 666–681. nance for cloud forensics,’’ in Proc. 3rd IEEE Int. Conf. Comput. Com-
[78] K. Li, H. Li, H. Hou, K. Li, and Y. Chen, ‘‘Proof of vote: A high- mun. (ICCC), Dec. 2017, pp. 2470–2473.
performance consensus protocol based on vote mechanism & consortium [106] (2020). Factom. [Online]. Available: https://www.factom.com/
blockchain,’’ in Proc. IEEE 19th Int. Conf. High Perform. Comput. Com- [107] (2020). Lazooz. [Online]. Available: http://lazooz.org/
munications; IEEE 15th Int. Conf. Smart City; IEEE 3rd Int. Conf. Data [108] Y. Yuan and F.-Y. Wang, ‘‘Blockchain and cryptocurrencies: Model,
Sci. Syst. (HPCC/SmartCity/DSS), Dec. 2017, pp. 466–473. techniques, and applications,’’ IEEE Trans. Syst., Man, Cybern. Syst.,
[79] X. Fu, H. Wang, P. Shi, and H. Mi, ‘‘PoPF: A consensus algorithm for vol. 48, no. 9, pp. 1421–1428, Sep. 2018.
JCLedger,’’ in Proc. IEEE Symp. Service-Oriented Syst. Eng. (SOSE), [109] (2020). PoA. [Online]. Available: https://www.poa.network/for-users/
Mar. 2018, pp. 204–209. whitepaper/poadao-v1/proof-of-authority
[80] (2020). PoI. [Online]. Available: https://github.com/Bit-Nation/Proof-of- [110] S. D. Angelis, L. Aniello, R. Baldoni, F. Lombardi, A. Margheri, and
Individuality-POI V. Sassone, ‘‘PBFT vs proof-of-authority: Applying the CAP theorem
[81] M. Borge, E. Kokoris-Kogias, P. Jovanovic, L. Gasser, N. Gailly, and to permissioned blockchain,’’ in Proc. Italian Conf. Cyber Secur., 2018,
B. Ford, ‘‘Proof-of-personhood: Redemocratizing permissionless cryp- pp. 1–11.
tocurrencies,’’ in Proc. IEEE Eur. Symp. Secur. Privacy Workshops (EuroS [111] (2020). Nuls. [Online]. Available: https://www.nuls.io/
PW), Apr. 2017, pp. 23–26. [112] (2020). Nuls. [Online]. Available: https://www.nuls.io/wp-content/
[82] C. Cachin, S. Schubert, and M. Vukolić, ‘‘Non-determinism in Byzantine uploads/2019/06/NULS_Whitepaper_2.0.pdf
fault-tolerant replication,’’ 2016, arXiv:1603.07351. [Online]. Available: [113] (2020). Nexty. [Online]. Available: https://nexty.io/
http://arxiv.org/abs/1603.07351 [114] (2020). Flashcoin. [Online]. Available: https://www.flashcoin.io/
[83] C. T. Nguyen, D. T. Hoang, D. N. Nguyen, D. Niyato, H. T. Nguyen, [115] (2020). Faircoin. [Online]. Available: https://fair-coin.org/
and E. Dutkiewicz, ‘‘Proof-of-stake consensus mechanisms for future [116] (2020). Faircoin. [Online]. Available: https://fair-coin.org/sites/default/
blockchain networks: Fundamentals, applications and opportunities,’’ files/FairCoin2_whitepaper_V1.2.pdf
IEEE Access, vol. 7, pp. 85727–85745, 2019. [117] (2020). Skycoin. [Online]. Available: https://www.skycoin.com/
[84] L. Chen, L. Xu, N. Shah, Z. Gao, Y. Lu, and W. Shi, ‘‘On security analysis [118] (2020). B3coin. [Online]. Available: https://b3coin.io/
of proof-of-elapsed-time (poet),’’ in Proc. Int. Symp. Stabilization, Saf., [119] (2020). Recaptcha. [Online]. Available: https://www.google.com/
Secur. Distrib. Syst. Boston, MA, USA: Springer, 2017, pp. 282–297. recaptcha/intro/v3.html
[85] D. Schwartz, N. Youngs, and A. Britto, ‘‘The ripple protocol consensus [120] Q. Zhuang, Y. Liu, L. Chen, and Z. Ai, ‘‘Proof of reputation: A reputation-
algorithm,’’ Ripple Labs, San Francisco, CA, USA, White Paper 8, 2014, based consensus protocol for blockchain based systems,’’ in Proc. Int.
vol. 5. Electron. Commun. Conf., Jul. 2019, pp. 131–138.
[86] S. Migliorini, ‘‘Enhancing blockchain smart-contracts with proof-of- [121] (2020). Foam. [Online]. Available: https://foam.space/publicAssets/
location,’’ in Proc. 10th Int. Conf. Geographic Inf. Sci., 2018. FOAM_Whitepaper.pdf
VOLUME 9, 2021 43649

[122] (2020). Foam. [Online]. Available: https://www.foam.space/ [153] (2020). VeriBlock. [Online]. Available: https://mirror1.veriblock.org/
[123] (2020). Solana. [Online]. Available: https://solana.com/ Proof-of-Proof_and_VeriBlock_Blockchain_Protocol_Consensus_
[124] (2020). POI. [Online]. Available: https://github.com/proofofindividuality/ Algorithm_and_Economic_Incentivization_v1.0.pdf
poi/blob/master/whitepaper.pdf [154] J. Lao, ‘‘A network-dependent rewarding system: Proof-of-mining,’’
[125] V. Gramoli, ‘‘From blockchain consensus back to Byzantine consensus,’’ 2014, arXiv:1409.7948. [Online]. Available: http://arxiv.org/abs/1409.
Future Gener. Comput. Syst., vol. 107, pp. 760–769, Jun. 2020. 7948
[126] L. M. Bach, B. Mihaljevic, and M. Zagar, ‘‘Comparative analysis of [155] (2020). Proof of Work Mining. [Online]. Available: https://www.m-
blockchain consensus algorithms,’’ in Proc. 41st Int. Conv. Inf. Commun. core.org/resources/mining.html
Technol., Electron. Microelectron. (MIPRO), May 2018, pp. 1545–1550. [156] J. Benet, D. Dalrymple, and N. Greco, ‘‘Proof of replication,’’ Protocol
[127] C. Cachin and M. Vukolic, ‘‘Blockchain consensus protocols in Labs, vol. 27, p. 20, Jul. 2017.
the wild,’’ 2017, arXiv:1707.01873. [Online]. Available: http://arxiv. [157] (2020). DPoS. [Online]. Available: https://bitshares.org/technology/
org/abs/1707.01873 delegated-proof-of-stake-consensus/
[128] F. Armknecht, G. O. Karame, A. Mandal, F. Youssef, and E. Zenner, [158] (2020). Komodo. [Online]. Available: https://docs.komodoplatform.com/
‘‘Ripple: Overview and outlook,’’ in Proc. Int. Conf. Trust Trustworthy whitepaper/introduction.html
Comput. Heraklion, Greece: Springer, 2015, pp. 163–180. [159] (2020). Bitshares History: Transactions as Proof-of-Stake
[129] (2020). DigixGlobal. [Online]. Available: https://digix.global/ (TAPOS)—Steemit. [Online]. Available: https://steemit.com/bitshares/
whitepaper.pdf#/ @testz/bitshares-history-transactions-as-proof-of-stake-tapos
[130] L. Lamport, ‘‘Fast Paxos,’’ Distrib. Comput., vol. 19, no. 2, pp. 79–103, [160] S. King, ‘‘Primecoin: Cryptocurrency with prime number proof-of-
2006. work,’’ Tech. Rep., vol. 1, no. 6, 2013.
[131] J.-P. Martin and L. Alvisi, ‘‘Fast Byzantine consensus,’’ IEEE Trans. [161] K. D. Bowers, A. Juels, and A. Oprea, ‘‘Proofs of retrievability: Theory
Dependable Secure Comput., vol. 3, no. 3, pp. 202–215, Jul. 2006. and implementation,’’ in Proc. ACM Workshop Cloud Comput. Secur.
[132] D. Ongaro and J. Ousterhout, ‘‘In search of an understandable consensus (CCSW), 2009, pp. 43–54.
algorithm,’’ in Proc. USENIX Annu. Tech. Conf., 2014, pp. 305–319. [162] (2020). Multichain. [Online]. Available: https://www.multichain.com/
[133] S. Liu, P. Viotti, C. Cachin, V. Quéma, and M. Vukolić, ‘‘FT: Practical download/MultiChain-White-Paper.pdf
fault tolerance beyond crashes,’’ in Proc. 12th Symp. Operating Syst. [163] M. Moniruzzaman, A. Yassine, and R. Benlamri, ‘‘Blockchain-based
Design Implement., 2016, pp. 485–500. mechanisms for local energy trading in smart grids,’’ in Proc. IEEE 16th
[134] P. J. Marandi, M. Primi, N. Schiper, and F. Pedone, ‘‘Ring Paxos: A high- Int. Conf. Smart Cities, Improving Qual. Life ICT IoT AI (HONET-ICT),
throughput atomic broadcast protocol,’’ in Proc. IEEE/IFIP Int. Conf. Oct. 2019, pp. 110–114.
Dependable Syst. Netw. (DSN), Jun. 2010, pp. 527–536. [164] (2020). LPoS. [Online]. Available: https://docs.wavesprotocol.org/en/
[135] P. J. Marandi, M. Primi, and F. Pedone, ‘‘Multi-ring Paxos,’’ in blockchain/leasing#leasing-benefits-for-the-node-owner
Proc. IEEE/IFIP Int. Conf. Dependable Syst. Netw. (DSN), Jun. 2012, [165] S. Andreina, J.-M. Bohli, G. O. Karame, W. Li, and G. A. Marson,
pp. 1–12. ‘‘Pots—A secure proof of TEE-stake for permissionless blockchains,’’
[136] S. Peluso, A. Turcu, R. Palmieri, G. Losa, and B. Ravindran, ‘‘Making IACR Cryptol. ePrint Arch., vol. 2018, p. 1135, Nov. 2018.
fast consensus generally faster,’’ in Proc. 46th Annu. IEEE/IFIP Int. Conf. [166] S. Biswas, K. Sharif, F. Li, S. Maharjan, S. P. Mohanty, and Y. Wang,
Dependable Syst. Netw. (DSN), Jun. 2016, pp. 156–167. ‘‘PoBT: A lightweight consensus algorithm for scalable IoT business
[137] I. Moraru, D. G. Andersen, and M. Kaminsky, ‘‘There is more consensus blockchain,’’ IEEE Internet Things J., vol. 7, no. 3, pp. 2343–2355,
in egalitarian parliaments,’’ in Proc. 24th ACM Symp. Operating Syst. Mar. 2020.
Princ., Nov. 2013, pp. 358–372. [167] Y. Gilad, R. Hemo, S. Micali, G. Vlachos, and N. Zeldovich, ‘‘Algorand:
[138] A. Ailijiang, A. Charapko, M. Demirbas, and T. Kosar, ‘‘WPaxos: Wide Scaling Byzantine agreements for cryptocurrencies,’’ in Proc. 26th Symp.
area network flexible consensus,’’ IEEE Trans. Parallel Distrib. Syst., Operating Syst. Princ., Oct. 2017, pp. 51–68.
vol. 31, no. 1, pp. 211–223, Jan. 2020. [168] T. Do, T. Nguyen, and H. Pham, ‘‘Delegated proof of reputation: A novel
[139] H. Howard, D. Malkhi, and A. Spiegelman, ‘‘Flexible Paxos: Quorum blockchain consensus,’’ in Proc. Int. Electron. Commun. Conf., Jul. 2019,
intersection revisited,’’ 2016, arXiv:1608.06696. [Online]. Available: pp. 90–98.
http://arxiv.org/abs/1608.06696 [169] D. Puthal, S. P. Mohanty, P. Nanda, E. Kougianos, and G. Das, ‘‘Proof-
[140] L. Lamport and M. Massa, ‘‘Cheap Paxos,’’ in Proc. Int. Conf. Depend- of-Authentication for scalable blockchain in resource-constrained dis-
able Syst. Netw., 2004, pp. 307–314. tributed systems,’’ in Proc. IEEE Int. Conf. Consum. Electron. (ICCE),
[141] F. Junqueira, Y. Mao, and K. Marzullo, ‘‘Classic Paxos vs. Fast Paxos: Jan. 2019, pp. 1–5.
Caveat emptor,’’ Proc. USENIX Hot Topics Syst. Dependability (HotDep), [170] A. Kiayias, A. Russell, B. David, and R. Oliynykov, ‘‘Ouroboros:
2007. A provably secure proof-of-stake blockchain protocol,’’ in Proc.
[142] (2020). FastPaxos. [Online]. Available: https://www.microsoft.com/en- Annu. Int. Cryptol. Conf. Santa Barbara, CA, USA: Springer, 2017,
us/research/wp-content/uploads/2016/02/tr-2005-112.pdf pp. 357–388.
[143] A. Chepurnoy, ‘‘Interactive proof-of-stake,’’ 2016, arXiv:1601.00275. [171] B. David, P. Gaži, A. Kiayias, and A. Russell, ‘‘Ouroboros praos:
[Online]. Available: http://arxiv.org/abs/1601.00275 An adaptively-secure, semi-synchronous proof-of-stake blockchain,’’ in
[144] (2020). Fiii. [Online]. Available: https://fiii.io/images/doc/whitepaper.pdf Proc. Annu. Int. Conf. Theory Appl. Cryptograph. Techn. Tel Aviv-Yafo,
[145] K. He, J. Chen, R. Du, Q. Wu, G. Xue, and X. Zhang, ‘‘DeyPoS: Israel: Springer, 2018, pp. 66–98.
Deduplicatable dynamic proof of storage for multi-user environments,’’ [172] C. Badertscher, P. Gaži, A. Kiayias, A. Russell, and V. Zikas, ‘‘Ouroboros
IEEE Trans. Comput., vol. 65, no. 12, pp. 3631–3645, Dec. 2016. genesis: Composable proof-of-stake blockchains with dynamic availabil-
[146] (2020). PoSBoo. [Online]. Available: https://tokens-economy.gitbook.io/ ity,’’ in Proc. ACM SIGSAC Conf. Comput. Commun. Secur., Oct. 2018,
consensus/chain-based-proof-of-stake/proof-of-stake-boo-pos-boo pp. 913–930.
[147] A. Shoker, ‘‘Sustainable blockchain through proof of exercise,’’ in Proc. [173] T. Kerber, A. Kiayias, M. Kohlweiss, and V. Zikas, ‘‘Ouroboros crypsi-
IEEE 16th Int. Symp. Netw. Comput. Appl. (NCA), Oct. 2017, pp. 1–9. nous: Privacy-preserving Proof-of-Stake,’’ in Proc. IEEE Symp. Secur.
[148] Z. Dong, Y. C. Lee, and A. Y. Zomaya, ‘‘Proofware: Proof of use- Privacy (SP), May 2019, pp. 157–174.
ful work blockchain consensus protocol for decentralized applica- [174] (2020). Fiiicoin. [Online]. Available: https://web.archive.org/web/
tions,’’ 2019, arXiv:1903.09276. [Online]. Available: http://arxiv.org/abs/ 20190402135904/https://www.fiii.io/images/doc/fiiicoin.yellowpaper.
1903.09276 v01.pdf
[149] A. Begicheva and A. Kofman, ‘‘Fair proof of stake,’’ Tech. Rep., 2018. [175] (2020). DPoC. [Online]. Available: https://sylvester-lee.blogspot.com/
[150] (Jul. 2018). Stake Net Whitepaper. Accessed: Jul. 14, 2018. [Online]. 2018/07/delegate-proof-of-capacity.html
Available: https://stakenet.io/Whitepaper_Stakenet_V3.0_EN.pdf [176] (2020). Shield. [Online]. Available: https://whitepaper.io/document/296/
[151] (2020). Applicability and Appropriateness of Distributed Ledgers shield-1-whitepaper
Consensus Protocols in Public and Private Sectors: A Systematic Review. [177] (2020). Stakenet. [Online]. Available: https://stakenet.io/
[Online]. Available: https://www.researchgate.net/publication/331942 [178] (2020). Stakenet. [Online]. Available: https://stakenet.io/Stakenet_
292_Applicability_and_Appropriateness_of_Distributed_Ledgers_ Whitepaper.pdf
Consensus_Protocols_in_Public_and_Private_Sectors_A_ [179] (2020). IZZZ. [Online]. Available: https://izzz.io/en/
Systematic_Review [180] (2020). LCPoA. [Online]. Available: https://docs.google.com/document/
[152] S. Sharkey and H. Tewari, ‘‘Alt-PoW: An alternative proof-of-work d/11ibTJ2-88r-vo64mLvAgKxsJ2vO3fR9sUre2ZPPrWwA/edit
mechanism,’’ in Proc. IEEE Int. Conf. Decentralized Appl. Infrastruct. [181] (2020). LCPoA. [Online]. Available: https://medium.com/@izzzio/lcpoa-
(DAPPCON), Apr. 2019, pp. 11–18. universal-as-pow-economical-as-pos-c26f6ba90017
43650 VOLUME 9, 2021

[182] (2020). Veriblock. [Online]. Available: https://www.veriblock.com/ [216] T. Crain, V. Gramoli, M. Larrea, and M. Raynal, ‘‘DBFT: Efficient
[183] (Aug. 2018). Casinocoin. [Online]. Available: https://casinocoin. leaderless Byzantine consensus and its application to blockchains,’’ in
org/consensus-the-core-of-blockchain-technology/ Proc. IEEE 17th Int. Symp. Netw. Comput. Appl. (NCA), Nov. 2018,
[184] (2020). Casinocoin. [Online]. Available: https://casinocoin.org/ pp. 1–8.
[185] (2020). DPoS. [Online]. Available: https://how.bitshares.works/en/ [217] J. Zhang, Y. Rong, J. Cao, C. Rong, J. Bian, and W. Wu, ‘‘DBFT:
master/technology/dpos.html A Byzantine fault tolerant protocol with graceful performance degra-
[186] (2020). Bitshares. [Online]. Available: https://bitshares.org/ dation,’’ in Proc. 38th Symp. Reliable Distrib. Syst. (SRDS), Oct. 2019,
[187] F. Schuh and D. Larimer. (2017). Bitshares 2.0: General Overview. pp. 123–12309.
Accessed: Jun. 2017. [Online]. Available: http://docs.bitshares.org/ [218] S. Jeon, I. Doh, and K. Chae, ‘‘RMBC: Randomized mesh blockchain
downloads/bitshares-general.pdf using DBFT consensus algorithm,’’ in Proc. Int. Conf. Inf. Netw. (ICOIN),
[188] (2020). DPoW. [Online]. Available: https://komodoplatform.com/ Jan. 2018, pp. 712–717.
delayed-proof-of-work/ [219] A. Bessani, J. Sousa, and E. E. P. Alchieri, ‘‘State machine replication for
[189] D. Larimer, ‘‘Transactions as proof-of-stake,’’ Invictus Innov. Group, the masses with BFT-SMART,’’ in Proc. 44th Annu. IEEE/IFIP Int. Conf.
Tech. Rep., 2013. Dependable Syst. Netw., Jun. 2014, pp. 355–362.
[190] (2020). Primecoin. [Online]. Available: https://primecoin.io/ [220] A. Kiayias and A. Russell, ‘‘Ouroboros-BFT: A simple Byzantine fault
[191] (2020). Multichain. [Online]. Available: https://www.multichain.com/ tolerant consensus protocol,’’ IACR Cryptol. ePrint Arch., vol. 2018,
[192] (2020). Waves Protocol. [Online]. Available: https://wavesprotocol.org/ p. 1049, Nov. 2018.
[193] Q. He, N. Guan, M. Lv, and W. Yi, ‘‘On the consensus mechanisms of [221] (2020). Hydrachain. [Online]. Available: https://github.com/
blockchain/DLT for Internet of Things,’’ in Proc. IEEE 13th Int. Symp. HydraChain/hydrachain
Ind. Embedded Syst. (SIES), Jun. 2018, pp. 1–10. [222] (2020). Hydrachain. [Online]. Available: https://www.blockchain-
[194] (2020). Dfinity. [Online]. Available: https://dfinity.org/pdf-viewer/ council.org/blockchain/what-is-hyderachain-technology-how-it-works/
library/dfinity-consensus.pdf [223] B. Expert. (2020). Hydrachain Daap. [Online]. Available: https://www.
[195] (2020). Dfinity. [Online]. Available: https://dfinity.org blockchainexpert.uk/blog/hydrachain-dapp
[196] (2020). ThresholdRelay. [Online]. Available: https://tokens-economy. [224] (2021). Boscoin. [Online]. Available: https://boscoin.io/boscoin/
gitbook.io/consensus/thresholdrelay [225] S. Duan, H. Meling, S. Peisert, and H. Zhang, ‘‘Bchain: Byzantine
[197] D. Puthal and S. P. Mohanty, ‘‘Proof of authentication: IoT-friendly replication with high throughput and embedded reconfiguration,’’ in Proc.
blockchains,’’ IEEE Potentials, vol. 38, no. 1, pp. 26–29, Jan. 2019. Int. Conf. Princ. Distrib. Syst. Cortina d’Ampezzo, Italy: Springer, 2014,
[198] (2020). Cardono. [Online]. Available: https://www.cardano.org/ pp. 91–106.
[199] E. B. Sasson, A. Chiesa, C. Garman, M. Green, I. Miers, E. Tromer, and [226] C. Dwork, N. Lynch, and L. Stockmeyer, ‘‘Consensus in the presence of
M. Virza, ‘‘Zerocash: Decentralized anonymous payments from bitcoin,’’ partial synchrony,’’ J. ACM, vol. 35, no. 2, pp. 288–323, Apr. 1988.
in Proc. IEEE Symp. Secur. Privacy, May 2014, pp. 459–474. [227] Z. Zheng, S. Xie, H. Dai, X. Chen, and H. Wang, ‘‘An overview
[200] (2020). Hydrachain. [Online]. Available: https://github.com/ of blockchain technology: Architecture, consensus, and future trends,’’
HydraChain/hydrachain in Proc. IEEE Int. Congr. Big Data (BigData Congr.), Jun. 2017,
[201] L. Smorgunov, ‘‘Blockchain and a problem of procedural justice of public pp. 557–564.
choice,’’ in Proc. Int. Conf. Digit. Transformation Global Soc. Saint [228] C.-T. Huang, L. Njilla, and T. Geng, ‘‘Consensus of whom? A spectrum
Petersburg, Russia: Springer, 2018, pp. 13–23. of blockchain consensus protocols and new directions,’’ in Proc. IEEE
[202] A. Miller, Y. Xia, K. Croman, E. Shi, and D. Song, ‘‘The honey badger of Int. Smart Cities Conf. (ISC2), Oct. 2019, pp. 1–8.
BFT protocols,’’ in Proc. ACM SIGSAC Conf. Comput. Commun. Secur., [229] L. Baird, ‘‘The Swirlds hashgraph consensus algorithm: Fair, fast,
Oct. 2016, pp. 31–42. Byzantine fault tolerance,’’ Swirlds, College Station, TX, USA,
[203] Hyperledger. [Online]. Available: https://www.hyperledger.org/ Tech. Rep. SWIRLDS-TR-2016, vol. 1, 2016.
wp-content/uploads/2017/08/Hyperledger_Arch_WG_Paper_1_ [230] Y. Jiang and Z. Lian, ‘‘High performance and scalable Byzantine fault
Consensus.pdf tolerance,’’ in Proc. IEEE 3rd Inf. Technol., Netw., Electron. Autom.
[204] H. Moniz, ‘‘The Istanbul BFT consensus algorithm,’’ 2020, Control Conf. (ITNEC), Mar. 2019, pp. 1195–1202.
arXiv:2002.03613. [Online]. Available: http://arxiv.org/abs/2002.03613 [231] (2020). Neo. [Online]. Available: https://neo.org/
[205] F. Borran and A. Schiper, ‘‘A leader-free Byzantine consensus algo- [232] A. Clement, M. Kapritsos, S. Lee, Y. Wang, L. Alvisi, M. Dahlin, and
rithm,’’ in Proc. Int. Conf. Distrib. Comput. Netw. Kolkata, India: T. Riche, ‘‘Upright cluster services,’’ in Proc. ACM SIGOPS 22nd Symp.
Springer, 2010, pp. 67–78. Operating Syst. Princ. (SOSP), 2009, pp. 277–290.
[206] F. Muratov, A. Lebedev, N. Iushkevich, B. Nasrulin, and [233] E. DestBest. (2020). Blocktivity. [Online]. Available: https://blocktivity.
M. Takemiya, ‘‘YAC: BFT consensus algorithm for blockchain,’’ 2018, info/
arXiv:1809.00554. [Online]. Available: http://arxiv.org/abs/1809.00554 [234] (2020). Blocktivity. [Online]. Available: https://blocktivity.info/
[207] (2020). FBFT. [Online]. Available: https://learning.oreilly.com/library/ [235] (2020). Eos. [Online]. Available: https://eos.io/
[236] (2020). Ethereum. [Online]. Available: https://ethereum.org/
view/foundations-of-blockchain/9781789139396/5d9e1324-88c9-4339-
[237] (2020). Vechain. [Online]. Available: https://www.vechain.org/
b974-31426ec72835.xhtml [238] (2020). Burstiq. [Online]. Available: https://www.burstiq.com/
[208] (Dec. 2016). Home. [Online]. Available: https://www.swirlds.com/ [239] (2020). Telosfoundation. [Online]. Available: https://www.
[209] J. Behl, T. Distler, and R. Kapitza, ‘‘Scalable BFT for multi-cores: telosfoundation.io/
Actor-based decomposition and consensus-oriented parallelization,’’ in [240] (2020). Hyperledger. [Online]. Available: https://www.hyperledger.org/
Proc. 10th Workshop Hot Topics Syst. Dependability (HotDep), 2014, projects/fabric
pp. 1–6. [241] (2020). Waltonchain. [Online]. Available: https://www.waltonchain.org/
[210] D. Mazieres, ‘‘The stellar consensus protocol: A federated model for [242] (2020). Stellar. [Online]. Available: https://www.stellar.org/
Internet-level consensus,’’ Stellar Develop. Found., vol. 32, pp. 1–45, [243] (2020). Hyperledger Sawooth. [Online]. Available: https://www.
Jul. 2015. hyperledger.org/projects/sawtooth
[211] B. Yu, J. Liu, S. Nepal, J. Yu, and P. Rimba, ‘‘Proof-of-QoS: QoS based [244] (2020). Ambrous. [Online]. Available: https://ambrosus.com/#home
blockchain consensus protocol,’’ Comput. Secur., vol. 87, Nov. 2019, [245] (2020). Guardtime. [Online]. Available: https://guardtime.com/
Art. no. 101580. [246] (2020). Kin. [Online]. Available: https://www.kin.org/
[212] M. Garg, S. Peluso, B. Arun, and B. Ravindran, ‘‘Generalized consen- [247] (2020). Hedera. [Online]. Available: https://www.hedera.com/
sus for practical fault tolerance,’’ in Proc. 20th Int. Middleware Conf., [248] (2020). Modum. [Online]. Available: https://modum.io/
Dec. 2019, pp. 55–67. [249] (2020). Iryo. [Online]. Available: https://www.iryo.io/
[213] R. Kotla, L. Alvisi, M. Dahlin, A. Clement, and E. Wong, ‘‘Zyzzyva: [250] (2020). Iost. [Online]. Available: https://iost.io/
[251] (2020). Ripple. [Online]. Available: https://ripple.com/
Speculative Byzantine fault tolerance,’’ ACM Trans. Comput. Syst.,
[252] (2020). Cargocoin. [Online]. Available: https://thecargocoin.com/
vol. 27, no. 4, pp. 1–39, 2010. [253] (2020). Encrypgen. [Online]. Available: https://encrypgen.com/
[214] (2020). Neo. [Online]. Available: https://docs.neo.org/docs/en-us/basic/ [254] (2020). Qourum. [Online]. Available: https://www.goquorum.com/
whitepaper.html [255] (2020). Cargox. [Online]. Available: https://cargox.io/
[215] F. Xiang, W. Huaimin, and S. Peichang, ‘‘Proof of previous transac- [256] (2020). Medrec. [Online]. Available: https://medrec.media.mit.edu/
tions (PoPT): An efficient approach to consensus for JCLedger,’’ IEEE [257] (2020). Bitcoin. [Online]. Available: https://bitcoin.org/en/
Trans. Syst., Man, Cybern. Syst., early access, May 8, 2019, doi: 10. [258] (2020). Hyperledger Iroha. [Online]. Available: https://www.
1109/TSMC.2019.2913007. hyperledger.org/projects/iroha
VOLUME 9, 2021 43651

[259] (2020). Shipchain. [Online]. Available: https://shipchain.io/ PETR MUSILEK (Senior Member, IEEE) received
[260] (2020). Metamedium. [Online]. Available: https://www.metadium.com/ the Ing. degree (Hons.) in electrical engineer-
[261] (2020). Bitcoinsv. [Online]. Available: https://bitcoinsv.io/ ing and the Ph.D. degree in cybernetics from
[262] (2020). Corda. [Online]. Available: https://www.corda.net/
[263] (2020). Origintrail. [Online]. Available: https://origintrail.io/
the Military Academy, Brno, Czech Republic,
[264] (2020). Evernym. [Online]. Available: https://www.evernym.com/ in 1991 and 1995, respectively. In 1995, he was
[265] (2020). Bitshares. [Online]. Available: https://bitshares.org/ appointed as the Head of the Computer Appli-
[266] (2020). [Online]. Available: https://www.taelpay.com/ cations Group, Institute of Informatics, Military
[267] (2020). Peermountain. [Online]. Available: https://www. Medical Academy, Hradec Kralove, Czech Repub-
peermountain.com/ lic. From 1997 to 1999, he was a NATO Sci-
[268] (2020). Tron. [Online]. Available: https://tron.network/ ence Fellow with the Intelligent Systems Research
[269] (2020). Stellar. [Online]. Available: https://www.stellar.org/
[270] (2020). Bext360. [Online]. Available: https://www.bext360.com/ Laboratory, University of Saskatchewan, Saskatoon, SK, Canada. In 1999,
[271] (2020). Pokitdok. [Online]. Available: https://pokitdok.com/who-we- he joined the Department of Electrical and Computer Engineering, Uni-
help-health-it/ versity of Alberta, Edmonton, AB, Canada, where he is currently a Full
Professor. Since 2016, he has been serving as the Director of Computer
Engineering Program and the Associate Chair (Undergraduate). He is cur-
rently the Associate Chair (Research and Planning). His research interests
BAHAREH LASHKARI (Graduate Student include artificial intelligence and energy systems. He developed a number of
Member, IEEE) received the MSc. degree in infor- innovative solutions in the areas of renewable energy systems, smart grids,
mation technology with computer networks orien- wireless sensor networks, and environmental monitoring and modeling.
tation. She is currently pursuing the Ph.D. degree
in software engineering and intelligent systems
with the University of Alberta. She worked as a
Software Engineer with Huawei Technologies Ltd.
She is also working as a Research Assistant on
blockchain-enabled smart grids under the super-
vision of Dr. P. Musilek. Her research interests
include blockchain, distributed ledger systems, and deep learning.
43652 VOLUME 9, 2021

A Comprehensive Review of Blockchain Consensus Mechanisms

Uploaded by

Copyright:

Available Formats

A Comprehensive Review of Blockchain Consensus Mechanisms

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

A Comprehensive Review of Blockchain Consensus Mechanisms

Uploaded by

Copyright:

Available Formats

Received February 1, 2021, accepted March 9, 2021, date of publication March 12, 2021, date of current version March

A Comprehensive Review of Blockchain

INDEX TERMS Blockchain, distributed ledger technology, consensus mechanisms, cryptocurrency.

I. INTRODUCTION consensus mechanisms that not only allows the examination

FIGURE 1. Publication outlets from industrial and academic perspective.

VOLUME 9, 2021 43621

FIGURE 2. Variants of distributed ledger technologies.

keys associated with transacting agents ensure the integrity

43622 VOLUME 9, 2021

C. DAG-BASED BLOCKCHAIN 4) VOTING

VOLUME 9, 2021 43623

1) STRUCTURE generated by the sidechain network and its underlying plat-

F. SIDECHAIN 3) SECURITY AND DECENTRALIZATION

43624 VOLUME 9, 2021

FIGURE 4. Classification of consensus mechanisms.

FIGURE 5. Venn diagram for different classes of consensus.

VOLUME 9, 2021 43625

FIGURE 6. Flow chart of Consensus classification procedure.

43626 VOLUME 9, 2021

VOLUME 9, 2021 43627

TABLE 1. Primitve consensus mechanism.

TABLE 2. Proof-Compliant hybrid consensus mechanism.

6) PROOF OF ACTIVITY 10) PROOF OF STAKE VELOCITY

43628 VOLUME 9, 2021

TABLE 3. BFT-Compliant hybrid consensus mechanism.

VOLUME 9, 2021 43629

TABLE 4. Cross-Compliant hybrid consensus mechanism.

43630 VOLUME 9, 2021

TABLE 5. Pure alternative consensus mechanism.

are burned intentionally to denote the investment in 6) PROOF OF TIME

VOLUME 9, 2021 43631

43632 VOLUME 9, 2021

14) PROOF OF CREDIT 18) PROOF OF COOPERATION

VOLUME 9, 2021 43633

43634 VOLUME 9, 2021

30) PROOF OF INDIVIDUALITY 35) RIPPLE/PROOF OF CORRECTNESS

31) PROOF OF PERSONHOOD

VOLUME 9, 2021 43635

permissive latency and bounds it to the network diameter, 2) FaB PAXOS

43636 VOLUME 9, 2021

TABLE 6. Primitive-compliant extensions.

VOLUME 9, 2021 43637

43638 VOLUME 9, 2021

TABLE 7. Proof compliant extensions.

VOLUME 9, 2021 43639

43640 VOLUME 9, 2021

VOLUME 9, 2021 43641

43642 VOLUME 9, 2021

TABLE 8. BFT-compliant extensions.

VOLUME 9, 2021 43643

10) HASHGRAPH/ABFT enables dynamic ownership transformations (as opposed to

16) DELEGATED BYZENTINE FAULT TOLERANCE

43644 VOLUME 9, 2021

FIGURE 7. Distribution plots.

VOLUME 9, 2021 43645

TABLE 9. Consensus in different blockchain platforms [234].

FIGURE 8. Distribution of proposed classification in BC applications.

43646 VOLUME 9, 2021

VOLUME 9, 2021 43647