CN114466024B

CN114466024B - Novel PoM consensus algorithm-based distributed account book system and method

Info

Publication number: CN114466024B
Application number: CN202210157526.5A
Authority: CN
Inventors: 薛立德; 于铭华; 邱定
Original assignee: CETC 32 Research Institute
Current assignee: CETC 32 Research Institute
Priority date: 2022-02-21
Filing date: 2022-02-21
Publication date: 2024-02-13
Anticipated expiration: 2042-02-21
Also published as: CN114466024A

Abstract

The invention provides a distributed account book method and a system based on a novel PoM consensus algorithm, wherein the method comprises the following steps: step S1: randomly selecting a preset hot candidate node from all candidate nodes; step S2: in leader election, a consumption node supports corresponding hot candidate nodes by submitting legal transaction data to a plurality of hot candidate nodes, and when the legal transaction data collected by the hot candidate nodes meet preset requirements, the current hot candidate node is a leader node; step S3: and integrating information supporting the hot candidate node into legal Avalon block data by submitting legal transaction data to the hot candidate node by the consumption node, storing the legal Avalon block data in a new legal block, broadcasting the current block in a P2P network by calling a gossip transmission protocol, and realizing that a new legal block is agreed by each node.

Description

Novel PoM consensus algorithm-based distributed account book system and method

Technical Field

The invention relates to the technical field of blockchains, in particular to a distributed ledger system and a method based on a novel PoM consensus algorithm.

Background

The blockchain is the most typical example in many existing distributed ledger systems, but the blockchain technique behind it is essential. The core aim is to make the public account book agree on the user without a trusted third party. The Nakamoto consensus, the core consensus algorithm of the present blockchain, uses proof of work (PoW) to make the cost of computing nodes (nodes that participate in the blockchain system and execute the PoW) very high, thereby ensuring the security of the overall system consensus. The Nakamoto consensus considers that a system is effectively secure as long as the computational power of malicious computing nodes in the system does not exceed 50% of the total computational power of the whole network, although later work revealed some drawbacks and potential attacks of the Nakamoto consensus.

In addition to these attacks and vulnerabilities, the availability of the Nakamoto consensus is poor compared to current online transaction requirements—on average only 7 transactions per second can be processed, and on average the transaction validation delays are up to one hour. In addition, poW-type blockchains also face the challenge of decentralization. The top-of-spike (PoS) and leader elections of PoW have shown a significant tendency to "center" and "unfairness", and excessive centering is also very likely to cause some security problems. While traditional BFT class protocols are completely decentralised, they limit nodes from freely joining or exiting the system and are easily damaged by DDoS attacks. Currently, many high-performance consensus algorithm protocols are based on the two general consensus schemes (PoW and BFT), however, they are all or more or have some defects and deficiencies, and the PoW scheme causes performance bottlenecks by excessive coupling of system efficiency and security; BFT is not suitable for large network multi-node consensus environments and requires additional node verification mechanisms to prevent malicious attacks such as DDoS, sybil types.

Classical blockchains suffer from a number of disadvantages, most of which should be scalability issues. For this reason, many solutions have also been proposed by the academy and industry, such as the middle blockchain-NG; the block chain-NG introduces the concepts of 'key block' and 'micro block', the key block uses classical PoW to carry out leader election, and the micro block is used for storing transactions. And to prevent double flower attacks, the middle blockchain-NG introduces a "poison transaction" mechanism. However, it still links PoW with leader node election and introduces additional complex consensus mechanisms (e.g., BFT), which would greatly reduce its efficiency in extreme cases. In addition, some modifications to the PoW (e.g., poS, DPoS, etc.) do not change the essence of the PoW.

Patent document CN110300969a (application number 201780086712.1) discloses a technique comprising: at an existing node of the distributed ledger network for operating according to a voting-based consensus algorithm, new candidate nodes of the distributed ledger network are identified, wherein the identification is based on an existing unique identifier that is independent of the distributed ledger network.

Disclosure of Invention

Aiming at the defects in the prior art, the invention aims to provide a distributed ledger system and a method based on a novel PoM consensus algorithm.

The invention provides a distributed account book method based on a novel PoM consensus algorithm, which comprises the following steps:

step S1: randomly selecting a preset hot candidate node from all candidate nodes;

step S2: in leader election, a consumption node supports corresponding hot candidate nodes by submitting legal transaction data to a plurality of hot candidate nodes, and when the legal transaction data collected by the hot candidate nodes meet preset requirements, the current hot candidate node is a leader node;

step S3: and integrating information supporting the hot candidate node into legal Avalon block data by submitting legal transaction data to the hot candidate node by the consumption node, storing the legal Avalon block data in a new legal block, broadcasting the current block in a P2P network by calling a gossip transmission protocol, and realizing that a new legal block is agreed by each node.

Preferably, the legal transaction data employs: only when the consumption node completes the preset consumption node PoW consensus, the transaction is considered to be in accordance with the preset requirement and belongs to legal transaction data;

the consumption node PoW consensus is that when the candidate node executes leader election, the consumption node needs to output corresponding legal transaction data submitted through the consumption node PoW and submit the legal transaction data to the corresponding candidate node in the leader election process.

Preferably, the leader election and the consumption node PoW consensus are decoupled, so that the correlation of the security and the block-out interval is weakened, the throughput is improved, and the delay of the transaction is reduced.

Preferably, the step S1 employs:

step S1.1: randomly selecting preset hot candidate nodes from all candidate nodes in each period T through a leader election scheme of VRF, and setting other candidate nodes as cold candidate nodes;

step S1.2: and the consumption node submits legal transaction data to the hot candidate node so as to support the current hot candidate node, and when the consumption node judges that the current hot candidate node is malicious or offline, the consumption node submits legal transaction data to the cold candidate node which is randomly selected.

Preferably, the consumption node judges that the current hot candidate node is malicious or offline adopted: and when the hot candidate nodes do not release new blocks after the preset time, the consumption node determines that the hot candidate nodes are all offline.

The invention provides a distributed account book system based on a novel PoM consensus algorithm, which comprises the following components:

module M1: randomly selecting a preset hot candidate node from all candidate nodes;

module M2: in leader election, a consumption node supports corresponding hot candidate nodes by submitting legal transaction data to a plurality of hot candidate nodes, and when the legal transaction data collected by the hot candidate nodes meet preset requirements, the current hot candidate node is a leader node;

module M3: and integrating information supporting the hot candidate node into legal Avalon block data by submitting legal transaction data to the hot candidate node by the consumption node, storing the legal Avalon block data in a new legal block, broadcasting the current block in a P2P network by calling a gossip transmission protocol, and realizing that a new legal block is agreed by each node.

Preferably, the module M1 employs:

module M1.1: randomly selecting preset hot candidate nodes from all candidate nodes in each period T through a leader election scheme of VRF, and setting other candidate nodes as cold candidate nodes;

module M1.2: and the consumption node submits legal transaction data to the hot candidate node so as to support the current hot candidate node, and when the consumption node judges that the current hot candidate node is malicious or offline, the consumption node submits legal transaction data to the cold candidate node which is randomly selected.

Compared with the prior art, the invention has the following beneficial effects:

1. although the invention still needs to execute the PoW operation, poM discretizes the PoW operation to the level of the consuming node, decouples the security and efficiency of the system, improves both, and reasonably utilizes the PoW to increase the cost for an attacker.

2. The invention adopts VRF & BFT mechanism for the first time, which uses verifiable random numbers to ensure mathematical fairness of leader node election.

3. The invention can serve as a bottom consensus mechanism for the segmentation and the side chains so as to obtain larger promotion.

4. The present invention comprehensively considers efficiency, fairness, decentralization, and energy conservation, and attempts to break through many of the limitations faced by classical blockchains. PoM is essentially a new type of PoW consensus, but it transfers the work of "hash operations" from the block producer to the consuming node. And a reasonable excitation mechanism is added, so that the attack of the malicious node is not profitable. Under the assumption of a rational attacker, the method has better performance and decentralization, and fully considers the willingness of the consumption node. The security is not inferior to the equally configured PoW type blockchain in the face of irrational attackers.

Drawings

Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, given with reference to the accompanying drawings in which:

FIG. 1 is a diagram of an Avalon system architecture.

FIG. 2 is a schematic diagram of a leader election process of Avalon.

Fig. 3 is a legal vot structure diagram of a consuming node.

FIG. 4 is a schematic diagram of the excitation mechanism of Avalon.

Fig. 5 is a graph of a relationship experiment between the parameter TAT and the throughput of the Avalon system.

FIG. 6 shows the parameter d _c And the relation experiment diagram between the throughput of the Avalon system.

Fig. 7 is a graph of TAT versus transaction confirmation delay.

FIG. 8 d _c And (3) a relation experiment diagram of transaction confirmation delay.

Fig. 9 is an experimental plot of TAT versus transaction rollback rate.

FIG. 10 d _c And a relation experiment diagram of the transaction rollback rate.

FIG. 11 is an experimental plot of TAT versus bifurcation number and height.

FIG. 12 d _c Experimental plot of the number and height of bifurcations.

Detailed Description

The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the present invention, but are not intended to limit the invention in any way. It should be noted that variations and modifications could be made by those skilled in the art without departing from the inventive concept. These are all within the scope of the present invention.

Example 1

the step S1 specifically adopts:

step S1.2: and the consumption node submits legal transaction data to the hot candidate node so as to support the current hot candidate node, and when the consumption node judges that the current hot candidate node is malicious or offline, the consumption node submits legal transaction data to the cold candidate node which is randomly selected. The consumption node judges whether the current hot candidate node is malicious or offline specifically adopts: and when the hot candidate nodes do not release new blocks after the preset time, the consumption node determines that the hot candidate nodes are all offline.

Specifically, the legal transaction data employs: only when the consumption node completes the preset consumption node PoW consensus, the transaction is considered to be in accordance with the preset requirement and belongs to legal transaction data;

Specifically, the leader election and the consumption node PoW consensus are decoupled, so that the correlation of the security and the block-out interval is weakened, the throughput is improved, and the delay of the transaction is reduced.

the module M1 specifically adopts:

module M1.2: and the consumption node submits legal transaction data to the hot candidate node so as to support the current hot candidate node, and when the consumption node judges that the current hot candidate node is malicious or offline, the consumption node submits legal transaction data to the cold candidate node which is randomly selected. The consumption node judges whether the current hot candidate node is malicious or offline specifically adopts: and when the hot candidate nodes do not release new blocks after the preset time, the consumption node determines that the hot candidate nodes are all offline.

Example 2

The invention provides a novel PoM consensus algorithm-based distributed ledger system, which comprises:

assuming that the hash function is a random predictor (oracle), any node can perform PoW operations, and all the consumption (e.g., power consumption, equipment depreciation, time cost, etc.) of this process per unit time is equal for any node, i.e., any node can purchase the same amount of money to obtain the same quality "PoW computing service". The present invention has standard cryptographic assumptions (e.g., public key signatures, collective signatures, etc.). Any node can obtain the correct and legal creation block. Any node has a fixed identity: consumers (or consumer nodes) or candidates (or candidate nodes). The consuming node generates a transaction and submits it to the candidate node, and the leader election will select a current round of leader (if there is no bifurcation) among the candidate nodes. The leader node then packages and broadcasts the legal block. In addition, all nodes are also classified into honest nodes and malicious nodes according to attributes (honest or malicious).

Network model

There are N nodes in the network (N candidates and m consuming nodes, i.e., n=n+m), and the honest nodes are considered to form a well-connected synchronous point-to-point (P2P) overlay network, and the size of any message (e.g., block or single transaction) that needs to be transmitted in the system is fixed. Thus, when one honest node broadcasts or transmits a message, the upper limit of the reception delay of any other honest node is also fixed and denoted as Δ. And all messages are transmitted between honest nodes via the gossip protocol.

Threat model

The invention tolerates that the computing power of all malicious nodes is less than 1/3 of the total network. The behavior of malicious nodes may deviate arbitrarily from our protocol. Further, malicious nodes may destroy any honest candidate nodes, but they cannot destroy honest consuming node nodes (which is also almost impossible in real situations). However, a malicious node may create any number of malicious consuming nodes (e.g., through a witch attack). Furthermore, they cannot break our standard cryptographic assumptions in polynomial time.

Consumption node preferences

The consuming node has some preference for the selection of candidate nodes and this is not considered by other blockchain protocols. Assume thatRepresenting consumption node i e m]For candidate node j e [ n ]]Is used (the preference degree represents the degree to which i wishes to submit his transaction data to j, e.g., n=10, x is the best candidate node for i, y is the second most preferred candidate node for i, then->) Then->Indicating that i prefers j to be more than k. And we define LCP _i A list of all candidate nodes ordered according to his preferences is i.

The invention needs to realize high throughput, low transaction delay, certain expandability, low candidate threshold (facilitating decentralization), simplicity, effectiveness and safety, and is specifically described as follows:

Validity and security: the transaction of the honest consumption node can finally obtain the confirmation of the blockchain in the invention; the invention can resist double-flower attack (namely, 51% attack), sybil attack and SPOF attack.

Throughput: when the system parameters such as the scale of the network (i.e., the number of nodes), bandwidth, etc. are determined, the average number of transactions acknowledged by the system per unit time is defined as the throughput of the system. The invention needs to improve the throughput of the system to the maximum extent.

Transaction delay: when system parameters such as the size of the network (i.e., the number of nodes), bandwidth, etc., are determined, the transaction delay is defined as the time that elapses from the transaction generation (i.e., the initiation of the transaction by the consuming node) until the transaction is ultimately validated on the blockchain. The present invention requires minimizing transaction delays.

Scalability: the invention needs to solve the conflict and coupling between the safety and the expandability in the PoW type consensus, and ensure that the expandability is improved as much as possible without sacrificing the safety. Furthermore, the present invention should be able to embed other slice protocols and layer 2 protocols to implement "scale-out".

Candidate node thresholding and decentralization: in the present invention, the candidate node assumes the obligation of the verifier, i.e., the threshold of the candidate node implies a condition to join the consensus mechanism (e.g., joining a licensed blockchain requires approval by all other members). The present invention requires that this threshold be reduced as much as possible to achieve a higher degree of decentration while ensuring that security is not sacrificed.

Consumption node preferences: order theFor the popularity of candidate j, then the Avalon system needs to satisfy: the probability that node j becomes a leader node has a positive correlation with the popularity of j.

To solve the above problems, the present invention introduces an Avalon scheme, a novel distributed consensus protocol. Avalon decouples PoW consensus and leader election, greatly weakens the correlation of security and block interval, improves system throughput, and reduces transaction delay. The Avalon's consensus mechanism, proof-of-Market (PoM), allows the Market and consumer nodes to decide on the outcome of leader election, poM without the assistance of any other consensus (e.g., BFT), which makes its code implementation very simple. And under the same environmental conditions, avalon is as safe as the middle blockchain.

The core content of Avalon-PoM is divided into leader election and consumption nodes PoW. In a leader election, the consuming node indicates that he supports this candidate node in the leader election by submitting legal transaction data to the candidate node he supports, i.e., the legal transaction data is considered to be a "vote (vote)". Only candidate nodes that have collected a predetermined number of vots become leader nodes. The consumer node PoW is to prevent malicious behaviour, which requires that a transaction is considered legal vot only when the consumer node completes a specified consumer node PoW. The present invention provides a set of incentives to match PoM and proves statistically unprofitable attacks. Notably, even if an inadvisable attacker insists on the attack, his attack is not less difficult than a PoW-like blockchain under the same settings (e.g., same hash operation difficulty and natural bifurcation rate) because PoW and PoM have the same defense mechanisms and defense characteristics. Irrational attacks only affect fairness and decentralization. Furthermore, the Avalon protocol has a larger controllable space than the PoW-type blockchain, such as by adjusting some parameters of PoM to suppress the natural bifurcation rate. In summary, poM implements the security of PoW-like blockchains while embodying consumer nodes and market willingness to leader election (rather than complete randomization), and analysis shows that the PoM mechanism can effectively suppress the constraint of blocktransfer delay on blockspacing and security. Thus, poM optimizes the throughput, transaction delay, decentralization, fairness, and energy consumption of the Avalon system.

Furthermore, a p2p network was modeled by omnetpp, which has a bandwidth of 30Mbps and a delay of 100 milliseconds, which contains over 1,000 distributed peer nodes. Experiments examined the effect of various parameters of the protocol on Avalon performance. The results show that with most parameter settings, avalon can achieve excellent throughput and transaction delay (beyond 4,000TPS and 10 s-40 s of transaction delay). Thus, under similar network environments, the throughput of Avalon reaches the level of 10 Xthe present blockchain-NG, 5 XByzCoin, and 4 XAlgornd, and the transaction delay reaches the same level as Algornd, and is half that of the present blockchain-NG and ByzCoin. The rollback rate of the transaction in the Avalon system after waiting 3 blocks is 0%, and in the Avalon blockchain, there is almost no bifurcation with a height exceeding 1, which indicates that the scheme has higher security. And Avalon maintains excellent decentration at all parameter settings.

Avalon System design

Fig. 1 illustrates the hierarchical system architecture of Avalon. At the data layer, the data structure of the Avalon blockchain is similar to that of a traditional medium blockchain, but the original transaction information is replaced by the vot information. The Avalon blockchain is formed by individual Avalon blocks, and the Avalon blocks comprise vot information, candidate node position information, leader digital signature and the like, wherein the vot information further comprises: UTXO transaction information, consumer node PoW certificates, consumer node digital signatures, former block hash pointers, etc. At the network layer, avalon adopts gossip transport protocol to ensure reliable broadcasting of messages in the distributed network. At the consensus and incentive layer, a core consensus algorithm-PoM mechanism of Avalon is provided, and a task of consensus incentive is completed by assisting an incentive strategy based on game theory and a cold and hot candidate node mechanism.

The relation among the hierarchical modules is as follows, and all candidate nodes in the system operate leader election modules of Avalon; meanwhile, the consumption node executes a consumption node PoW module, the consumption node PoW needs to input UTXO transaction information, the UTXO transaction information can output legal volte information, the consumption node submits the legal volte information to corresponding candidate nodes (in the process, a game theory excitation module and a cold and hot candidate node mechanism also work at the same time, and the excitation node executes the steps); the output of leader election, i.e., the node that becomes the leader, integrates the collected vot information into legal Avalon block data and invokes the gossip transport protocol to broadcast this block. To this end, a new legal Avalon block is agreed upon by each node in the system.

Market proving (PoM) mechanism

PoM consensus mechanisms include vot, election, and block broadcasting. The PoW of the consuming node constitutes the vot mechanism. Unlike classical leader election, a "one-ticket multiple throw (one voter for multiple investments, OVMI)" mechanism is proposed that eliminates the limitation of the PoW mechanism on system throughput and simplifies our protocol. The consuming nodes use the transaction data they submit as vot to select their favorite candidate nodes as leader nodes, while the cost generated by the consuming node PoW can filter out malicious nodes.

Leader election for candidate nodes

The election of the next leader node will occur immediately after any on the previous leader node. This is an uninterrupted process. Fig. 2 shows an election process in the Avalon system, which is specifically as follows:

first, in one round of election, all candidate nodes participating in the election "freeze" a certain number of digital assets $ in their local ledger _d As a "deposit", this information is then digitally signed and broadcast. The consuming node receiving the message can be according to $ _d And other factors that select candidate nodes that they support and submit local transaction data to him (or them). Note that $ _d Is temporarily frozen from the candidate node's ledger, and the candidate node cannot use, transfer this candidate unless the leader election fails. If the candidate node becomes a leader, $ _d The specific amount of distribution to be distributed to the consumer node to which the transaction information was previously submitted is proportional to the amount of the transaction submitted by the consumer node.

The candidate node can immediately perform the next round of leader election after the previous round of leader node broadcasts the block and compare the candidate node with the desired candidate node _d The value is broadcast into the network. Total transaction amount ta collected when candidate node i is current _i Upon reaching a certain threshold (which relates to the total amount of transactions collected by the preceding leader node), he will immediately verify all collected transactions. If no other candidate node reaches the threshold earlier than i (i.e., i does not receive any other node broadcast blocks in this round), then he will immediately package and broadcast all legitimate transactions. The candidate node of the received block knows that the candidate node fails in the round of election, and can immediately join in the next round of election.

Many details are hidden in the leader election, here we first describe the transaction amount threshold TAT that the candidate node needs to collect. If the transaction amount collected by the previous leader node l is ta _l (i.e., sum of amounts contained in vot collected is ta _l ) The transaction amount of the next leader is at least lambda ta _l . The coefficient lambda corresponds to the difficulty of hash operation in the block chain system to a certain extent, lambda is controlled by the transaction amount and trend of the market during the period, for example, the transaction amount of the predecessor is ta _l =100, the current market demand is in the rising phase, λ=1.07, then the transaction amount collected by the next leader must be greater than 107. Lambda mainly acts to control the gap of the blocks.

Consumption node PoW scheme

When the candidate node executes the leader election, the consumption node outputs corresponding vot data through the consumption node PoW to be submitted to a specific candidate node in the leader election, so that normal execution of the leader election process in the Avalon system is ensured. The specific scheme is as follows:

the message structure submitted by the consuming node to the candidate node is: the digital signature of the consuming node, transaction information, hash information of the latest block (for preventing the node from executing the consuming node PoW in advance), the number of candidates supported, and the random number x. This type of message is called the vot of the consuming node, as shown in fig. 3.

The consumer node must perform PoW operation on its vote, i.e. adjust the random number x so that the hash value, transaction amount, etc. of the entire consumer node vote meets the current consumer node difficulty condition (similar to the block hash of the middle-cost blockchain). Only the vot of the PoW is legal at the prescribed consumption node difficulty.

One-ticket multiple throw (OVMI): the Avalon system does not prevent consumer nodes from submitting different vots to multiple candidate nodes in a round of elections as long as the vots they submit are legitimate.

Consumption node PoW difficulty (hereinafter d _c Representation). Similar to the hash operation difficulty of PoW type block chain, d _c It is the difficulty of the consumer node to perform the PoW operation. d, d _c Is a variable related to the total hash capability of all current consuming nodes, and d _c Related to the upper bound of the transaction amount submitted by the consuming node. However, unlike the definition of the hash operation difficulty of PoW, here we consume node difficulty d in definition 1 from a new perspective (economic perspective) _c Related to consumption node PoW, transaction amount, etcIs defined as follows.

Definition 1 (consumption node PoW difficulty): based on the assumption above, if a node invests $ _PoW The funds of the amount are used for consuming node PoW (mainly used for e.g. electric power, equipment lease, manpower, etc. of hash operation), and the current consumption node difficulty is d _c Then the upper limit of the expected transaction amount he can legalize is d _c $ _PoW 。

For example, if the amount of a transaction is 10 units of currency, and d _c =100, then the consuming node would take 0.1 units of currency to complete the cost incurred by the consuming node PoW to make the transaction legal.

By combining definition 1 we can derive d _c Some of the characteristics of (a):

for two different transactions tx ₁ And tx ₂ The amounts are respectivelyAnd->The cost of the consumer node PoW for which it is legitimate is also different. If these costs are quantified as monetary value, i.e. +. >And->Then we have +> And->

As with the PoW-type blockchain, the computing power of the consuming nodes in Avalon also changes continuously, so d _c Corresponding changes are required according to the total calculation power of the network to keep the calculation time stable.

In addition, the consuming node PoW needs to ensure the cost linearity of the consuming node PoW, that is, when a transaction tx is split into two parts tx ₁ And tx ₂ At this time, the expected value of the consumption node PoW cost of tx should be equal to the completion tx ₁ And tx ₂ Is the desired cost of PoW. And this is easily achieved: assuming that the probability of the consuming node PoW to complete a unit currency transaction is p in one hash function call, the probability of the consuming node PoW to complete an s unit currency transaction should be in one hash function call

In fact, the PoM mechanism unhooks the hash operation from the leader node election, avoiding centralization and vicious hash operation competition. Whereas the consuming node PoW makes the PoW have the function of "punishing" the attacker.

Excitation mechanism

The consuming node PoM is not sufficient to make the Avalon system incentive compatible, so we need to introduce a companion incentive mechanism. The game theory is the key to setting a reasonable incentive mechanism, and the whole incentive mechanism of the protocol is introduced below.

In addition to the aforementioned threshold TAT of transaction amount and consumption node PoW difficulty d _c As shown in fig. 4, avalon also has the following two parameters:

calculation of excitation k _t : when a leader succeeds in a transaction with a $in uplink amount, the consuming node should give the leader a proportion of the transaction amount as "calculation incentive", i.e., k _t And (4) the cost is reduced. This is also a similar mechanism in the present blockchain.

Delay penalty k _d : if the candidate node fails in the election, he should pay k to the consuming node that supports him in the current round of election _d $is a deferred penalty for the transaction. Delay fees are a "placebo" for transaction delays caused by election failures, and may prevent some malicious attacks, which are analyzed in detail later.

This k _t And k _d The following relationship (based on game theory strategy) should be satisfied:

and

Wherein n represents the number of nodes in the system; $ for ₀ Representing the amount of the deposition; f ($) ₀ ) Indicating when a candidate is broadcasting $ ₀ At the point of the amount, a mathematical expectation of the amount of vot submitted by the consuming node is obtained; d, d _c Representing the consumption node PoW difficulty.

Calculating incentive and delay penalties pays according to the following rules:

we require candidate nodes to broadcast the signed delivery information to the whole network when opting in. If the candidate node wins, the collected transactions are immediately packed and the desired is paid to each consuming node in proportion, otherwise the transaction set is considered illegal.

Since the whole network does not agree on the transactions collected by the failure candidate nodes, we use a "claims" mechanism to assign deferred fees. The legal vot itself may be regarded as a "check" and when this "check" is used in the future for payment, the corresponding amount will be automatically deducted from the failed candidate account.

To prevent candidate node funds from being in shortage and unable to pay deferral fees, we can specify that the candidate node must freeze a certain amount of "prepaid deferral fees" in each election. And if the candidate node wins, thawing, otherwise, waiting for a plurality of rounds and automatically thawing. But a certain amount of "prepaid deferral fees" must be maintained as long as the candidate node is in the election state.

In order for the deferred fees to have a positive incentive to Avalon (i.e., malicious nodes cannot profit from the deferred fees), this requires that the mathematical expectation of the cost (investment) of the nodes on the consuming node PoW be largeThe delay fee available to him, i.e.,

cold and hot candidate node mechanism

In the previous description of Avalon, all candidate nodes may participate in leader election at the same time. However, this approach spreads the computing power of the consuming nodes and may impair the security of Avalon. Therefore, we introduce a hot-cold candidate node mechanism:

The operation process of the Avalon system is divided according to time T and divided into a plurality of periods with equal length.

Within each period T, the Avalon system randomly selects 3 special candidate nodes (via Algorand [10] election scheme, i.e., VRF-based leader election scheme) from all candidate nodes, which are called hot candidate nodes, and other candidate nodes are called cold candidate nodes.

The consuming node can only submit legal transactions to the hot candidate node unless the consuming node determines that the hot candidate node is malicious or offline.

When the consuming node determines that the hot candidate node is malicious or offline, the consuming node may submit a legal vot to the cold candidate node.

Here we explain some details of the cold and hot candidates. First, only 3 hot candidates are selected per round to ensure that the computing power of all consuming nodes is concentrated on these three candidates. Thus, a malicious node cannot perform a double-flower attack as long as its computational power does not exceed 1/3 of the full network computational power (i.e., the computational power of all consuming nodes). The process of randomly selecting popular candidate nodes is detailed in the committee election scheme of algornd [10] that uses globally non-counterfeitable and verifiable random numbers to prevent cheating. Method for discovering candidate offline with respect to consuming nodes: since the completion time of the consuming node PoW satisfies the exponential distribution, the probability of the hot candidate node coming out of the block will be continuously close to 1 as time increases. If waiting long enough for the hot candidate node to issue a new block, the consuming node may determine that the hot candidate nodes have all gone offline.

Gossip transport protocol

The Gossip protocol is a typical distributed message transmission protocol, and the bottom layer of the Gossip protocol generally adopts the UDP protocol, which can realize low-load, high-reliability, scalability and other performances at the same time, and is simple and easy to realize. The Gossip protocol is generally described as follows, where first the message transmission is initiated by a seed node that randomly picks up several neighbor points and sends the message to be synchronized to them. The node that received the message becomes the new seed node and repeats the process until all nodes in the network have received the message. The Gossip protocol has the final consistency that, in theory, there will always be a time node after which all nodes in the system receive the message to be synchronized.

Bifurcation of

We first introduce the link rules for the Avalon blockchain: all transactions within a block must be directed to the same previous block, which is also the parent of this block. Thus, when a bifurcation occurs, all consuming nodes need to select the appropriate parent tile, and candidate nodes must also select the appropriate consuming node (to ensure that transactions within their tile are directed to the same parent tile).

When forking occurs and different network partitions produce different consensus, we use the same solution as the PoW type blockchain: the longest chain is selected. Shortening of the block interval may result in an increase in bifurcation. And the probability of bifurcation depends to a greater extent on the market situation due to the mechanism of PoM. For example, if there is a "super candidate node" with strong attractive transaction capabilities (through its preferential policy, premium services, etc.), the probability of bifurcation is very low, as it is difficult for other candidate nodes to issue legal blocks at about the same time as it is. If there are multiple equally popular candidate nodes (i.e., worst case), the probability of bifurcation increases.

However, unlike blockchains, in this case our protocol can still affect the bifurcation rate by adjusting the parameters. For example, we can adjust d _c TAT orOVMI is limited (this allows the consuming node to select only a single candidate node, thereby reducing the bifurcation rate).

Avalon pseudocode

This section we use the pseudo code in algorithm 1 to elucidate the operation of Avalon.

Algorithm 1: avalon main algorithm

Input: transaction amount threshold TAT, consumption node PoW difficulty d _c Calculating excitation k _t Calculating excitation k _d 。

For any one candidate j:

step one: starting a new round of leader election, j broadcasting own desired information $ _d The method comprises the steps of carrying out a first treatment on the surface of the j collecting the transaction of the consuming node and verifying the legitimacy thereof (for example, the consuming node PoW, digital signature, UTXO record, etc.); all legal transactions are added to the transaction set TX collected by this round j. And executing the second step.

Step two: if sigma _tx∈TX $ _tx >TAT without any candidate node broadcasting a new legal block, where $ _tx And (3) representing the amount of tx, executing the third step, and otherwise executing the fourth step.

Step three: j immediately signing, packaging and broadcasting all transactions; j pays the consuming node the corresponding desired (i.e., $ _d ) These transactions are contained in TX and tax the transaction (i.e., k) from the consuming node _t ). And executing the fifth step.

Step four: if there are other candidate nodes broadcasting new legal blocks, j pays deferred fees to all consumer nodes submitting vot to him in this round (i.e., k _d ). And executing the fifth step.

Step five: if j continues to participate in the leader node election, entering the next round of election and executing the step one, otherwise exiting.

For any candidate node i:

step one: if a new legal block b, i is received, stopping the ongoing operation and executing the following operations: i updating its uncommitted information list UTX according to the latest blockchain _i The method comprises the steps of carrying out a first treatment on the surface of the i collecting candidate node position information and adjusting LCP _i (according toEvery examinee $ _d And i own preferences). And (3) entering a step two.

Step two: if all candidate nodes j epsilon LCP are traversed from high to low _i And j is a hot candidate node, then i is from UTX _i The earliest transaction tx is selected; i executing CPoW function (see algorithm 2 for details) to get legal transaction ltx =cpow (i, j, b, tx, d) _c ). i submits ltx to j. And (3) monitoring the network, if a new legal block is received, entering the first step, otherwise, continuing to execute the second step.

Algorithm 2: function CPoW (i.e., consumption node PoW)

Input: consumption node i, candidate node j, previous block b, transaction content tx, consumption node PoW difficulty d _c 。

And (3) outputting: legal transaction ltx.

Step one: generating a random number x; calculating h=hash (i, j, HASH (b), tx, x), wherein HASH (·) represents the HASH function; if H satisfies the amount $ _tx Consumption node PoW difficulty d corresponding to transaction of (2) _c And executing the second step, otherwise executing the first step.

Step two: function return ltx =sig _i (<i,j,HASH(b),tx,x>) Wherein sig _i (. Cndot.) represents the digital signature of i.

Prototype test of Avalon

In this section, we evaluate the performance of Avalon. We pay particular attention to throughput of transactions, acknowledgement delay of transactions, rollback rate of transactions, convergence of chains and decentralization (see table 1 for detailed explanation). Furthermore, two important variables are the threshold TAT of leader election and the difficulty d of the operation of the consuming node PoWhash _c . Note that for ease of understanding we redefine d here _c I.e. the number of transactions per unit time (i.e. 1 s) of a unit amount (i.e. 1 BTC) that the consuming node can legalize by the consuming node PoW, i.e. d _c The larger the node the more efficient the hash operation (this is different from definition 1, but does not affect our analysis, which is set only for ease of experimentation). We specify TAT and d _c Default values for (c) are 80,000 and 6.

Table 1.Avalon experimental parameters and their description table.

Parameters (parameters)	Description of the drawings
		Transaction Throughput (TT)	Average number of transactions confirmed per second by the Avalon system.
Transaction Confirmation Delay (TCD)	The transaction is from appearance to the average length of time that is experienced by the Avalon ul acknowledgement.
		Rollback Rate of Transaction (TRR)	After several blocks, the transaction is rolled back due to a split transmission.
Convergence of blockchain (CC)	The Avalon blockchain needs to wait for the number of blocks to converge when forking occurs.
		Decentralization	Each candidate node successfully produces a block probability.

2.1 prototype implementation

We realized a prototype of Avalon in a simulated p2p network by omnetpp. To simulate a truly global distributed network, we limit the bandwidth of all connections between nodes to 30Mbps and impose a delay of 100 milliseconds on all communication links. And to simulate transactions at the consuming node, we have collected a set of all real transactions in the middle blockchain from blocks from 500,000 to 505,000 in height. Our prototype simulates a total of 1,003 nodes, including 1,000 consuming nodes and 3 hot spot candidate nodes, each producing 5 raw transactions per second (to balance the performance of our experimental machine, we control the network size to 1,003 nodes, but this has been comparable to many classical blockchain prototypes, e.g., the midst blockchain-NG, byzCoin, etc.).

Transaction Throughput (TT)

Here we have studied TAT and d _c Effect on Avalon throughput under our experimental setup. The results of the specific experiments are shown in FIGS. 5 and 6, where d _c With the default values in fig. 5, TAT with the default values in fig. 6. TAT and d _c The increase in (a) has a positive regulatory effect on Avalon throughput, but is not readily apparent. Avalon can reach above 4,000TPS at most parameter settings. This is comparable to the average throughput of Visa, which is much higher than ByzCoin (almost 4 times the ByzCoin throughput) under the same conditions.

Transaction Confirmation Delay (TCD)

The average TCD experimental results for Avalon are shown in fig. 7, 8. Note that the TCD calculation here is based on waiting for 6 blocks to acknowledge the transaction, i.e. we use the standard acknowledgement model for the current blockchain. TAT has a clear positive correlation with TCD, and d _c Little effect on TCD (in practice weak negative correlation). The potential reason for this is that the threshold requires more transactions, however the frequency of transactions generated by the consuming node is fixed, and therefore requires waiting for a longer time (the time to wait for the consuming node to generate a transaction increases). When d _c When the consumption node is not less than 3, the consumption node can submit information to all three hot candidates basically, so d is continuously added _c Only a small increase in TCD can be brought about. In most cases, avalon's TCD will not exceed 40s, and in the best case, the transaction can be confirmed within 10s (i.e., the transaction will not be rolled back).

Overall, the TCD of Avalon far exceeds the level of PoW-type blockchain, comparable to the currently prevailing blockchain protocols, indicating that Avalon is well acceptable for delay while achieving high utility.

Transaction Rollback Ratio (TRR)

Rollback rate is an important indicator for measuring the blockchain protocol and is directly related to the security of the system. PoM is essentially a PoW-type consensus, so Avalon also has the convergence of PoW-type blockchains, which makes submitted transactions immutable after validation. Fig. 9 and 10 are related experimental results in which data were taken from blockchains that were highly over 100. TAT and d _c The increase in (2) can effectively inhibit TRR. But only when the transaction is confirmed after waiting more than 3 blocks, no rollback phenomenon exists in the Avalon system. Avalon waits 6 blocks to acknowledge the transaction as in the middle blockchain, so it is very secure.

Convergence of blockchain (CC)

The convergence experiment of the chain judges the safety of the blockchain by recording the number and the height of bifurcation. The experimental results are shown in fig. 11 and 12. Typically, most of the prongs have a height of 1, with only a few prongs reaching a height of 2. Increasing TAT is effective in inhibiting bifurcation. As can be seen by comparing fig. 10 and 12, d _c The increase in (c) can inhibit TRR without affecting bifurcation rate. This is due to d _c The increase in (c) allows the consuming node to submit the same transaction to more candidate nodes, and the transaction may be confirmed by the blocks of other candidate nodes, even if there is rollback.

Decentralization

The decentralization experiment verifies the fairness and safety of Avalon. If the narrowing of the block interval gives the node a "run-in" advantage, the security of the system is greatly compromised. Our experimental results are shown in tables 2 and 3, where all experimental data were obtained at least 3 times in steady runs, indicating that Avalon was found in different TATs and d _c Good decentration is maintained under set-up. The winning rate of hot candidates is maintained in the interval of 25% -40%, TAT and d _c The variation of (c) does not bring about a significant fluctuation of the winning rate. Considering the TCD (10 s to 40 s) and bandwidth setting (30 Mbps) of Avalon, the PoW-type blockchain under equal conditions will have serious congestion and "run-in" advantages (this is also a PoW-type blockchain)The chain cannot increase the chunk-out capacity or reduce the root cause of the chunk-out interval), but Avalon can ensure the normal operation of the system.

Table 2. A table of the winning rate (i.e., the blocking rate) of the hot candidate node versus the threshold TAT.

TAT	20000	40000	60000	80000
					Thermal candidate node 1	30.20％	28.60％	29.50％	25.60％
Thermal candidate node 2	34.00％	30.70％	31.10％	43.30％
					Thermal candidate node 3	35.80％	40.70％	39.40％	31.10％

TABLE 3 winning rate (i.e., blocking rate) of hot candidate nodes and consuming node PoW difficulty d _c Is a relationship of (3).

d _c	4	6	8	10
					Thermal candidate node 1	23.90％	25.60％	26.40％	26.70％
Thermal candidate node 2	37.50％	43.30％	34.10％	33.30％
					Thermal candidate node 3	38.60％	31.10％	39.50％	40.00％

Those skilled in the art will appreciate that the systems, apparatus, and their respective modules provided herein may be implemented entirely by logic programming of method steps such that the systems, apparatus, and their respective modules are implemented as logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc., in addition to the systems, apparatus, and their respective modules being implemented as pure computer readable program code. Therefore, the system, the apparatus, and the respective modules thereof provided by the present invention may be regarded as one hardware component, and the modules included therein for implementing various programs may also be regarded as structures within the hardware component; modules for implementing various functions may also be regarded as being either software programs for implementing the methods or structures within hardware components.

The foregoing describes specific embodiments of the present invention. It is to be understood that the invention is not limited to the particular embodiments described above, and that various changes or modifications may be made by those skilled in the art within the scope of the appended claims without affecting the spirit of the invention. The embodiments of the present application and features in the embodiments may be combined with each other arbitrarily without conflict.

Claims

1. A distributed ledger method based on a novel PoM consensus algorithm, comprising:

step S3: integrating information supporting the hot candidate node into legal Avalon block data by submitting legal transaction data to the hot candidate node by the consuming node, storing the legal Avalon block data in a new legal block, broadcasting the current block in a P2P network by calling gossip transmission protocol, and realizing that a new legal block is agreed by each node;

the step S1 includes: randomly selecting preset hot candidate nodes from all candidate nodes in each period T through a leader election scheme of VRF, and setting other candidate nodes as cold candidate nodes;

the legal transaction data adopts: only when the consumption node completes the preset consumption node PoW consensus, the transaction is considered to be in accordance with the preset requirement and belongs to legal transaction data;

2. The distributed ledger method based on the novel PoM consensus algorithm according to claim 1, wherein leader election and consumption node PoW consensus are decoupled, thereby weakening the correlation of security and block out interval, improving throughput, reducing transaction delay.

3. The distributed ledger method based on the novel PoM consensus algorithm according to claim 1, wherein the step S1 further comprises: and the consumption node submits legal transaction data to the hot candidate node so as to support the current hot candidate node, and when the consumption node judges that the current hot candidate node is malicious or offline, the consumption node submits legal transaction data to the cold candidate node which is randomly selected.

4. The distributed ledger method based on the novel PoM consensus algorithm according to claim 3, wherein the consuming node determines that the current hot candidate node is malicious or offline employing: and when the hot candidate nodes do not release new blocks after the preset time, the consumption node determines that the hot candidate nodes are all offline.

5. A distributed ledger system based on a novel PoM consensus algorithm, comprising:

module M3: integrating information supporting the hot candidate node into legal Avalon block data by submitting legal transaction data to the hot candidate node by the consuming node, storing the legal Avalon block data in a new legal block, broadcasting the current block in a P2P network by calling gossip transmission protocol, and realizing that a new legal block is agreed by each node;

the module M1 includes: randomly selecting preset hot candidate nodes from all candidate nodes in each period T through a leader election scheme of VRF, and setting other candidate nodes as cold candidate nodes;

6. The distributed ledger system based on the novel PoM consensus algorithm according to claim 5, wherein leader election and consumption node PoW consensus are decoupled, thereby weakening the correlation of security and out-block interval, improving throughput, reducing transaction delay.

7. The distributed ledger system based on the novel PoM consensus algorithm according to claim 5, wherein the module M1 further comprises: and the consumption node submits legal transaction data to the hot candidate node so as to support the current hot candidate node, and when the consumption node judges that the current hot candidate node is malicious or offline, the consumption node submits legal transaction data to the cold candidate node which is randomly selected.

8. The distributed ledger system based on the novel PoM consensus algorithm according to claim 7, wherein the consuming node determines that the current hot candidate node is malicious or offline employing: and when the hot candidate nodes do not release new blocks after the preset time, the consumption node determines that the hot candidate nodes are all offline.