A computer-implemented method for exchanging information through DAG / hash graph consensus via a gossip protocol

Abstract

To provide a method of, a system for, and a network for exchanging information relating to an action list or transaction among distribution computer nodes.SOLUTION: A method according to the present invention has a step in which a first given node randomly selects a second given node to transmits reference information to it, a step in which it is determined based on reference information which node event is prior to others, a step in which, if the reference information indicates that an event of the first given node is prior to an event of the second given node, cancel information is transmitted to the first given node to prevent the first given node from transmitting a DAG-event information to the second given node, and a step in which, if the reference information indicates that the event of the second given node is the event of the first given node, the first given node transmits the DAG-event information to the second given node.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates to the field of exchange of information, particularly related to lists of actions or transactions, between distributed computer nodes connected within a particular network.

[0002] The present invention relates to systems and methods for implementing Directed Acyclic Graph (DAG) consensus algorithms such as hash graph consensus algorithms via gossip protocols. [Background technology]

[0003] Hashgraph is known in the art as a new way to reach consensus in a decentralized network. It is often referred to as a voting algorithm. Hashgraph is specifically a Directed Acyclic Graph (DAG).

[0004] The goal of Hashgraph is primarily to reach consensus among nodes in a distributed system. The Hashgraph algorithm is a method for solving the Byzantine Generals Problem.

[0005] Furthermore, the hashgraph algorithm prevents an attacker from shutting down or freezing the network, thus preventing it from reaching consensus.

[0006] One idea of ​​this application is to have an efficient way to exchange information between nodes and also to support systems for hash graphs or other DAG systems that need to exchange DAG information.

[0007] FIG. 1 shows a conventionally developed schematic hash graph. A hash graph is a set of computer nodes (N A , N B , N C , ND , N E ) network. Each computer node (N A , N B , N C , N D , N E ) can record what are called Events 1, shown as circularized, and shown in more detail in Figure 2. Events are generally sequences of information / bytes and memory that are then hashed. Events are recorded by nodes. Events consist of pieces of information that are distributed across the network. Information is generally hashed to make it secure. Thus, Hashgraph allows for secure information to be distributed within the network.

[0008] Events are signed by nodes.

[0009] In a hashgraph implementation, a node then randomly selects another node and sends it all of its information, which is hashed over the communication. This is repeated until the information has propagated through the network.

[0010] Hashgraph implies an infinite number of transmissions of information within the network. Additionally, in the hashgraph algorithm, events always contain a timestamp that leads to a consensus timestamp within the network.

[0011] Referring to Figure 2, the event information is The hash from SelfParent2 received from SelfParentNode, shown vertically in Figure 2. The hash from SelfParent2 is associated with the previous event in the memory of SelfParentNode; Hashes from other parents 3 received from other parent nodes shown diagonally in Figure 2. Hashes from other parents 3 are those related to previous events in the memory of the other parent nodes; Information on current events1; and Contains the timestamp of the current event. Summary of the Invention [Problem to be solved by the invention]

[0012] The objective of the present invention is to limit / reduce the number of transmissions of information and to reduce the use of computer resources for calculation, storage and transmission of information.

[0013] Another general object of the present invention is the efficient exchange of information between computer nodes and preventing nodes from going into an infinite echo where they transmit information back and forth forever.

[0014] Another objective is node security and the ease of determining what information needs to be shared between nodes without transmitting unnecessary information between them. [Means for solving the problem]

[0015] To achieve these objectives, the present invention relates to a computer-implemented method for exchanging information among distributed computer nodes connected in a particular network, the method implementing a DAG (Directed Acyclic Graph) consensus algorithm, such as a hash graph consensus algorithm, through the use of gossip in a gossip protocol, the method comprising: The first granting node randomly selects a second granting node and transmits the reference information; The method preferably comprises: determining which node events are prior to the others based on the reference information; If the reference information is Noi The second node Noi If you suggest it was before the vent, sending cancellation information to the first Granting node, thereby preventing the first Granting node from sending DAG-event information to the second Granting node; If the reference information is Noi The first node Noi If you suggest it was before the vent, causing the first granting node to send DAG-event information to the second granting node; and preferably, then causing the second granting node to transmit the DAG-event information to the first granting node, And so on to other nodes to synchronize and propagate the DAG-event information across the network.

[0016] More precisely, depending on the aspect, the method comprises: causing the second Granting node to send cancellation information to the first Granting node; When the second granting node receives the reference information from the first granting node, If the second granting node has previously sent reference information to the first granting node, The first granting node does not transmit DAG-event information to the second granting node.

[0017] This DAG-event information refers to the information exchanged related to an event in the DAG. In particular, each piece of information is connected via a hash pointer, that is, the hash of an event is a unique hash (fingerprint), and the self-parent and other parent hash fingerprints are contained within the event and act as unique identifiers / pointers between events in the DAG.

[0018] More advantageously, the method can limit the number of times of transmitting information via the cancellation information, so that the node can stop communicating information, which implies reducing the use of computer resources in computing, storing and transmitting information.

[0019] The idea behind the present invention is to have an efficient way of exchanging information between nodes (gossip about gossip), so the wavefront method is used as a support system for hash graphs or other DAG systems that need to exchange DAG information.

[0020] Furthermore, it limits the amount of information by transmitting the state of each node's chain of events, indicated by the number of references / altitudes. In this way, the node can determine if the last event in the chain of events is after or before another node's event chain.

[0021] Additionally, cancellation can prevent infinite echoes between two nodes.

[0022] Additionally, the wavefront method easily secures nodes to determine what information needs to be shared between nodes by sharing a wavefront (discussed in more detail below) without transmitting unnecessary information between them.

[0023] Depending on other aspects of the method taken individually or in any other technically possible combination, The reference information relates to the order of node events, and / or The order of the node events is determined by tracking the ordinate values ​​of all events in succession; and / or The ordering of node events is determined by having each node keep track of an integer value called an altitude that increases for each event generated within the node; and / or The reference information is sent via tidal waves, and / or The cancellation information is sent via breaking waves, and / or the second transmission of DAG-event information from the second Granting node to the first Granting node is a second and final wave of DAG-event information; and / or The method includes transmitting another cancellation information if the node receives information that is not in a predetermined sequence of information exchange and / or if the node receives information from a node with which it has already exchanged information.

[0024] The invention further relates to a system for exchanging information, comprising means for carrying out the steps of the method according to the invention.

[0025] More generally, the present invention also relates to a system for exchanging information among distributed computer nodes connected in a particular network, the system implementing a DAG (Directed Acyclic Graph) consensus algorithm, such as a hash graph consensus algorithm, through the use of gossip in relation to a gossip protocol, the system comprising: means for causing a first granting node to randomly select a second granting node and transmit reference information thereto; The system preferably comprises: means for determining which node events precede others based on the reference information; If the reference information is Noi The second node Noi means for transmitting cancellation information to the first granting node if the first granting node indicates that the event is occurring before the event, thereby preventing the first granting node from sending DAG-event information to the second granting node; and If the reference information is Noi The first node Noi means for causing the first granting node to send DAG-event information to the second granting node if the first granting node indicates that the event is before the event; and preferably, means for causing the second granting node to transmit the DAG-event information to the first granting node, And so on to other nodes to synchronize and propagate the DAG-event information across the network.

[0026] The advantages described above for the method also apply to the system for carrying out the method.

[0027] Depending on other aspects of the system taken individually or in any other technically possible combination, The reference information relates to the ordering of the node events; and / or the system comprising means for tracking the change in ordinate value of each event to determine the order of the node events; and / or the system comprises means for causing each node to track the change of an integer value, called altitude, which increases for each event generated at the node in order to determine the order of the node events; and / or the system comprises means for transmitting the reference information via tidal waves (8) and means for transmitting the cancellation information via breaking waves (9); and / or the system comprises means for limiting the number of transmissions of DAG-event information between the second Granting node and the first Granting node; and / or The system comprises means for transmitting further cancellation information if the node receives information that is not in a predetermined sequence of information exchange and / or if the node receives information from a node with which it has already exchanged information.

[0028] Another object of the invention is a network comprising one or more central units, in particular one or more computerized central units, and connections to additional command units, in particular computerized command units, which execute transactions, the central unit comprising the system according to the invention. [Brief explanation of the drawings]

[0029] The present invention will now be described in detail through non-limiting descriptions and drawings that illustrate the invention. [Figure 1] FIG. 1 shows a schematic diagram of a hash graph in a prior art method. [Figure 2] Figure 2 shows a schematic diagram of events in the hashgraph of Figure 1. [Figure 3] FIG. 3 shows a schematic diagram of a system according to an aspect of the present invention. [Figure 4] FIG. 4 shows a schematic diagram of a hash graph in a method according to an aspect of the present invention. [Figure 5] FIG. 5 shows a detailed diagram of the exchange of information between two nodes in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0030] The present invention relates to the exchange of information.

[0031] The present invention implements a DAG consensus algorithm over a gossip about gossip protocol, such as a hash graph consensus algorithm over a gossip about gossip protocol.

[0032] The means of the invention distribute information via a gossip protocol, transmitting information about data received from other nodes within a particular network 5 .

[0033] The present invention relates to a computer-implemented method and also to a corresponding system 5a and network 5. The method according to the invention allows for processing and exchanging information, such as what is needed for blockchain transactions and hashgraph applications.

[0034] DAGs, such as hashgraphs, are used to exchange information between distributed computer nodes connected within a particular network. This information is secure and hashed. The information transmitted in a DAG / hashgraph is called DAG-event information.

[0035] Gossip protocols suggest that computer nodes transmit information to other nodes and have that information in memory.

[0036] The method of the present invention includes steps for determining the order of node events to determine which node events are earlier and which are later.

[0037] In this embodiment, the order is not based on time. This embodiment does not require time consensus. Advantageously, this prevents the problem of having echo computer nodes that are on the same time.

[0038] By determining the order of events, it is possible to know which events are earlier and which events are later, thus providing more accurate event information. This determination is preferably not limited to having the timestamps of the nodes in the events.

[0039] In accordance with aspects of the present invention, the order of node events is determined by tracking the change in ordinate value of each event. Advantageously, the use of ordinate values ​​prevents problems with having echo computer nodes that are on the same time.

[0040] In particular, the ordering of node events is determined by having each node track the change of an integer value called the altitude, which increases for each event generated at the node. This altitude plays a role in the processing of information.

[0041] In another step of the method of the present invention, the first giving node N1 randomly selects the second giving node N4 and transmits the DAG-event information to it within the network 5. More generally, the system is arranged to cause the first giving node N1 to randomly select the second giving node N4 and preferably transmit the DAG-event information (6, 7).

[0042] All nodes N0, N1, ..., N5, ... in network 5 process the DAG-event information (6, 7) until it is propagated within network 5.

[0043] The present invention aims to prevent some transactions of DAG-event information, but still have related information propagated within the network 5 .

[0044] According to the present invention, the method includes a step in which, when the first assigning node N1 randomly selects the second assigning node N4, the first assigning node N1 transmits specific reference information (8) to the second assigning node N4 before transmitting any DAG-event information (6, 7). The transmission of this reference information is called a tidal wave. A tidal wave is a flow of information.

[0045] This tidal wave defines an altitude value for each node. Each node tracks the changes in the event-DAG. This means that each node has a current altitude that is known by the node for each node.

[0046] In particular, the reference information (8) relates to the order of node events. More precisely, the reference information 8 comprises the previous altitudes, here a list of all altitudes for each node.

[0047] Furthermore, the method according to the present invention includes a step of transmitting a cancellation message (9) to the first node N1 under certain circumstances to prevent further transmission of the DAG-event information 6 from the first node N1 to the second node N4. The transmission of the cancellation message 9 is called a breaking wave. This breaking wave is a flow of information transmitted in response to return the node to its initial state.

[0048] The cancellation information 9 is sent when the second node N4 receives the reference information 8 from the first node N1 if the event of the second node N4 is not before that of the first node N1 in terms of the reference information 8. This is easily done when analyzing the altitude.

[0049] In a preferred embodiment, the cancellation information 9 goes directly from the second grant node N4 to the first grant node N1.

[0050] Additionally, the method includes preventing the first grant node N1 from sending the DAG-event information to the second grant node N4 as a result of the cancellation information.

[0051] In other words, the second grant node N4 transmits cancellation information to the first grant node N1, and as a result, the first grant node N1 does not transmit DAG-event information to the second grant node N4.

[0052] The advantage is that this method can limit the number of transactions through cancellation information, which results in nodes stopping communication of information, which implies reducing the use of computer resources in the calculation, storage and transmission of information.

[0053] In view of the reference information (8), if the event of the second giving node N4 is before the event of the first giving node N1, the transmission of the DAG-event information continues.

[0054] In particular, the second giving node N4 sends DAG-event information, including the events it has in memory, to the first giving node N1. This first transmission of DAG-event information is called First Wave 6. This First Wave is a well-known flow of information known to those skilled in the art. The DAG-event information in First Wave 6 includes a list of all events that the second giving node N4 has in memory.

[0055] More particularly, after the first grant node N1 receives the DAG-event information from the second grant node N4, the first grant node N1 transmits the DAG-event information including the events it has in memory to the second grant node N4. This second transmission of the DAG-event information is called a second wave (7). This second wave is a well-known flow of information known to those skilled in the art.

[0056] After the first and second waves, the exchange of information between the first and second providing nodes is stopped.

[0057] This is a second way to prevent infinite echoing of transmitted information, i.e. when a second wave is received / transmitted the wave state for this node-id to return to the initial state.

[0058] In a more preferred embodiment, the method includes the step of sending another (second) cancellation message if the node receives information that is not in a predetermined sequence of information exchange, in particular in the sequence of reference information (8), then the first wave (6) and then the second wave (7) of DAG-event information, in particular the cancellation message is of the same type as before, but is triggered, for example, if the second granting node N4 receives the second wave first.

[0059] According to another embodiment, the method includes a step of transmitting another (third) cancellation information if the node receives information from a node with which it has already exchanged information. This cancellation information is in particular of the same type as before, but is triggered, for example, if the second granting node N4 subsequently receives another reference information 8 from the first node N1. Advantageously, these steps further limit redundancies / echoes in the transmission of information.

[0060] The present invention further relates to a system 5a for carrying out the above-mentioned method. This system 5a is a computerized system comprising hardware and software modules configured to carry out the above-mentioned method, which may be called a wavefront system 5a.

[0061] This system 5a is capable of exchanging information between distributed computer nodes connected within a particular network, and executes a DAG / hash graph consensus algorithm via gossip on a gossip protocol.

[0062] This system 5a comprises means for determining the order of node events, in order to determine which node events are before and which are after, i.e. means for the first assigning node N1 to randomly select the second assigning node N4 and send reference information thereto, if possible, DAG-event information (6, 7); and means for propagating DAG-event information (6, 7) within the network 5.

[0063] According to one aspect, the system 5a comprises: means for causing the second Granting node N4 to send cancellation information to the first Granting node N1 when the second Granting node N4 receives reference information from the first Granting node N1 if the second Granting node N4 has previously sent reference information to the first Granting node N1; Consequently, the first granting node N1 includes means for preventing the first granting node N1 from sending DAG-event information to the second granting node N4.

[0064] Furthermore, the system 5a comprises means for transmitting further cancellation information if the node receives information that is not in a predetermined sequence of information exchange and / or if the node receives information from a node with which it has already exchanged information.

[0065] The advantages described above for the method also apply to the system for carrying out the method.

[0066] Another object of the invention is a network comprising one or more central units (10), in particular one or more computerized central units (10), and connections (12) to additional command units (11) for executing transactions, in particular computerized command units, central units comprising the system 5a described above. This is called a Wavefront network. This network is connected to the additional command units via the Internet.

[0067] The advantages described above for the method and system also apply to a network 5 including a corresponding system 5a.

[0068] (Example) The embodiment is explained in detail in FIGS.

[0069] 4 shows a hash graph of gossip information showing the information flow between network nodes N0, N1, N2,..., N5. Other nodes are preferably in the same network, however, not shown.

[0070] In a finite time, all nodes in the network can construct the same event-DAG.

[0071] Each vertical line represents a computer node, each circle represents an event, and the lines between events represent the communication of events between nodes.

[0072] Events seen only by each node are in solid white circles. Events seen by more than one node are in dashed white circles. The last event of each node is in a flat circle.

[0073] The final event we create is a dotted circle.

[0074] The hash graph algorithm uses a gossip protocol called "gossip about gossip" to propagate information between nodes, where a first giver node N1 sends all the information of the communication it knows to a randomly selected second giver node N4, allowing node N4 to build the same hash graph as node N1.

[0075] In the network 5 of the present invention, a protocol called Wavefront is used to exchange information between two nodes, ensuring that nodes N1 and N4 only need to communicate three times to exchange the state of the graph.

[0076] Each node tracks the change in an integer value called altitude. Altitude increases by 1 for each event generated by the node. Each node stores its current estimate of the altitude for each node in the network.

[0077] By exchanging information about altitude between two nodes, they can both understand which event came first.

[0078] To determine which event is the current event and which event is the received event, subtract this value greater than 0. That means the current event is before the received event, and vice versa if the received event is before the current event. JPEG0007727277000001.jpg1976 If the received event is before the current event.

[0079] Wavefront information exchanges four states. 1. Node N1 randomly selects node N4 and transmits a list of the last known altitudes for each node. This state is called a tidal wave. 2. Node N4 receives a tidal wave from node N1. If, based on the tidal wave, node N4's event is not before node N1's event, then node N4 sends what is called a breaking wave to node N1. Otherwise, node N4 returns a list of all events before node N1's tidal wave. This state is called the first wave. 3. If node N1 receives the first wave from node N4, it returns a list of all events before node N4. Reaching this state ends the wavefront exchange. 4. If node N1 or N4 receives a breaking wave, the wavefront communication is dropped.

[0080] Regarding state 2, in a change or combination, if node N4 has already exchanged information with node N1 through the previous tidal wave, node N4 sends a breaking wave. Additionally, in a change or combination, if node N4 has received the first and second waves from node N1, node N4 sends a breaking wave.

[0081] This prevents both nodes from getting into an infinite echo where they send information back and forth forever.

[0082] In a network, a node often has many simultaneous wavefront connections, and so occasionally receives the same package of events from other nodes. It simply drops any duplicate events it receives.

[0083] The examples in FIGS. 4 and 5 are diagrams showing events and altitude values.

[0084] Node 1 synchronizes with node N4. Event b5 (colored) is the last event generated by node N1, and all previous events known by b5 are below the dash-dotted line.

[0085] Event e5 (colored) is the last event generated by node N4, and all previous events known by this event are below the dashed line.

[0086] All events known by both b5 and e5 are shown as dotted circles.

[0087] The wavefront from node N1 (marked with the dashed line of dots above) is a list of all altitudes of the most recent known events.

[0088] This means that the wavefront node N1 is: JPEG0007727277000002.jpg1686

[0089] Node N4 marked with a blue dashed line has the following formula: JPEG0007727277000003.jpg1294

[0090] The communication steps are as follows:

[0091] The first node N1 is initially in the idle state at the bottom of the figure. The first node N1 starts gossiping towards the second node N4, which is in the initial state. 1. Initial wavefront

[0092] Node N1 connects to node N4 and is connected to wavefront W 1,b5 to node N4.

[0093] This message is marked as a tidal wave type, which causes node 4 to go from the idle state to the tidal wave state T. 2. If a message marked as Tidal Wave type is received

[0094] Node N4 receives the wavefront and decomposes the wavefront W 4,e5 Compared with W 1,b5 This means that events e3, e4, e5, f3 and f4 are returned to node N1.

[0095] This message is marked as a first wave type, which causes node N1 to go to first wave state F. 3. If a message is received that is marked as the first wave.

[0096] Node N1 receives the wavefront from node N4 and accumulates events not known by node N1.

[0097] The received wavefront is compared to the wavefront and events before the received wavefront are collected and returned to node N1.

[0098] In this example, the collections are a1, b1, b2, b4, b5, c2, c3 and d3.

[0099] This message is marked as a second wave type, which causes node N4 to go to second wave state S.

[0100] If node N4 has already sent a message of the first wave to node N1, node N4 sends a message marked as breaking wave type to node N1 and changes the state of communication from node N1 to node N E must be set to 4. If a message is received that is marked as a second wave.

[0101] The event is accumulated if the event is not known up to node N4. 5. If you receive a message that is marked as a breaking wave.

[0102] The communication state of the receiving channel is noN E is set to

[0103] An example of the communication flow is shown in FIG.

[0104] In this example, the Tidal Wave 8 is listed as Altitude's List W 1,b5 =[-3,-2,1,-10,2].

[0105] The first wave 6 is Altitude List W 4,e5 = [-4,-6,-1,-2,3,4] and also list of events 1: {e3, e4, e5, f3, f4}.

[0106] The second wave 7 contains the list of events 1: {a1, b1, b2, b4, b5, c2, c3, d3}.

Claims

1. A computer-implemented method for exchanging information among distributed computer nodes connected in a particular network, the method implementing a Directed Acyclic Graph (DAG) consensus algorithm, such as a hash graph consensus algorithm, through the use of gossip in a gossip protocol, the method comprising: First node (N 1 ) is randomly assigned to the second node (N 4 ) and transmitting reference information (8), The method comprises: Which node events are different from others (N 1 ) based on the reference information (8), If the reference information (8) is 1 ) event of the second given node (N 4 ) event, the first given node (N 1 ) and, as a result, the first grant node (N 1 ) is the second given node (N 4 ) from sending DAG-event information to the If the reference information (8) is 4 ) event of the first given node (N 1 ) event, the first given node (N 1 ), and the DAG-event information (6) is added to the second node (N 4 ), And preferably, then, the second grant node (N 4 ), and the DAG-event information (7) is added to the first node (N 1 ), and similarly at other nodes to synchronize and propagate the DAG-event information (6, 7) across the network.

2. 2. The computer-implemented method of claim 1, wherein the reference information relates to an order of the node events.

3. 3. The computer-implemented method of claim 1 or 2, wherein the order of the node events is determined by tracking the change in ordinate value of each event.

4. 4. The computer-implemented method of claim 1, wherein the order of node events is determined by having each node track an increasing integer value called altitude for each event generated at the node.

5. 5. A computer-implemented method according to any one of claims 1 to 4, characterized in that the reference information is transmitted via a tidal wave (8) and / or the cancellation information is transmitted via a breaking wave (9).

6. The present invention relates to a system for exchanging information among distributed computer nodes connected in a particular network, the system implementing a DAG (Directed Acyclic Graph) consensus algorithm, such as a hash graph consensus algorithm, through the use of gossip related to a gossip protocol, the system: First node (N 1 ) is randomly assigned to the second node (N 4 ) and transmit the reference information; The system preferably comprises: means for determining which node events precede others based on reference information (8); If the reference information (8) is 1 ) event of the second given node (N 4 ) event, the first given node (N 1 ) and, as a result, the first grant node (N 1 ) is the second given node (N 4 ) to prevent the sending of DAG-event information (6, 7); and If the reference information (8) is 4 ) event of the first given node (N 1 ) event, the first given node (N 1 ), and the DAG-event information (6) is added to the second node (N 4 ), and And preferably, then, the second grant node (N 4 ), and the DAG-event information (7) is added to the first node (N 1 and means for causing the The system is characterized in that the DAG-event information (6, 7) is synchronized and propagated through the network in the same way in other nodes.

7. 7. The system of claim 6, wherein the reference information relates to an order of the node events.

8. 8. A system according to claim 6 or 7, characterized in that it comprises means for tracking the change in the ordinate value of each event in order to determine the order of the node events.

9. 9. A system according to any one of claims 6 to 8, characterized in that, in order to determine the order of the node events, it comprises means for causing each node to track the change of an integer value called altitude, which increases for each event generated at the node.

10. 10. The system according to any one of claims 6 to 9, characterized in that it comprises means for transmitting the reference information via tidal waves (8) and means for transmitting the cancellation information via breaking waves (9).

Images