Gossip protocol optimization method and system based on dynamic adjustment

Abstract

The invention discloses a Gossip protocol optimization method and system based on dynamic adjustment, and belongs to the technical field of distributed systems and network communication. The method specifically comprises the following steps: in a distributed system, acquiring and initializing network parameters; the method comprises the following steps of: executing previous R-1 rounds of propagation on nodes for updating messages of a distributed system according to a Gossip protocol, and constructing an expected number of updated nodes at the end of an ith round; calculating the total number of messages sent in the first R-1 rounds; constructing a CCP model, and solving the CCP model through a binary search method to obtain the minimum number of forwarding times; dynamically adjusting the last round of forwarding times K '; and executing the Rth round of propagation, forwarding messages to the K'random nodes by all updated nodes, and realizing Gossip protocol optimization based on dynamic adjustment. According to the method, through a staged propagation strategy and CCP modeling, the number of forwarding times K'of the last round is dynamically adjusted, and redundant messages are remarkably reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of distributed systems and network communication, and more specifically, relates to an optimized message propagation method based on the Gossip protocol, which significantly reduces redundant messages and maintains the completely decentralized feature of the protocol by dynamically adjusting the message forwarding times in multiple rounds. Background Art

[0002] The Gossip protocol is a decentralized communication protocol, and its core process includes the following steps:

[0003] 1. Initialization: There are N nodes in the network. Initially, the information states of all nodes are the same. When a certain node generates updated data, at this time, it is necessary to synchronize the update message to other nodes, and the following steps are executed.

[0004] 2. Multi-round propagation: The protocol executes a predetermined R rounds of operations. In each round, the updated nodes randomly select K other nodes to forward the message.

[0005] 3. Exponential growth: Through multi-round propagation, the number of updated nodes grows exponentially, and finally the whole network is covered.

[0006] However, the traditional Gossip protocol has the following problems:

[0007] 1. Too many redundant messages: In the later stage of propagation, most nodes have been updated, but the nodes still forward messages according to a fixed number (K), resulting in a sharp increase in network load.

[0008] 2. Insufficient efficiency: The number of forwardings in the last round is not adjusted according to the actual coverage situation, which may cause waste of resources or insufficient coverage.

[0009] Existing improvement schemes are mostly based on simple linear adjustment or empirical formulas, without combining probability statistical models to quantify the redundancy boundary, and the optimization effect is limited. Summary of the Invention

[0010] To solve the deficiencies in the prior art, the present invention provides an optimized method and system for the Gossip protocol based on dynamic adjustment, specifically including a dynamic optimization method for the Gossip protocol based on CCP (Coupon Collector's Problem). Using a phased propagation strategy, in the first R - 1 rounds, it is executed according to the traditional Gossip protocol, and in the last round, the forwarding times are dynamically adjusted to K′, calculated based on K1 of CCP. The full-coverage problem of the Gossip protocol is transformed into a CCP model, and the minimum number of messages T is determined through a binary search algorithm, and the value of K′ is deduced inversely.

[0011] The present invention adopts the following technical solutions.

[0012] The first aspect of the present invention provides an optimization method for the Gossip protocol based on dynamic adjustment, specifically including:

[0013] In a distributed system, obtain and initialize network parameters, where the network parameters include the set total number of network nodes N, the total number of propagation rounds R, and the fixed forwarding times K in the first R - 1 rounds;

[0014] According to the initialized network parameters, the nodes with updated messages in the distributed system perform propagation in the first R - 1 rounds according to the Gossip protocol, and construct the expected number E of updated nodes at the end of the i-th round i ;

[0015] According to the constructed expected number E of updated nodes at the end of the i-th round, calculate the total number S of messages sent in the first R - 1 rounds;

[0016] Construct a CCP model based on the minimum number of forwarding times required to update to the total number of network nodes N, and solve the CCP model by the binary search method to obtain the minimum number of forwarding times T;

[0017] According to the calculated total number S of messages sent in the first R - 1 rounds and the obtained minimum number of forwarding times T, dynamically adjust the forwarding times K' in the last round;

[0018] Perform the R-th round of propagation, and forward the messages of all updated nodes to the K' random nodes, so as to realize the optimization of the Gossip protocol based on dynamic adjustment.

[0019] Preferably, the nodes with updated messages in the distributed system perform propagation in the first R - 1 rounds according to the Gossip protocol, specifically including:

[0020] Perform propagation in the first R - 1 rounds on the updated nodes. In each round of operation, all updated nodes randomly select K other nodes to forward messages, and the newly updated K nodes start to participate in forwarding messages from the next round.

[0021] Preferably, the constructed expected number E of updated nodes at the end of the i-th round i , is represented by the following formula:

[0022]

[0023] In the formula,

[0024] E i represents the expected number of updated nodes at the end of the i-th round.

[0025] Preferably, the calculation of the total number S of messages sent in the first R - 1 rounds is represented by the following formula:

[0026]

[0027] wherein,

[0028] E i represents the expected number of updated nodes at the end of the i-th round.

[0029] Preferably, the CCP model is solved by the binary search method to obtain the minimum number of forwarding times T, which is expressed by the following formula:

[0030] Initialize the binary search interval [Left, Right];

[0031] If Left > Right, directly output T = Right, otherwise, calculate the midpoint mid;

[0032] Determine whether all nodes can be covered with a probability of ≥ 99.99% when the number of messages is the midpoint mid;

[0033] If all nodes can be covered with a probability of ≥ 99.99%, then Right = mid, otherwise Left = mid + 1;

[0034] Repeat the above steps until Left ≥ Right, and output T = Right.

[0035] Preferably, initializing the binary search interval [Left, Right] specifically includes:

[0036] Left = E R-1 , Right = the total number of messages of the traditional Gossip protocol, where E R-1 represents the expected number of updated nodes at the end of the (R - 1)-th round, Right represents the right interval of the binary search interval, and Left represents the left interval of the binary search interval.

[0037] Preferably, the calculation of the midpoint mid is expressed by the following formula:

[0038]

[0039] wherein,

[0040] Right represents the right interval of the binary search interval;

[0041] Left represents the left interval of the binary search interval.

[0042] Preferably, the determination of whether all nodes can be covered with a probability of ≥ 99.99% when the number of messages is the midpoint mid is expressed by the following formula:

[0043] P(CCP|mid) ≥ 99.99% (4)

[0044] wherein,

[0045] P(CCP|mid) represents the probability of covering all nodes when the number of messages is the midpoint mid.

[0046] Preferably, the dynamic adjustment of the number of forwarding times K' in the last round is represented by the following formula:

[0047]

[0048] where

[0049] E R-1 is the expected number of updated nodes at the end of the (R - 1)-th round.

[0050] Preferably, the number of forwarding times K' in the last round ≤ the fixed number of forwarding times K in the previous (R - 1) rounds.

[0051] The second aspect of the present invention provides an optimized Gossip protocol system based on dynamic adjustment, which runs an optimized Gossip protocol method based on dynamic adjustment described in the first aspect, including:

[0052] Data initialization module: used to obtain and initialize network parameters in a distributed system, where the network parameters include the set total number of network nodes N, the total number of propagation rounds R, and the fixed number of forwarding times K in the previous (R - 1) rounds;

[0053] Expected solution module: used to perform the first (R - 1) rounds of propagation on the nodes with message updates in the distributed system according to the initialized network parameters according to the Gossip protocol, and construct the expected number of updated nodes E i ;

[0054] Message total number solution module: used to calculate the total number of messages S sent in the previous (R - 1) rounds according to the constructed expected number of updated nodes at the end of the i-th round;

[0055] CCP model construction and solution module: used to construct a CCP model according to the minimum number of forwarding times required to update to the total number of network nodes N, and solve the CCP model by the binary search method to obtain the minimum number of forwarding times T;

[0056] Forwarding times solution module: used to dynamically adjust the number of forwarding times K' in the last round according to the calculated total number of messages S sent in the previous (R - 1) rounds and the obtained minimum number of forwarding times T;

[0057] Last round update module: used to perform the R-th round of propagation, and forward the messages of all updated nodes to the K' random nodes to achieve the optimization of the Gossip protocol based on dynamic adjustment.

[0058] Compared with the prior art, the beneficial effects of the present invention at least include:

[0059] Through CCP modeling, the present invention uses a probabilistic statistical model to quantify the redundancy boundary of the node propagation times, and can dynamically adjust the forwarding times to make the forwarding times close to the theoretical lower limit of CCP, thereby reducing redundant messages. At the same time, a phased propagation strategy is used to dynamically adjust the last-round forwarding times K', significantly reducing redundant messages, improving network efficiency, and reducing resource waste. The first R-1 rounds are executed according to the traditional Gossip protocol, and the K' in the Rth round is calculated based on the statistical model. Experiments show that on the premise of ensuring that the network coverage probability reaches 99.99%, compared with the traditional Gossip protocol, the redundant messages of the present invention are reduced by 16.40% - 38.07%. The present invention is applicable to large-scale distributed scenarios such as blockchain networks and the Internet of Things. Description of the Drawings

[0060] Figure 1 is a schematic diagram of the optimization process of the Gossip protocol based on dynamic adjustment provided according to an embodiment of the present invention;

[0061] Figure 2 is a schematic diagram of the propagation process of dynamically adjusting K' provided according to an embodiment of the present invention;

[0062] Figure 3 is a schematic diagram of the T-value calculation flow chart (showing the binary search steps) based on the CCP model provided according to an embodiment of the present invention;

[0063] Figure 4 is a schematic diagram of the comparison of the number of redundant messages between the traditional Gossip protocol and the present invention provided according to an embodiment of the present invention. Detailed Embodiment

[0064] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0065] As Figure 1 shown, Embodiment 1 of the present invention provides an optimization method for the Gossip protocol based on dynamic adjustment, including the following steps:

[0066] Step 1: In a distributed system, obtain and initialize network parameters.

[0067] In a preferred but non-limiting embodiment of the present invention, in a distributed system, the total number of network nodes N, the total number of propagation rounds R, and the fixed forwarding times K in the first R-1 rounds are set.

[0068] Further preferably, in the blockchain network, standard values of the number of nodes N for block propagation, the predefined number of rounds R, and the fixed forwarding times K in the first R - 1 rounds are determined; in the Internet of Things device network, corresponding parameters N, R, and K are initialized according to the number of devices, the communication range, and the expected message coverage efficiency.

[0069] More preferably, the initialized parameters include, but are not limited to, the total number of network nodes N, the total number of propagation rounds R, and the fixed forwarding times K in the first R - 1 rounds. The initial values of these parameters can be determined according to the specific distributed system or communication network type and application requirements. For example, in a large-scale distributed storage system, N can correspond to the number of storage nodes, and the preset values of R and K are based on the expected file synchronization efficiency and redundancy tolerance.

[0070] Step 2: According to the network parameters initialized in Step 1, the nodes for message update in the distributed system perform the first R - 1 rounds of propagation according to the Gossip protocol, and the expected number of updated nodes E at the end of the i-th round is constructed. i 。

[0071] In a preferred but non-limiting embodiment of the present invention, the first R - 1 rounds of propagation are performed on the updated nodes. As Figure 2 shown, in each round of operation, all updated nodes randomly select K other nodes to forward the message, and the newly updated K nodes start to participate in forwarding the message from the next round.

[0072] Further preferably, the expected number of updated nodes E at the end of the i-th round i is calculated for the coverage expectation, and is expressed by the following formula:

[0073]

[0074] In the formula,

[0075] E i represents the expected number of updated nodes at the end of the i-th round,

[0076] N represents the total number of network nodes,

[0077] K represents the fixed forwarding times in the first R - 1 rounds.

[0078] Step 3: Based on the expected number of updated nodes E at the end of the i-th round constructed in Step 2 i , calculate the total number of messages S sent in the first R - 1 rounds.

[0079] In a preferred but non-limiting embodiment of the present invention, the total number of messages S sent in the first R - 1 rounds is calculated, and is expressed by the following formula:

[0080]

[0081] In the formula,

[0082] K represents the fixed forwarding times in the first R - 1 rounds.

[0083] Step 4: Construct a CCP model based on the minimum forwarding times T required to update to the total number of network nodes N, and solve the minimum forwarding times T through the binary search method.

[0084] In a preferred but non - restrictive embodiment of the present invention, a CCP model is constructed based on the minimum forwarding times T required to update to the total number of network nodes N, and the minimum forwarding times T is solved through the binary search method. The specific steps include:

[0085] Step 4.1: Convert the problem of the minimum forwarding times T required to update to the total number of network nodes N in the Gossip protocol into a CCP (Coupon Collector's Problem), and calculate the minimum number of attempts T required to collect N kinds of "coupons".

[0086] It should be noted that for the traditional Gossip protocol in the later stage of propagation, most nodes have been updated, but the nodes still forward messages according to the fixed number of times (K), resulting in a sharp increase in network load due to excessive redundant messages. The present invention converts the problem of the minimum forwarding times T required to update the traditional Gossip protocol to the total number of network nodes N into a probability - statistical model of CCP, calculates the minimum number of attempts T required to collect N kinds of "coupons", and quantifies the network propagation problem after conversion to CCP. The forwarding times can be dynamically adjusted to make the forwarding times close to the theoretical lower limit of CCP, thereby reducing redundant messages and improving network efficiency.

[0087] Step 4.2: Rapidly approximate the value of T through the binary search algorithm.

[0088] As Figure 3 shown, the binary search algorithm includes the following steps:

[0089] Step 4.2.1: Initialize the binary search interval [Left, Right], Left = E R-1 , Right = the total number of messages of the traditional Gossip protocol, where E R-1 represents the expected number of updated nodes at the end of the (R - 1) - th round, Right represents the right interval of the binary search interval, and Left represents the left interval of the binary search interval.

[0090] Step 4.2.2: If Left > Right, directly output T = Right; otherwise, calculate the mid - point mid, which is expressed by the following formula:

[0091]

[0092] Step 4.2.3: Determine whether all nodes can be covered with a probability of ≥ 99.99% when the number of messages is the midpoint mid calculated in Step 4.2.2, which is expressed by the following formula:

[0093] P(CCP|mid) ≥ 99.99% (4)

[0094] In the formula,

[0095] P(CCP|mid) represents the probability of covering all nodes when the number of messages is the midpoint mid.

[0096] If the condition of formula (4) is satisfied, then Right = mid; otherwise, Left = mid + 1;

[0097] Step 4.2.4: Repeat Step 4.1.2 - Step 4.1.3 until Left ≥ Right, and output T = Right.

[0098] Step 5: Dynamically adjust the number of transmissions K' in the last round based on the minimum number of transmissions T calculated in Step 4 and the total number of messages S sent in the first R - 1 rounds calculated in Step 3.

[0099] In a preferred but non - limiting embodiment of the present invention, the number of transmissions K' in the last round is calculated by the following formula:

[0100]

[0101] Wherein,

[0102] E R-1 is the expected number of updated nodes at the end of the (R - 1) - th round.

[0103] It should be noted that the present invention uses phased propagation. In the first R - 1 rounds, it is executed according to the traditional Gossip protocol, and in the R - th round, K' is calculated based on T solved by the probability - statistical model CCP, reducing redundant messages and resource waste.

[0104] Step 6: Execute the R - th round of propagation, and forward the messages of all updated nodes to K' randomly selected nodes obtained in Step 5, realizing the optimization of the Gossip protocol based on dynamic adjustment.

[0105] In a preferred but non - limiting embodiment of the present invention, the number of transmissions K' in the last round ≤ the fixed number of transmissions K in the first R - 1 rounds.

[0106] Compared with the prior art, the beneficial effects of the present invention at least include:

[0107] Through CCP modeling, the present invention uses a probability and statistics model to quantify the redundancy boundary of the number of node propagations, and can dynamically adjust the number of forwarding times, making the number of forwarding times close to the theoretical lower limit of CCP, thereby reducing redundant messages. At the same time, a phased propagation strategy is used to dynamically adjust the number of forwarding times K' in the last round, significantly reducing redundant messages, improving network efficiency, and reducing resource waste. The first R - 1 rounds are executed according to the traditional Gossip protocol, and the Rth round calculates K' based on the statistical model. Experiments show that on the premise of ensuring that the network coverage probability reaches 99.99%, compared with the traditional Gossip protocol, the redundant messages of the present invention are reduced by 16.40% - 38.07%. The present invention is applicable to large-scale distributed scenarios such as blockchain networks and the Internet of Things.

[0108] Embodiment 2 of the present invention provides an optimized system for the Gossip protocol based on dynamic adjustment, which runs the optimized method for the Gossip protocol based on dynamic adjustment described in Embodiment 1, including:

[0109] Data initialization module: used to obtain and initialize network parameters in a distributed system, where the network parameters include the set total number of network nodes N, the total number of propagation rounds R, and the fixed number of forwarding times K in the first R - 1 rounds;

[0110] Expected solution module: used to perform the first R - 1 rounds of propagation on the nodes that update the messages in the distributed system according to the initialized network parameters, and construct the expected number E of updated nodes at the end of the ith round i ;

[0111] Total message number solution module: used to calculate the total number of messages S sent in the first R - 1 rounds according to the constructed expected number E of updated nodes at the end of the ith round;

[0112] CCP model construction and solution module: used to construct a CCP model for the purpose of solving the minimum number of forwarding times required to update to the total number of network nodes N in the Gossip protocol, and solve the CCP model by the binary search method to obtain the minimum number of forwarding times T;

[0113] Forwarding number solution module: used to dynamically adjust the number of forwarding times K' in the last round according to the calculated total number of messages S sent in the first R - 1 rounds and the obtained minimum number of forwarding times T;

[0114] Last round update module: used to perform the Rth round of propagation, and forward the messages of all updated nodes to the K' random nodes to achieve the optimization of the Gossip protocol based on dynamic adjustment.

[0115] Such as Figure 4As shown in the figure, Embodiment 3 of the present invention provides a comparison of the number of redundant messages of the present invention with the traditional Gossip protocol in the distributed system network scenario. In a distributed system network, information synchronization between nodes is the key to ensuring the consistency and efficient operation of the system. The present invention significantly reduces the number of redundant messages and improves the network transmission efficiency by dynamically adjusting the forwarding times of the Gossip protocol.

[0116] In a distributed system network containing 1000 nodes, after a node is updated, when the total number of message propagation rounds is set to 5 rounds and the fixed forwarding times in the first 4 rounds are 10 times, the total number of messages of the traditional Gossip protocol is 18,286, while the total number of messages of the optimized protocol of the present invention is reduced to 11,321, and the redundant messages are reduced by 38.07%.

[0117] Taking N = 1000, K = 10, and R = 5 as an example:

[0118] Propagation in the first 4 rounds:

[0119] The number of forwarding times per round K = 10, and the total number of messages in the first 4 rounds is counted as S = 8336.

[0120] The expected number of updated nodes E at the end of the 4th round 6 = 995.

[0121] Calculate the T value:

[0122] Through the CCP model and binary search, it is determined that T = 10392 (ensuring 99.99% probability of full coverage).

[0123] Calculate K':

[0124] Δ = 10392 - 8336 = 2056.

[0125]

[0126] Propagation in the 5th round:

[0127] The forwarding times are adjusted to K1 = 3, and the total number of messages is reduced to 11321.

[0128] If the K value is not changed, the total number of messages is 18286. The redundancy is reduced by 38.07%.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the specific implementation manners of the present invention can still be modified or equivalently replaced, and any modification or equivalent replacement without departing from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A Gossip protocol optimization method based on dynamic adjustment, characterized in that: In a distributed system, obtain and initialize network parameters, which include setting the total number of network nodes N, the total number of propagation rounds R, and the fixed forwarding number K in the first R-1 rounds; According to the initialized network parameters, the nodes that update the message of the distributed system perform the first R-1 rounds of propagation according to the Gossip protocol, and construct the expected number of updated nodes E at the end of the i-th round. i ; Calculate the total number of messages S sent in the previous R-1 rounds based on the expected number of updated nodes at the end of the i-th round constructed as described above; The CCP model is constructed based on the minimum forwarding times required to update the total number of network nodes N. The CCP model is solved by binary search to obtain the minimum forwarding times T. Dynamically adjust the number of forwarding times K′ in the last round according to the total number of messages S sent in the first R-1 rounds and the obtained minimum number of forwarding times T; Perform the Rth round of propagation, forward all updated nodes to the K′ random nodes, and implement Gossip protocol optimization based on dynamic adjustment.

2. According to a Gossip protocol optimization method based on dynamic adjustment according to claim 1, it is characterized in that: According to the initialized network parameters, the nodes that update the message of the distributed system perform the first R-1 rounds of propagation according to the Gossip protocol, which specifically includes: The first R-1 rounds of propagation are performed on the updated nodes. In each round of operation, all updated nodes randomly select K other nodes to forward the message, and the newly updated K nodes participate in forwarding messages starting from the next round.

3. According to a Gossip protocol optimization method based on dynamic adjustment according to claim 1, it is characterized in that: The expected number of updated nodes E at the end of the i-th round of construction i , expressed as follows: In the formula, E i Represents the expected number of updated nodes at the end of round i.

4. According to a Gossip protocol optimization method based on dynamic adjustment according to claim 1, it is characterized in that: The total number of messages S sent in the first R-1 rounds of calculation is expressed as follows: In the formula, E i Represents the expected number of updated nodes at the end of round i.

5. According to a Gossip protocol optimization method based on dynamic adjustment according to claim 1, it is characterized in that: The CCP model is solved by binary search to obtain the minimum forwarding times T, which is expressed as the following formula: Initialize the binary search interval [Left,Right]; If Left>Right, directly output T=Right, otherwise, calculate the midpoint mid; Determine whether all nodes can be covered with a probability of ≥ 99.99% when the number of messages is the midpoint mid; If all nodes can be covered with a probability of ≥99.99%, then Right = mid, otherwise Left = mid + 1; Repeat the above steps until Left ≥ Right, and output T = Right.

6. According to a Gossip protocol optimization method based on dynamic adjustment according to claim 5, it is characterized in that: Initialize the binary search interval [Left,Right], including: Left=E R-1 ,Right=total number of messages in traditional Gossip protocol, where E R-1 It represents the expected number of updated nodes at the end of the R-1th round, Right represents the right interval of the binary search interval, and Left represents the left interval of the binary search interval.

7. The Gossip protocol optimization method based on dynamic adjustment according to claim 5 is characterized in that: The calculation midpoint mid is expressed by the following formula: In the formula, Right represents the right interval of the binary search interval; Left represents the left interval of the binary search interval.

8. The Gossip protocol optimization method based on dynamic adjustment according to claim 5, characterized in that: The judgment of whether all nodes can be covered with a probability of ≥ 99.99% when the number of messages is the midpoint mid is expressed by the following formula: P(CCP|mid)≥99.99% (4) In the formula, P(CCP|mid) represents the probability of covering all nodes when the number of messages is the midpoint mid.

9. The Gossip protocol optimization method based on dynamic adjustment according to claim 1, characterized in that: The dynamic adjustment of the last round of forwarding times K′ is expressed by the following formula: in, E R-1 is the expected number of updated nodes at the end of round R-1.

10. A Gossip protocol optimization system based on dynamic adjustment, running a Gossip protocol optimization method based on dynamic adjustment according to any one of claims 1 to 9, characterized in that: Data initialization module: used to obtain and initialize network parameters in a distributed system. The network parameters include setting the total number of network nodes N, the total number of propagation rounds R, and the fixed forwarding number K in the first R-1 rounds; Expected solution module: used to perform the first R-1 rounds of propagation for the nodes of the distributed system message update according to the Gossip protocol based on the initialized network parameters, and construct the expected number of updated nodes E at the end of the i-th round i ; A message total number solving module is used to calculate the total number of messages S sent in the previous R-1 rounds according to the expected number of updated nodes at the end of the i-th round constructed; CCP model construction and solution module: used to build a CCP model based on the minimum forwarding times required to update the total number of network nodes N, and solve the CCP model through binary search method to obtain the minimum forwarding times T; Forwarding times solving module: used for dynamically adjusting the last round forwarding times K′ according to the total number of messages S sent in the first R-1 rounds and the obtained minimum forwarding times T; The last round of update module: used to perform the Rth round of propagation, forwarding messages from all updated nodes to the K′ random nodes, and realizing Gossip protocol optimization based on dynamic adjustment.

Images