Blockchain technology is rapidly integrating into various spheres of human activity. Private companies, government agencies, and international organizations are gradually adapting this technology to solve a wide range of tasks. The main areas of its use include financial transactions, document management, digital identification, control of logistics links, and tokenization of physical and classical financial assets. The more a technology develops, the more it needs to be updated and improved. For blockchain, the chosen consensus algorithm is very important. There is a need to ensure control over data and resources and their copies on different nodes to avoid conflicts between nodes. After all, any conflict between nodes can lead to inefficient and inconsistent data storage. As a blockchain is a specialized type of database that stores this data by distributing it among several completely independent nodes, i.e. computers or devices, blockchains allow data to be added to databases and make it impossible to attempt to change or delete them. Therefore, this article is based on the study of the criteria that can help in the selection of a consensus algorithm. Four main criteria are identified, which in combination, allow us to select a consensus algorithm more accurately. The main selection criteria are energy consumption, decentralization, security, and bandwidth. Each feature of these criteria has been considered during the study and highlighted in this article. It is very important to combine different criteria and their parameters to choose the most successful consensus algorithm. Different approaches make it possible to find the most optimal option. Based on the introduced criteria and the proposed methodology, a program for selecting the optimal consensus algorithm has been created using the Python programming language.
Blockchains are decentralized, meaning that data can be distributed across multiple host servers, which distinguishes them from conventional databases. Decentralization is the main feature of blockchain.
Since control over data or resources is shared among several nodes simultaneously, it makes it very difficult for an attacker to delete or use resources. The process of changing data or using resources can only take place with the consent of the majority of nodes in the blockchain if the blockchain is large and contains sufficiently independent nodes, which makes it impossible for an attacker [1].
Blockchain decentralization also comes with a social and technical challenge, which is maintaining consensus between nodes. Consensus maintenance can ensure that all participants agree with the decisions made by the network [2, 3].
Consensus maintenance is very important because, without it, copies of data on different nodes may conflict with each other, leading to inefficient and inconsistent data storage. Consensus also makes it possible to update the underlying blockchain protocol, i.e. the rules for node interaction and regulation of its organization. If it became necessary to modify the protocol due to security or performance issues, all nodes would have to agree to accept the change, as different nodes cannot use different protocols in the same blockchain.
Because all nodes in a blockchain are decentralized and work independently, maintaining consensus creates a complex problem, which different types of blockchains have solved in different ways. Different solutions for maintaining consensus between nodes provide us with technical innovations for further updates of algorithms for blockchain technology.
The first blockchain networks used the Proofof-Work (PoW) consensus mechanism, the main disadvantages of which are high energy costs, low network bandwidth, and high transaction costs. Alternative mechanisms have been developed to overcome these issues.
Currently, the most common replacement for PoW is the Proof-of-Stake (PoS) consensus mechanism.
The essence of this approach is that to generate a new block, the holder of the validator node must stake a certain number of the native tokens and interact with other nodes to reach a consensus on the chain. In case the validator node engages in malicious activity, it risks losing the blocked tokens partially or completely.
This approach, inherently, has several vulnerabilities; the main one is the possibility of accumulating a significant number of tokens with a small number of related holders, which can lead to catastrophic consequences for the entire network. This problem is called capital centralization or oligopoly.
A typical solution to ensure the consensus of PoS networks is a scheme for distributing rewards and penalties to validator nodes.
The purpose of this paper is to analyze the criteria and parameters of a blockchain network to determine the optimal consensus algorithm.
A successful consensus algorithm should help a blockchain application achieve its goal. The purpose of the blockchain is to provide a tool for decentralized decision-making. The complex nature of blockchain consensus stems from its original goal of making decisions without a central authority. Therefore, the level of decentralization of the consensus algorithm is included in the evaluation criteria.
A consensus algorithm with a higher level of decentralization is considered good [4].
Decentralization plays a key role in a blockchain network as it determines the level of power and control distribution among participants. Let’s consider several criteria that can be taken into account when choosing a consensus algorithm in terms of decentralization: • Level of equality (how equally the power is distributed among the network participants, the fewer centralized control points there are and, the more decentralized is the system); • Decentralized decision-making mechanisms (how decisions are made about the direction of the network development and whether there are mechanisms, which allow participants to contribute to the decision-making process). • Scalability of decentralization (how the system decentralizes as the number of participants and the volume of transactions increase; and whether the network can remain decentralized over time and grow in scale). • Fault tolerance to attacks (i.e. how the system responds to attempts to hack into the decentralized structure and whether there are mechanisms to prevent concentration of power or a controlled attack). • The number of nodes (the more nodes in the network, the more decentralized the system can be considered, but it is also important to consider how these nodes are selected and controlled). • Sybil Resistance (how the system prevents attacks in which an attacker can create many artificial identities to gain control over the network). • Algorithmic implementation (how the consensus algorithm promotes decentralization) [5–11]. Taking these criteria into account helps to show that the chosen consensus algorithm promotes the highest possible
level of decentralization in the blockchain network.
Another important criterion is whether the transaction speeds, such as payment systems or micropayments. Second (TPS), transaction confirmation time.
The calculation of transaction speed in the consensus algorithm can fulfill its own goal, form of formulas can look like this: (1) (2) = , where is the speed of transactions; n is the number of transactions that have
been successfully processed over a certain period; t is the time interval, during which the transactions have been processed.
2. Scalability: Scalability determines how easily the network can be increased in size and load to serve more users and transactions without losing performance.
Metrics: Infrastructure to scale up the network, network bandwidth, horizontal and
The formula for calculating the scalability vertical scalability. coefficient: which is to achieve consensus. In the context of blockchain and cryptocurrencies, this can be defined as the speed at which the algorithm reaches consensus or the
amount of bandwidth. For example: • Fast action and number of transactions per second (TPS) (i.e. how fast a particular consensus algorithm can be processed and how many transactions can be processed in one second. This is especially important for blockchain networks that have a high transaction flow). • Scalability (how well the algorithm scales when the number of participants and the flow of transactions increase). • Processing
multitasking (whether the algorithm can efficiently process many transactions simultaneously). • Latency (how quickly transactions can be
confirmed in the network). • Resource dependence (the system must be
efficient in terms of resource use). • Network efficiency (whether
the algorithm works well in a network with a large number of participants and a large amount of data). • Protocol flexibility (the ability to adjust algorithm parameters to achieve optimal bandwidth in specific conditions) [12–19].
These criteria help to establish that the selected consensus algorithm meets the network capacity requirements and ensures efficient transaction
processing in the blockchain system. the us algorithm:
The blockchain network criteria play an important role in the selection of a consensus algorithm and in the process of developing a blockchain solution. Let’s consider, in detail, formulas for calculating the
main blockchain network metrics, which will allow to determine the
optimal consensus
Transaction
speed: This criterion determines how
quickly the network can process and confirm
transactions. It is important for applications that require high where M is the scalability coefficient; P is the maximum number of transactions that the network can process within one second; k is the average number of active users in the network.
3.
Decentralisation:
criterion indicates the degree of distributed control and participation in the network. A higher level of decentralization means greater independence and security.
Metrics: Number of nodes, participating in the consensus, concentration of power, location of geographically different nodes.
To find the decentralization ratio, use the following formula: = 1 − , (3) where D is the decentralization factor and K is a value that determines the degree of centralization of control in the network, it can be a numerical value from 0 (full decentralization) to 1 (full centralization).
4. Security: Security is defined as the level of protection a network has against attacks and misuse. The data must be securely protected from unauthorized access.
Metrics: Level of cryptographic security, resistance to 51% attack, protection against double-spending, vulnerability detection, and remediation.
The formula for calculating the security factor: = (1 − ), (4) where S is the security factor, Rs is the level of cryptographic security, i.e. an assessment of the level of data protection in the network, which can be a numerical value from 0 (no security) to 1 (high security); Ra is the risk of attack, i.e. the probability of an attack on the network.
5. Costs: Costs include the cost of operating the network, including equipment, electricity, and transaction fees. Low costs can be important for maintaining network stability and attracting users [20–24]. 2. Transactions flow per second (TPS): 100.
4. Time between blocks: 10 minutes.
1. Number of active master nodes: 500. 2. Transactions flow per second (TPS): 200.
4. Time between blocks: 5 minutes.
Metric calculation—Transaction speed (TPS):
= (100*1,000)/600 = 166,67
= (200*500)/300 = 333,33
As can be seen, the “Transaction Rate” (TPS) metric for network B (PoS) is higher than for network
A (PoW), which indicates that network B can process more transactions per second. According to this data, network B may be more suitable for applications that require high transaction speeds, while network A has more miners and possibly more security due to PoW.
M = (10 minutes/block)*(1 block/10 minutes) = 1 block/minute.
M = (5 minutes/block)*(1 block/5 minutes) = 1 block/minute.
For Network A (PoW): Decentralisation is difficult to calculate without up-to-date concentration data.
D = 1–0.9 = 0.1.
For Network A (PoW): The security level will be high due to the powerful hash power.
For Network B (PoS): Security can be determined based on the level of cryptographic security and the risk of attack, which are difficult to calculate without specific data.
includes the cost of equipment and electricity to mine the blocks. The number of transactions is also difficult to estimate without specific data.
PoS costs include master node maintenance and transaction fees.
A consensus algorithm is considered more secure if it can protect against different types of security threats.
Therefore, security is a critical aspect when choosing a consensus algorithm in blockchain technology. Here are some key security criteria: • Attack resistance (how well the system can resist various types of attacks, such as double costs, white hat attacks, consensus attacks, etc;). • Decentralization (the gradient decentralization can determine difficult it is to carry out successful attacks on the system, the more different participants in the network, the more difficult it is to interfere with the consensus). • Key protection methods (what methods are used to store and protect the private keys of network participants and the security of the key storage system).
of how • Fault tolerance (restoring the system after errors or attacks is important for the continuous operation of the network). • Algorithmic strength (determining the strength of the cryptographic algorithms, which are used in the consensus algorithm. For example, if a hash function is used, how resistant it is to collisions and attacks). • Sybil Attack Resistance (how the system prevents attacks in which one participant can create many pseudo-identities to dominate the network). • Active cooperation (Byzantine Fault Tolerance—BFT) (the requirement for the network to be immune to attacks that may lead to a discrepancy in information between participants). • Protocol upgrades and changes (how easy it is to implement changes to the consensus protocol, and how much this can affect network security) [25–27].
Finally, one should not consider only theoretical aspects. There is another criterion— the energy consumption of the algorithm. If an algorithm consumes too much energy, it should be changed and more environmentally friendly options should be considered.
The following energy consumption criteria for choosing a consensus algorithm in blockchain technology should be considered: Proof of Work (PoW): • Computational complexity (the more difficult the task is for miners, the more energy is consumed). • Algorithm efficiency (some PoW algorithms may be more energy efficient than others). 1. Proof of Stake (PoS): • Selective efficiency (the less currency a participant has, the less energy he uses). • Methods for controlling misuse (it is important to have mechanisms to prevent concentration of power that may affect the effectiveness of PoS). 2. Delegated Proof of Stake (DPoS): • Chosen legitimacy (some participants may spend more energy to obtain delegate status). • Flexibility of the voting policy (if the voting system is not efficient, it can lead to an incorrect distribution of power and energy costs). 3. Proof of Burn and other alternatives: • Spending strategies (determining exactly how currency is spent in the consensus process and how this affects energy costs. • Innovative approaches (alternative methods such as Proof of Space (PoSpace) or Proof of Time (PoT) may offer lower energy costs).
However, the energy consumption criteria must be balanced with other important aspects, such as security, decentralization, and bandwidth, when choosing a consensus algorithm for blockchain technology.
Here is a comparative description of consensus algorithms for different types of cryptocurrencies (Table 1).
Energy consumption Medium Low Low Medium Medium Hight Low Medium Medium Medium Low Low Low Hight Low
Bandwidth OK Not So Fast OK OK Fast Not So Fast Very Fast Fast OK OK Fast Fast Fast Very Fast Fast The main steps of determining the consensus algorithm by the selected criteria are shown in Fig. 1.
The obtained results formed the basis of the methodology for determining the consensus algorithm for the blockchain network. Based on the obtained data, a program for determining the optimal algorithm has been created in the Python programming language. The program allows choosing one of the available consensus algorithms by the established criteria. The program can be extended to take into account more metrics and conditions for choosing an algorithm.
The criteria for helping to select a consensus algorithm, by reviewing popular algorithms and providing a solution in the form of a decision tree have been presented in this article. This solution is very useful for the further selection of consensus algorithms and the development of blockchain technology.
In general, the criteria, discussed in this article, make it clear how important the method of selecting a consensus algorithm is.
After all, a successfully selected consensus increases the efficiency of blockchain technology.
This study is limited in time. For the whole population, the consensus algorithm may be useful to study more algorithms and create a larger decision tree, as they all have their characteristics. Also, this article is an impulse to dive deeper and compare performance using benchmarking.
For any future work, it would be interesting to have a study on the selection of a consensus algorithm and it is possible to choose a wider list of criteria or reduce it to a minimum. In addition, the blockchain application industry is developing rapidly and new algorithms are being created that can quickly replace the old ones on the market. There is always a need for future research, so it is a necessity to keep up to date with the latest developments in consensus algorithms and blockchain technology. [4] F. Kipchuk, et al., Assessing Approaches and Telecommunication Systems, Vol. of IT Infrastructure Audit, in: IEEE 8th 3421 (2023) 223–230.
International Conference on Problems of [15] D. Mingxiao, et al., A Review on Infocommunications, Science and Consensus Algorithm of Blockchain, Technology (2021). doi: 10.1109/ IEEE International Conference on picst54195.2021.9772181. Systems, Man, and Cybernetics (2017). [5] B. Sriman, et al., Blockchain Technology: doi: 10.1109/SMC.2017.8123011.
Consensus Protocol Proof of Work and [16] R. Pass, E. Shi, Hybrid Consensus: Proof of Stake, Intelligent Computing and Efficient Consensus in the Applications (2020) 395–406. doi: Permissionless Model, Initiative for 10.1007/978-981-15-5566-4_34. CryptoCurrency and Contracts (2016) 1– [6] S. Bamakan, et al., A Survey of Blockchain 55. doi: 10.4230/LIPIcs.DISC.2017.39.
Consensus Algorithms Performance [17] S. King, S. Nadal, PPCoin: Peer-to-Peer Evaluation Criteria, Expert Systems with Crypto-Currency with Proof-ofStake Applications 154 (2020) 113385. doi: (2012).
10.1016/j.eswa.2020.113385. [18] L. Shi, et al., Pooling Is Not Favorable: [7] H. Cho, ASIC-Resistance of Multi-Hash Decentralize Mining Power of POW Proof-of-Work Mechanisms for Blockchain Using Age-of-Work, IEEE Blockchain Consensus Protocols, IEEE Transactions on Cloud Computing 11(3) Access 6 (2018) 66210–66222. doi: (2022). doi: 10.1109/TCC.2022.3226496. 10.1109/ACCESS.2018.2878895. [19] B. Sriman, et al., Blockchain Technology: [8] T. Duong, et al., 2-Hop Blockchain: Consensus Protocol Proof of Work and Combining Proof-of-Work and Proof-of- Proof of Stake, Intelligent Computing and Stake Securely, Computer Security – Applications (2020) 395–406. doi: ESORICS 2020 (2020) 697–712. doi: 10.1007/978-981-15-5566-4_34. 10.1007/978-3-030-59013-0_34. [20] Y. Yu, Analysis of POW in Bitcoin and POS [9] A. Guru, et al., A Survey on Consensus in Peercoin, Highlights in Science, Protocols and Attacks on Blockchain Engineering and Technology 39 (2023) Technology, Applied Sci. 13(4) (2023) 784–788. doi: 10.54097/hset.v39i.6645. 2604. doi: 10.3390/app13042604. [21] Q. Zhang, et al., Blockchain Model [10] S. King, S. Nadal. PPCoin: Peer-to-Peer Testing and Implementation Based on Crypto-Currency with Proof-of-Stake, Improved PBFT Consensus, 11th IEEE (2012) 1–6. International Conference on Intelligent [11] Y. Liu, et al., Hybrid Consensus Protocols DataAcquisition and Advanced and Security Analysis for Blockchain, Computing Systems: Technology and International Conference on Data Applications (IDAACS) (2021). doi: Analytics, Computing and Artificial 10.1109/idaacs53288.2021.9660959. Intelligence (ICDACAI) (2022). doi: [22] S. Zhang, J.-H. Lee, Analysis of the Main 10.1109/ICDACAI57211.2022.00046. Consensus Protocols of Blockchain, ICT [12] B. Lucas, R. V. Paez, Consensus Express 6(2) (2020) 93–97. doi: Algorithm for a Private Blockchain, IEEE 10.1016/J.ICTE.2019.08.001. 9th International Conference on [23] W. Zhao, et al., On Peercoin Proof of Electronics Information and Emergency Stake for Blockchain Consensus, 3rd Communication (ICEIEC) (2019). doi: International Conference on Blockchain 10.1109/ICEIEC.2019.8784500. Technology (2021) 129–134. doi: [13] A. Mackenzie, Memcoin2: A Hybrid 10.1145/3460537.3460547.
Proof-of-Work, Proof-of-Stake Crypto- [24] W. Zhao, et al., On Peercoin Proof of Currency (2013). Stake for Blockchain Consensus, 3rd [14] V. Malinov, et al., Biomining as an International Conference on Blockchain Effective Mechanism for Utilizing the Technology (2021) 129–134. doi: Bioenergy Potential of Processing 10.1145/3460537.3460547. Enterprises in the Agricultural Sector, [25] V. Zhebka, et al., Optimization of Machine Cybersecurity Providing in Information Learning Method to Improve the Management Efficiency of Heterogeneous Telecommunication Network, Cybersecurity Providing in Information and Telecommunication
Systems Vol. 3288 (2022) 149–155. [26] Z. Zheng, et al., An Overview of
Blockchain Technology: Architecture, Consensus, and Future Trends, IEEE International Congress on Big Data (BigData Congress) (2017). doi: 10.1109/BIGDATACONGRESS.2017.85. [27] Z. Zheng, et al., An Overview of
Blockchain Technology: Architecture, Consensus, and Future Trends, IEEE International Congress on Big Data (BigData Congress) (2017). doi: 10.1109/BIGDATACONGRESS.2017.85.