3.1

PoW algorithm

PoW(Proof of Word) is the neutralization core algorithm of the Bitcoin system. In the Bitcoin system, the sender (certifier) must prove that it has completed a certain amount of computing work within a certain time interval¹⁰. The idea for this consensus comes from the work in the Dwork paper¹¹. PoW is proposed for a peer-to-peer version of an electronic cash system (Bitcoin), where online payments happen directly between two parties without any intermediaries in between. It is believed to solve the double spending problem due to the reversibility of online transactions¹². Its core formula is shown in (1), where data is the data that combines information such as timestamp, version number, and block height, nonce is a random value, D(d) is the target value, and d is the mining As the difficulty of mining increases, it becomes more difficult to match the target value with the hash value.

There are three main problems in the PoW algorithm: a) Low efficiency. The block capacity in the Bitcoin system is only 1MB, the transaction volume in the network is about 7 transactions/s, and the block generation time is about 10 minutes. The throughput of this specification is very unsuitable for high-concurrency business scenarios. b) High resource consumption. Since the miner nodes need a large amount of power and stream processors and other mining machines during the mining process, according to the Digiconomist Bitcoin Energy Consumption Index and the Cambridge Bitcoin Power Consumption Index, the electricity consumption of the Bitcoin network in 2018 is equivalent to Electricity consumption in the Philippines or Finland. c) The degree of decentralization is low. The original intention of Satoshi Nakamoto to design Bitcoin mining is that everyone uses CPU to mine, but now most of the computing power is in the hands of computing power mining pools, and mining pools and mining pools will also form alliances to compete for bookkeeping rights.

3.2

PoS algorithm

PoS(Proof of Stake) was proposed by King and others, and was later adopted by Nextcoin. PoS proposes solutions to the problems of excessive waste of PoW resources and slow block generation time, so that each participating node has two attributes of currency holdings and currency age. The calculation is shown in formula (2). Whenever a new transaction message in the blockchain network begins to appear, the weight is calculated according to the currency age of each node, and the node conducts a round of elections according to the weight. The probability of the node with the greater weight being selected as the bookkeeping node is also The bigger it is, then the bookkeeping node starts to package transaction messages, its currency age will be reduced, and the bookkeeping node will get incentives and some transaction fees. Although PoS solves the problems caused by PoW to a certain extent, the degree of decentralization is fundamentally low, and the bookkeeping rights are still in the hands of a small number of nodes with older currency ages, which is likely to cause polarization among currency holders The phenomenon. In theory, the attacker needs to have 51% of the currency holdings to successfully attack¹³.

The first problem faced by the original PoS algorithm is the non-interest attack (the bookkeeping node maliciously forks to obtain double benefits). The main reason for this problem is that there is no clear equity distribution and punishment measures for the bookkeeping nodes in the application framework. Another problem is long-range attacks. Since most public chains are based on the longest chain principle, if malicious nodes create a pseudo-chain longer than the main chain, all transaction records will be tampered with, and the public chain If there are many nodes that have been offline for a long time or have newly added nodes when synchronizing data, the blockchain will fork. According to the principle of the longest chain, the pseudo-chain will be selected as the main chain.

3.3

PoSpace algorithm

The core idea of PoSpace (proof of space)¹⁴ is to use the user’s hard disk space as the cost of proof to replace the computing resources of PoW. The size of the user’s workload can be judged by the size of the content downloaded by the node. After the node pays for the disk capacity, subsequent mining There is no need to pay additional costs, and nodes with larger storage space are more likely to become bookkeeping nodes. In the PoSpace algorithm, the participating nodes create a nonce value through the Shabal algorithm. The nonce is very difficult to calculate, so it is necessary to pre-calculate and store the nouce value. Every two hash values constitute a scoop. A nouce is composed of multiple scoops. The more space the user has, The more the nouce value, the node needs to use its available disk space to continuously store the nouce value before starting mining. If there is a hash value that is closest to the puzzle in the network, it wins the mining This algorithm has the characteristics of a high degree of decentralization, low requirements for obtaining tokens, and low energy consumption. A quality function is defined in PoSpace as a competition indicator:

Among them, hash is the hash value of the ordinary node being verified, and S is the size of the storage space occupied by the data. In each round of competition, the node with the smallest quality function result wins.

The PoSpace consensus algorithm uses storage space instead of computation, saving power resources; also, under this consensus protocol, the storage space size is determined when the node first connects to the network and cannot be expanded afterwards, preventing the emergence of “mining pools” in the PoW consensus algorithm (pooling the computing power of different nodes into one large computing power node) and, to a certain extent, avoiding the problem of increased centralisation and reducing security risks. The problem of increased centralisation is avoided to a certain extent, and security risks are reduced.

4.1

DPoS algorithm

DPoS was designed by Larimer and first implemented in the BitShares project¹⁵. Its purpose is to solve the problem of centralized accounting in PoS. DPoS introduces a proxy mechanism, and currency holders can elect super nodes as accounting representatives, and elect several representative nodes to keep accounts in turn. Each super node has a cycle. It will be removed when there is an abnormality in the cycle. DPoS consumes less energy, reduces network operating costs, is more decentralized than PoS, and achieves consensus faster, but the participation in community election voting is very low. When dealing with abnormal super nodes, the election system cannot solve abnormalities in time. Problems posed by malicious nodes.

Using DPoS can reduce the number of participating verification and accounting nodes, so as to achieve second-level consensus verification. However, the centrality of the blockchain system using the DPoS consensus algorithm is relatively weak, and its security is weaker than that of POW. At the same time, the node agent is manually selected, and its fairness is lower than that of POS. At the same time, the entire consensus mechanism still relies on tokens. Additional issuance to maintain the stability of proxy nodes. DPoS introduces an election mechanism into PoS. DPoS divides nodes into two roles: ordinary nodes and trusted nodes. Ordinary nodes can vote for trusted nodes or be voted as trusted nodes. In DPoS, nodes consume rights and interests as voting rights, and the most trusted N nodes are generated according to the weighted results of rights and interests to become the trust node set N{N₁…N_n}, and each trust node is given a fixed period to become the master node in turn, after the authorization time is over, the authority is handed over to the next trusted node. After the set N ends the cycle, a new N is re-elected, and malicious nodes will lose trust in this round of elections. The flow of the DPoS consensus algorithm is shown in Figure 2.

Figure 2.

DPoS consensus algorithm flow chart

At present, the blockchain application systems that use the DPoS consensus algorithm mainly include Bitshares¹⁶ (bit shares), Steem¹⁷, and EOS¹⁸, and the number of principals in both Steem and EOS is 21.

4.2

PBFT algorithm

The PBFT consensus algorithm is implemented in Fabric v0.6.0¹⁹. The consensus algorithm selects a master node from the entire network nodes to be responsible for creating blocks, and then reaches a consensus through three-stage voting: pre-preparation stage, preparation stage, and submission stage. As shown in Figure 3, the main process is as follows:

Figure 3.

PBFT consensus process

PBFT can provide strong consistency and overcome the shortcomings of the original BFT algorithm, which is not efficient. The application of this consensus algorithm in the blockchain system can greatly improve the block generation speed of the system, and at the same time provide a certain degree of security, which can ensure the safe operation of the system when less than 1/3 of the nodes fail. However, this algorithm also has some shortcomings. Since the three-stage protocol requires nodes to broadcast messages to the entire system, if the number of nodes in the system is too large, the number of messages in the network will increase significantly, resulting in network congestion, so it is not suitable for large-scale Large-scale blockchain system.

4.3

DBFT algorithm

The DBFT algorithm is proposed in the NEO project²⁰. DBFT is improved from the PBFT algorithm. The PBFT algorithm consensus process requires the participation of all nodes, and requires three rounds of network request confirmation to reach a consensus. The complexity of network communication is O(N2), where N is the number of nodes in the entire network, and the scalability is poor. As the network size increases, it is difficult to reach consensus quickly. NEO’s solution is to vote for some nodes to participate in the consensus, thereby reducing network communication consumption, improving scalability, and increasing transaction processing speed. The DBFT formula process is shown in Figure 4.

Figure 4.

DBFT consensus process

DBFT consensus process: All nodes will vote for a bookkeeping node (that is, a credible node, here is done according to the original PBFT algorithm), while other nodes are not bookkeeping, not Will participate in the consensus process, but accept the final consensus. Then, the speaker is selected from the collection of bookkeeping nodes (the speaker is the main node responsible for establishing blocks), and the remaining bookkeeping nodes are councilors; after each consensus, the speaker will propose a new block, and then Confirmation and broadcast by MPs. When a node receives the confirmation information sent by more than 2/3 of the bookkeeping nodes, it will add a block locally, that is, reach a consensus on the latest block; after that, select a new master node and carry out A new round of consensus. The advantages of the DPoS consensus algorithm are: no need to consume huge energy, higher operating efficiency, faster block generation, and less prone to forks. The disadvantage is: the degree of decentralization is not high, and the problem of bribery is prone to occur.