CN115333673A

CN115333673A - Block coding transmission method based on ILT in block chain network

Info

Publication number: CN115333673A
Application number: CN202210905529.2A
Authority: CN
Inventors: 张佩云; 郭威峰; 陶言昊; 束俊良; 何思开; 黄天林; 潘朝君; 谢荣见
Original assignee: Nanjing University of Information Science and Technology
Current assignee: Nanjing University of Information Science and Technology
Priority date: 2022-07-29
Filing date: 2022-07-29
Publication date: 2022-11-11

Abstract

The invention discloses an ILT-based block coding transmission method in a block chain network, which comprises the steps of firstly predicting the channel state of a neighbor node, then determining how to divide and group blocks according to the channel state and transmitting grouped codes; meanwhile, an improved coded packet transmission protocol based on a Pichu block transmission protocol is provided, in the protocol, a node sends INV information to a neighbor node as long as receiving a coded packet, and simultaneously verifies a block. The invention can transmit the block by utilizing the channel to the maximum extent, thereby avoiding the block retransmission phenomenon caused by overlarge block and poor state of the transmission channel; even if the transmitted coded packet is lost, no retransmission is required.

Description

Block coding transmission method based on ILT in block chain network

Technical Field

The present invention relates to a block coding transmission method, and more particularly, to an ILT-based block coding transmission method in a block chain network.

Background

The impact of communication cost on block transmission performance is mainly expressed in two aspects of block size and communication bandwidth. Larger blocks indicate that more transactions can be carried in the bitcoin network, which is beneficial to improving the scalability of the blockchain, but also causes problems, i.e., more time is required to propagate the blocks, which is not beneficial to block synchronization. In contrast, small blocks increase the block synchronization time, but do not facilitate the extension of the block chain. The communication bandwidth is the communication capacity of the node itself, and the higher the communication bandwidth, the less time the node spends in transmitting the block. Conversely, the more time a node spends.

As shown in fig. 1, in the conventional block transmission protocol, the protocol method has the time cost that nodes are continuously synchronized due to inconsistent transactions, which consumes unnecessary bandwidth, and simultaneously, the entire block needs to be transmitted during transmission, which results in low transmission efficiency of a single block and increased retransmission probability. Within the current protocol, several important information needs to be noted: 1) Verifying the computing power requirement of the block and verifying the transaction by the node; 2) Whether a transmission object of a node is on line or not and whether the state of a block can be accepted or not are required to be confirmed; 3) Only the first two passes can complete the block transfer. Under the condition that the 3-point important principle is not violated, if the node A needs the same time after receiving the block information due to the block verification, the node B can be informed of sending the current node state and the current channel communication information by selecting to preferentially send the inv message in the time period; after receiving the information, the node A determines the sending mode according to the received channel information and the state information of the node B. The sending mode is mainly based on the premise that the block body information and the transaction information in the block are not changed, and aims to transmit the block to a destination node quickly, safely and completely under different state conditions. However, in the block chain network, when a node transmits a block, the block is retransmitted due to the fact that the block is too large and the state of a transmission channel is not good, and the final confirmation of the block is affected due to the instability of the transmission block, so that the safety of the network is reduced.

Disclosure of Invention

The purpose of the invention is as follows: the invention aims to provide an ILT-based block coding transmission method in a block chain network, which reduces the size of a single block, reduces the probability of retransmission and improves the transmission efficiency.

The technical scheme is as follows: the block coding transmission method of the invention comprises the following steps:

s10, when the node receives the block, predicting the state of the transmission channel and dividing the state of the transmission channel;

s20, coding the block by adopting a block coding algorithm;

s30, after the node finishes block coding, an improved Pichu block transmission protocol is used for sending a coding packet to a neighbor node;

and S40, when the node receives enough encoding packets, decoding by adopting a block decoding algorithm.

Further, in step S10, the transmission status of the channel is predicted by calculating an average transmission bandwidth of the channel, where the calculation formula of the average transmission bandwidth is as follows:

wherein P is the average transmission bandwidth; q. q of _max Representing the maximum amount of data transmitted during transmission, q _max ＝max{q ₁ ,q ₂ ,q ₃ ,…,q _i },1≤i≤N；q _min Represents a minimum amount of transmission data; 1200 denotes two unit times 1200 seconds;

the estimated time for a block to pass through the channel is as follows:

wherein D is the distance between nodes, and S/N is the signal-to-noise ratio;

let the node channel threshold be T _r When T is _b Less than T _r When, it means that the channel is in idle state; otherwise, the channel is in a congested state.

Further, in step S20, encoding the block is a process of linearly combining a plurality of packets to generate a new encoded packet, where each encoded packet is independent, and the specific steps are as follows:

s211, arranging all transaction groups in sequence;

s212, generating a degree d according to the degree distribution function;

s213, randomly selecting non-repeated transaction groups as adjacent elements of the coding groups, and recording the original transaction group positions;

s214, carrying out XOR on the randomly selected transaction groups to form a coding packet;

s215, generating a coding packet set, wherein the set comprises the degree and the adjacent meta-information of the coding packet, and sending the set information to a receiving party.

Further, a degree d is generated through a degree distribution function, then d random numbers are generated, and d groups are selected according to the random numbers, so that the size of the random numbers is larger than 0 and smaller than the group number w:

wherein M is _t Representing a total number of transactions in the block; h represents the Mi-Merkle Tree height formed by the exchanges that make up the group; 2 ^h-1 Representing the number of transactions divided into a group;

carrying out exclusive or on the selected groups and recording the original positions of the groups to form a coding packet;

the degree means that one coding packet needs several groups for coding; the degree distribution function is then as follows:

wherein s is the number of packets with the degree of 1, W represents the total number of packets, and delta represents the decoding failure probability; e represents the total number of the encoding packets, and E > W > s;

when the degree d =1, the degree distribution function is the proportion of the number of packets with the degree 1 to the total number of packets; when the degree d is other values, a 1-delta probability recovery function is added on the basis of the ideal soliton distribution.

Further, in step S30, the block transmission protocol uses an improved PiChu block transmission protocol to transmit the encoded packet, the verification of the block is to verify the block header, and the verification of the block is to identify the block information.

Further, in step S40, the specific steps of decoding by the block decoding algorithm are as follows:

s231, the receiving end receives a certain amount of coded packets, obtains the degree and the grouping in the coded packets according to the coding, and forms a Tanner graph according to the relation between the coded packets and the grouping;

s232, finding out the code packet symbol with the degree d of 1, recovering the group S1 corresponding to the code packet symbol, deleting the code packet with the degree of 1, carrying out XOR on the group S1 and other code packets connected with the group S1, and then storing the group S1 in a block, so that the number of the code packets connected with the group S1 can be reduced by 1;

s233, repeating the step S232 until all the groups are recovered and the decoding is successful;

and S234, constructing the intra-group transaction according to the Mi-Merkle Tree sequence to obtain a complete block.

Further, the Mi-Merkle Tree is subjected to Hash grouping according to the original Merkle Tree in the block and the transaction with a certain height;

when the number of transactions is M _t Value, the height of the Merkel tree is

And when the height of the tree is more than or equal to 8, dividing and grouping the transactions in the block.

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

1. the invention reasonably groups the transactions in the transmission blocks according to the channel state, and provides an ILT algorithm to encode the grouped information, the size of a single block is reduced through the encoded information, and the retransmission probability is reduced;

2. the invention divides the block into groups and codes the groups into coded packets, thereby changing the complete block transmission into the propagation of small blocks; by adopting the method, retransmission is not needed even if errors occur in the transmission process, and the coding mode has certain fault tolerance, so that the block transmission efficiency is improved.

Drawings

FIG. 1 is a diagram of a conventional block transport protocol;

FIG. 2 is a general flow diagram of the present invention;

FIG. 3 is a block transmission process diagram according to the present invention;

FIG. 4 is a diagram of new block header information;

FIG. 5 is a schematic view of transaction grouping information;

FIG. 6 is a schematic diagram of the original MerkelTree structure;

FIG. 7 is a schematic diagram of the structure of Mi-MerkelTree;

FIG. 8 is a schematic diagram of an encoding process;

FIG. 9 is an encoding flow diagram;

FIG. 10 is a diagram illustrating the format of an encoded packet;

fig. 11 (a) is a schematic diagram of the determination packet S1,

figure 11 (b) is a schematic diagram of the xor of the S1 value with its concatenated packet,

fig. 11 (c) is a schematic diagram of the encoded packet and the connectivity corresponding to S1 after being deleted,

fig. 11 (d) is a diagram showing S2 determined by the coded packet having the transition 1,

FIG. 11 (e) is a diagram illustrating the process of XOR-connecting S2 with other encoded packets and deleting the connectivity,

fig. 11 (f) is a packet S3 determined by the passage 1;

FIG. 12 is a diagram of a new transport protocol;

FIG. 13 is a diagram of a block transmission algorithm based on channel conditions;

FIG. 14 is a block synchronization timing diagram for different node numbers;

FIG. 15 is a graph showing block synchronization times for different transaction amounts;

fig. 16 is a diagram illustrating a block transmission success rate for different node numbers;

FIG. 17 is a diagram illustrating block transmission success rates for different transaction amounts;

FIG. 18 is a diagram of the retransmission rate of coded packets at different nodes;

fig. 19 is a diagram illustrating the retransmission rate of coded packets for different transaction amounts.

Detailed Description

The invention is described in further detail below with reference to the drawings and the detailed description.

As shown in fig. 2, the present invention reasonably groups transactions in a transmission block according to a channel state, and proposes an ILT algorithm (Improved liby Transform) to encode packet information, so as to reduce the size of a single block and reduce the probability of retransmission by using encoded information. Meanwhile, the transmission efficiency is accelerated on the basis of the block transmission stability through a new transmission protocol.

The invention considers the following factors for the transmission protocol: 1) Whether a node has received a block; 2) The number of transactions per transmission is reduced, thereby reducing the block size for a single transmission.

Three elements of the block transmission process that constructs a blockchain network are as follows: blocks, packets, and coded packets. The specific definition is as follows:

block: a Block is defined by Block, different transaction information is stored in the Block, and the transaction information forms a Merkel Tree in a full binary Tree structure.

Grouping: and defining grouping by using packet, wherein the grouping is to divide the transaction information in the Merkle Tree in the block into a group according to the height h, and the structure formed by the transaction information in each group is defined as Mi-Merkel Tree.

And (3) encoding the packet: and defining an Encoding packet by using an Encoding packet, wherein the Encoding packet is obtained by Encoding the grouping information, and the obtained Encoding packet is transmitted to other nodes which do not receive the blocks.

Blocks are important elements of block transmission; and grouping is to divide the blocks in order to reduce the block size for a single transmission; the encoded packet is the unit that is ultimately transmitted between the nodes.

Channel state prediction

Channel prediction is a process of measuring a transmission channel in advance when a node needs to transmit a block. For this reason, the following assumptions are made for the channel: a1 Node has sufficient download bandwidth, upload bandwidth is limited; a2 Node download bandwidth is known and each node connects to multiple other nodes; a3 Does not limit the topology of the blockchain network, the dilution of the node distribution. Predicting the transmission state of the channel by calculating the average transmission bandwidth of the channel, as shown in formula (1):

in formula (1), P is the average transmission bandwidth; q. q of _max Representing the maximum amount of data transmitted during transmission, q _max ＝max{q ₁ ,q ₂ ,q ₃ ,…,q _i },q _i (1 ≦ i ≦ n), n representing the number of transmissions; q. q.s _min Representing the minimum amount of data to be transmitted. 1200 denotes two unit times of 1200 seconds, and in order to more accurately judge whether the current channel is in a congestion state when the block is transmitted, the block passing channel time needs to be predicted.

The estimated time for the current block to pass through the channel is calculated by equation (2):

in the formula (2), D is the distance between nodes, and S/N is the signal-to-noise ratio. The approximate time of the block passing through the channel can be obtained by equation (2).

For this purpose, T is introduced _r A value, which is a node channel threshold, by which the congestion status of the transmission of the channel is reflected: when T is _b Less than T _r When, it means that the channel is in idle state; trans formThe channel is in a congested state.

Block transmission process based on ILT algorithm

In a blockchain network, a block can be viewed as one contiguous segment of data. And the block transmission among the nodes is to divide the block into groups by adopting ILT coding, thereby reducing the transmission pressure of a single node and improving the transmission efficiency and stability of the block. The block transfer process based on ILT encoding is a process from encoding, transmission and decoding. The specific process is shown in fig. 3, assuming that node a has verified that the block has passed, it is necessary to transmit the block to other neighboring nodes b, c, d. The transmission channel status of nodes b and d is better than that of node c, so that node b and node d can receive the block information faster than node c. At this time, the node a transmits the encoded packet to the nodes b, c and d by means of ILT encoding. When the node b receives enough packet information, the complete block information is obtained by means of decoding, and the block information is verified. When the node c may not receive the packet in time due to transmission limitation, the node b transmits the received coded packet to the node c.

The blocks need to be modified before block coding in order to adapt to the transmission state of the channel and reduce the additional communication cost in transmission. Therefore, a new block structure is adopted to divide the complete block into block header information and grouping information. The block head information is used for storing block information and transaction number information, and the grouping information stores specific transaction information. As shown in fig. 4, the block header information is added with a Number of 2 bytes in the original block header to indicate the transaction Number. The reason is that the block header and the grouping information are transmitted separately, and the receiver can obtain the specific transaction quantity information after receiving the block header information and can confirm whether all transaction information is received in the verification process.

Since thousands of transaction information are stored in one block, transactions need to be grouped, the grouping information is shown in fig. 5, a symbol part represents a hash value (32 bytes) of the Mi-merckle Tree, and the rest stores the transaction information.

The Mi-Merkle Tree carries out Hash grouping according to the original Merkle Tree in the block and the transaction with the height of h, and the obtained grouping does not disturb the sequence of original transaction information and is also beneficial to reconstructing the block and verifying by the node. The original Merkle Tree structure is shown in fig. 6, where Tx1 represents transaction 1, ha1 represents the hash value of the transaction, ha1,2 represent the values obtained by hashing Ha1 and Ha2, and so on. Therefore, 16 transactions are stored in the block, and the root hash value of the Merkel Tree is finally generated through two-two hashing.

To reduce the size of the single-pass transport block, the hashing process needs to be improved. Assuming that 16 transactions in fig. 6 are divided into groups, the Merkel Tree has 5 layers in total, when h =3 is selected, the transaction number in the group is 4, the group number is 4 groups, and the structure of the sub-Merkel Tree formed by the transactions in the 4 groups is the proposed Mi-Merkel Tree, specifically, as shown in fig. 7, the structure of the Mi-Merkel Tree is a generation process of selecting an original Merkel sub-Tree, thereby reducing the block size of single transmission. Therefore, when the number of transactions is M _t When the value is, the height of Merkel tree is

When the height of the tree is 8 or more, it means that the block is too large, and it is necessary to divide and group the intra-block transaction.

The encoding process, the decoding process and the transmission process of the ILT algorithm are as follows:

(21) Encoding process

The block coding process is mainly a process of linearly combining a plurality of groups to generate a new coded packet. When the node a in fig. 3 transmits the block information to the node b, all the packets need to be arranged in sequence, then the degree d is determined according to the degree distribution function, and the packets are randomly selected according to the degree d. And finally, carrying out exclusive OR on the selected groups to generate a new coding packet. ILT coding requires that when d packets are selected for coding, d is probabilistically satisfied with the degree distribution function. As shown in fig. 8, when the degree of selection is 3, 3 transaction packets are selected from the transaction group for encoding. Therefore, each encoded packet of the ILT encoding is independent of each other, and the encoding flowchart is shown in fig. 9, where a specific encoding step is as follows:

211 Group all transactions in order;

212 Determining a degree d of the encoded packet according to the optimized degree distribution function;

213 Randomly selecting non-repeated transaction groups as adjacent elements of the coding groups, and recording the original transaction group positions;

214 Xor randomly selected transaction groups to form a coding packet;

215 Generates a set of encoded packets containing degree and adjacency meta-information of the encoded packets, and transmits the set information to the receiving side.

In fig. 9, a degree d is generated by the degree distribution function, then a random number is generated, and d packets are selected according to the random number, so that the size of the random number should be larger than 0 and smaller than the packet number w.

Where h represents the Mi-Merkle Tree height formed by the exchanges that make up the group.

And carrying out exclusive OR on the selected groups and recording the original positions of the selected groups until all the group codes are finished. The format of the generated coded packet is shown in fig. 10, and Degree (Degree) is used to indicate the number of packets constituting the coded packet, and for example, the Degree is 3 if the coded packet has 3 different packets. The Original Packet location (Original Packet location) is the order in which the packets were recorded as Original, and the last Packet data (Packet data) is used to hold the packets.

(22) Optimizing degree distribution function

The degree distribution function is the core of the ILT algorithm, and the selection of the degree directly influences the coding superiority and inferiority of the coding. The degree distribution function is a probability distribution function, and the essence is to select the appropriate degree with probability. The degree of the invention mainly refers to that one coding packet needs several groups for coding. Therefore, two points should be considered in the design degree score function:

b1 One degree 1 packet occurs in each iteration;

b2 To ensure that the degree of encoding is linear with time complexity.

The optimized degree distribution function is then as follows:

in the formula (4), s is the number of packets with the degree of 1, W represents the total number of packets, and delta represents the decoding failure probability; e represents the total number of encoded packets (E > W > s).

Equation (4) is explained by pitching into the barrel: assuming s buckets, at least s ln (s/delta) balls are needed to ensure that the balls are thrown into the buckets with a probability of 1-delta, and the number of coding packets needed for the number of groups is at least equal to that of the remaining W-s groups to ensure that the remaining W-s groups can be recovered with a probability of 1-delta

When the degree d is 1, the degree distribution function is the proportion of the number of packets with the degree 1 to the total number of packets. When the degree is other values, a 1-delta probability recovery function is added on the basis of ideal soliton distribution, and the decoding efficiency is increased.

(23) Decoding process

ILT code decoding requires two pieces of information: grouping and degree. The packet sum degree is included in the encoded packet, and the packet sum degree is acquired by the encoded packet. The decoding algorithm is a belief propagation algorithm, and comprises the following specific steps:

231 Receiving end receives a certain amount of coded packets, obtains degree and grouping in the coded packets according to coding, and forms a Tanner graph by the relation between the coded packets and the grouping, as shown in fig. 11 (a) - (f);

232 Finding the code packet symbol with the degree d of 1, recovering the corresponding grouping S1, deleting the code packet with the degree of 1, carrying out XOR on the grouping S1 and other code packets connected with the grouping S1, and then storing the grouping S1 in a block, so that the code packet degree connected with the grouping S1 can be reduced by 1;

233 Repeat step 232 until all packets can be recovered and decoding is successful;

234 The packets are constructed in order, resulting in complete blocks.

In fig. 11, (a) - (f), circles indicate packets, and squares indicate encoded packets. Fig. 11 (a) determines the packet S1 by encoding the packet number of degrees to 1. Fig. 11 (b) xors the S1 value with its concatenated packet and deletes the encoded packet and the connectivity corresponding to S1, which results in fig. 11 (c). Then, the next coded packet with the degree of 1 is found in fig. 11 (c), S2 is determined by the coded packet with the degree of 1 in fig. 11 (c), and the coded packet with the degree of 1 connected to S2 is deleted. Next, the S2 in (d) in fig. 11 is xored with other coded packets, and the connectivity is deleted to obtain (e) in fig. 11. Finally, by determining the grouping S3 for the number 1, (f) in fig. 11 is obtained.

(24) Block transmission protocol

Block transport protocols based on the ILT algorithm are proposed in order to reduce bandwidth consumption. The new transport protocol is to transport encoded packets according to the modified Pichu Block transport protocol, as shown in FIG. 12. The verification of the block is mainly to verify the block head, and the verification of the block is concentrated into identifying the block information rather than verifying the transaction in the block. This protocol takes into account the following factors:

c1 Security and efficiency issues for the peer nodes need to be considered;

c2 Need to note whether the node trades for consistency when reconstructing the block;

c3 Needs to account for transport block interruptions and retransmissions due to churn (churn) by the nodes.

In fig. 12, after improvement according to the LT algorithm, once node a receives a block, it needs to send an Inv (blockhead) message to node B at the same time for block verification, and if it receives Get information sent by node B at this time, it needs to group the block after the block information is verified, and it performs ILT encoding after grouping into k encoded packets (Sym _ chunk) for sending. Node B only needs to receive

When a packet is encoded, the block can be reconstructed. Therefore, the transmission protocol of the present invention has the following advantages:

d1 Ensuring the integrity and safety of block transmission, wherein the integrity is realized when the ILT algorithm does not damage the block, and simultaneously ensuring that the nodes can receive consistent blocks, and the safety is realized when verifying the information of the block and verifying the transaction information in the block;

d2 Ensure sufficient "knowledge" between nodes, meaning that the nodes know explicitly the neighbor node channel information before selection;

d3 Reduced consumption of transmission bandwidth, and the number of packets is adjusted according to channel conditions, reducing the consumption of transmission bandwidth.

(III) Block Transmission Process

The block transmission process is displayed through an ILT encoding algorithm, an ILT decoding algorithm, and a block transmission protocol, and as shown in fig. 13 in detail, two algorithms and one protocol are called in total: an encoding algorithm (algorithm 1), a decoding algorithm (algorithm 2) and a transmission protocol (protocol 1), when a node receives a block, the channel state of a neighbor node is predicted, and the channel state is divided. The block is then encoded (algorithm 1), and when the node completes the block encoding, the modified Pichu Block transfer protocol (protocol 1) is used to transfer the encoded packet to the neighboring node. When the coded packet is received, decoding is performed, and a decoding algorithm (algorithm 2) is called. The algorithms are as follows:

(31) Block coding algorithm

The block coding algorithm of the present invention is used for block coding of blocks. Specifically, as shown in algorithm 1:

algorithm 1

In algorithm 1, when a node i receives a block, it needs to measure the channel, and determines the current transmission channel state by comparing the calculated average transmission time with the node channel threshold (steps A1 to A3). Determining the number of packets according to the channel state, then determining the included packets through a degree distribution function, calculating the degree distribution function (steps A5 to A8), and carrying out XOR on the randomly selected packets to obtain a coding packet (steps A9 to A11). The encoded packet set S is transmitted (step a 12). If the block itself is small, no coded transmission is required (steps A13 to A15).

(32) Block transmission protocol

The block transmission protocol of the invention transmits the coded packet to other nodes. The specific information is shown in protocol 1:

protocol 1

A detailed description is given in connection with fig. 12 and the block transport protocol as follows:

firstly, sym _ chunk in the protocol is expressed as an ILT-coded packet, and the node A receives the block information and propagates the INV message to the neighbor nodes after verifying the block header information.

Secondly, if node B receives INV (chunk head) information and node B does not receive the chunk, it returns a Get message, which is for the following purposes: 1) The node B needs to be prompted to have no block information and is in a state of being capable of receiving the block; 2) Testing the transmission distance between the AB nodes; 3) The transmission channel state information of the node B is required.

Then, the node B forwards the verified INV (blockhead) message to the node C, and the node C also sends a GET message to the node B if there is no block.

In order to prevent node a from transmitting without the verification block, node a does not immediately transmit the block when it receives the Get message, and must wait until the verification block message is finished before transmitting the block. If the node A verifies that the block information is not received and the node B sends the Get message, only 1 unit time interval is needed to be waited, and then the node B can be considered to leave, the node B is abandoned to receive the Get message from the node B, and the sending task is finished.

And finally, the node A sends the coding packet to the node B, and the node B receives the coding packet and then decodes the coding packet to obtain a packet in the block and verifies the transaction information in the packet. And then according to the channel information sent by the node C, if the channel state information is consistent with the channel state information between the nodes AB, directly sending the coding packet. If not, a new coded packet is sent to the node C. When the node C receives and verifies the block, it sends a FIN message to the node B indicating that the block verification has been completed.

(33) Block decoding algorithm

The decoding algorithm of the invention is used for decoding the received coding packet to obtain grouping information and finally obtain the block. Specifically, as shown in algorithm 2:

algorithm 2

In algorithm 2, when node j receives the encoded packet S, the degree in the encoded packet is first taken out, and the original packet position is saved (steps B1 to B2). And carrying out XOR on the packet data in the coding packet according to the coding packet with the degree of 1, and continuously iterating to recover the original packet (steps B3 to B5). If all the packets have been restored, the packet information is verified, and the block is restored according to the saved packet positions, finally obtaining the complete block (steps B6 to B11).

According to the algorithm 1, the algorithm time complexity is mainly influenced by the selection of the degree distribution function, and according to the degree distribution function, the probability of selecting the transaction number in each group is about ln (| W |/M) _t ) W is the number of packets, M _t Is the transaction number. The temporal complexity in algorithm 1 is therefore O (| W |/M) _t )). The time complexity of the algorithm 2 mainly consists of decoding and iteration degrees, so the time complexity is O (k). The spatial complexity for algorithm 1 and algorithm 2 is mainly determined by the code packets transmitted by the nodes, and therefore the spatial complexity is O (k).

(IV) verification

In order to test the transmission performance and reliability of the ILT algorithm and the protocol based on the block transmission, experiments are designed and compared with the block transmission method of Velocity and Kadcast. The method comprises the steps of firstly constructing an experimental environment, designing an experimental process and specific experimental parameters, then expressing and comparing the specific direction of the experiment by setting three indexes (block synchronization time, block transmission success rate and coding packet retransmission rate), and finally obtaining the result and analysis of the experiment.

(41) Experimental setup

Specific parameter settings are as in table 1:

table 1 parameter setting table

Parameter(s)	Description of the invention	Value taking
			N	Total number of nodes	100、500、1000、5000
B _t	Size of transaction	500bytes
			P _o	Initial node participation probability	1
Θ	On-line threshold	0.3
			Z _c	Probability of current node joining network	0.5
B _l	Maximum node bandwidth	155Mbps
			M _t	Number of transactions	100、500、1000、2000
h	Height of group	8、10
			Tr	Threshold value	30s
S/N	Signal to noise ratio	10dB
			δ	Probability of decoding failure	0.15

The initial node participation rate is set to 1, the online threshold is set to 0.3 and the current node participation rate is set to 0.5.

(42) Index of experiment

In the experiment, three indexes of block transmission time, block transmission success rate and block retransmission rate are measured, and the calculation formulas of the three indexes are as follows:

(d1) Block synchronization time (T) _g )

The block synchronization time represents the time a block is finally acknowledged on the chain, and the expression is as follows:

T _g ＝max{T _i ^a } (7)

in the formula (7), T _i ^a Indicating the time when node i received a block.

(d2) Block Transmission success Rate (P) _b ^t )

The ratio of the number of successfully received blocks to the total number of blocks in time t is represented by the following expression:

in the formula (8), R represents the total number of blocks, I _i Representing the number of blocks received by node i and F representing the total number of nodes.

(d3) Coded packet retransmission rate (Q) _i )

The retransmission rate of the coded packet retransmitted after the block is unsuccessfully transmitted is expressed as follows:

in the formula (9), M represents the total number of transmitted coded packets, K _i Represents the number of nodes i neighbor, V _i Indicating the number of retransmitted coded packets by node i.

(43) Results and analysis of the experiments

(431) Block synchronization time

In the block synchronization time, the transmission performance of the block transmission algorithm is expressed by measuring the difference of the block synchronization time under different nodes, as shown in fig. 14, different block transmission algorithms show great difference when the number of the measured nodes is 100, 500, 1000 and 5000, respectively. When the number of the nodes is 100 and 500, the difference of the three transmission algorithms is not obvious, the three algorithms all adopt a non-rate erasing algorithm, and under the condition that the number of the nodes is less, the difference of the synchronization time is not large. When the number of nodes reaches 1000 and 5000, the block synchronization time of the Velocity algorithm is obviously longer than that of the other two algorithms, because the algorithm adopts the equal-block division principle in the process of dividing blocks, the blocks are divided into equal-size segments, and the block structure and the channel state are not fully considered. The block synchronization time of 1000 and 5000 nodes in the Kadcast algorithm is longer than that of the algorithm of the invention, when the Kadcast algorithm transmits the block, a structured topology transmission block is adopted, and as the block chain network is unstructured, the algorithm needs to construct a tree-shaped transmission model similar to a k-bucket according to the node distribution before transmission, and at the moment, the block is transmitted by optimizing a transmission path, but along with the increase of the number of the nodes, the time and the cost required by constructing the structured transmission path are continuously increased. The block transmission algorithm of the invention combines the channel state and the ILT algorithm, and can reduce the number of transmitted coding packets by reducing the number of packets under the condition of not reducing the decoding efficiency in a congested channel, thereby greatly reducing the synchronization time of the blocks.

The block synchronization time for different block sizes is shown by increasing the transaction number, as shown in fig. 15, the transaction number in the block is different, and the block synchronization time for different block transmission algorithms is greatly different. The same 100 transaction amount is not very different among the three. It does not perform very well when facing smaller block transfers. This is because the transmission time and the transmission rate of the small blocks in the conventional bitcoin network or the ethernet network are short. When the number of transactions in a block is increased to 500, the highest block synchronization time of the Velocity algorithm can be seen through the block synchronization time, and the block synchronization time is increased along with the increase of the transaction amount, because the Velocity algorithm increases the number of divided blocks while the block is increased, and the number of the increased code packets is more than that of the other two algorithms. When the transaction amount reaches 1000, the synchronization time of the Kadcast algorithm is increased compared with that of the method of the invention, and when the transaction amount reaches 2000, the difference between the two is more obvious, because the Kadcast algorithm transmits blocks through different paths in the block transmission process; and this approach does not adequately account for channel congestion caused by the transmission of blocks in the same channel. The ILT algorithm of the invention can increase transaction in the block and adaptively adjust the packet number according to the state of the channel, reduce the transmission size of the coding packet and improve the block synchronization time. Thus, as intra-block transactions increase, the ILT algorithm can also exhibit less synchronization time than other algorithms.

(432) Block transmission success rate

The block transmission success rate is the probability that a block can be finally and successfully determined on a block chain through transmission between nodes, and the difference of the block transmission success rates of different transmission algorithms can be seen through different node numbers. As shown in fig. 16, when the number of nodes is 500 and the number of nodes is 5000, the block transmission success rate characteristics of the three algorithms are displayed. When the node is 500, the median of the Velocity algorithm is about 0.8, while the Kadcast algorithm is between 0.87 and 0.89, and the ILT algorithm is about 0.9. Therefore, the Velocity algorithm is obviously insufficient in block transmission success rate compared with other algorithms because the algorithm does not consider the limited condition of a transmission channel when transmitting the coded packet, and for a part of nodes with limited transmission bandwidth, a certain amount of coded packet information cannot be received in time, which results in the reduction of the block transmission success rate. For the Kadcast algorithm, the advantage is that the transmission path is considered and the packet is lost during transmission, so the block transmission success rate at the node 500 is not much different from that of the ILT algorithm. However, when the node is 5000, the block transmission success rate is reduced due to instability of part of nodes in the network. However, the ITL algorithm of the present invention fully considers the status of the transmission channel and ensures that the block is quickly and stably transmitted to other nodes by preferentially transmitting appropriate coded packet information, even when the bandwidth of some nodes is limited and unstable, the block can be transmitted to the limited node through other neighboring nodes by a new transmission protocol, thereby improving the reliability of transmission.

As shown in fig. 17, the block transfer success rate at different transaction amounts shows a tendency to decrease as the transaction amount increases. Since as transactions increase, the synchronization time and the verification time of the blocks also increase. The verification time is an important factor of the block transmission success rate, and the specific verification time is divided into calculation power verification and intra-block transaction verification, wherein the calculation power verification is to verify whether the block is Fu Gesuan power requirement, and the transaction verification is to confirm whether the transaction information exists. In the Velocity algorithm, block verification is performed only after a block is completely received, which not only affects the transmission of the block, but also delays the block verification time; meanwhile, the Velocity algorithm equally transmits the blocks, and missing coding packet information is constructed through a non-rate erasure algorithm; therefore, the Velocity algorithm is superior to the Kadcast algorithm in the block transmission success rate; but its verification is delayed somewhat, which greatly affects the final validation of the block. While the Kadcast algorithm increases with the transaction quantity, the block transmission success rate is reduced rapidly, and the Kadcast algorithm adopts UDP transmission, so that the stability of the received packet is poor; although the Kadcast algorithm employs the FEC algorithm to reduce packet retransmission, the integrity of the block cannot be guaranteed in the case of limited transmission channel. The ILT algorithm of the invention divides the transaction according to the Mi-Merkl Tree structure, which ensures that each group is a complete transaction information, and also provides a way of verifying the transaction information. When the transaction amount is increased, the influence of a transmission channel can be reduced and the block transmission success rate can be increased according to the number of the channel control coding packets.

(433) Coded packet retransmission rate

The retransmission probability of the coded packet is considered that the algorithm can accelerate the block transmission and simultaneously ensure the stability of the transmission, which is beneficial to the verification and the final confirmation of the block. As shown in FIG. 18, when the number of nodes is 500, the difference in overall data performance of the three algorithms is not too large, the coded packet retransmission rate of the Velocity algorithm is concentrated in the range of 0.19 to 0.25, and the Kadcast and ILT algorithms are concentrated in the ranges of 0.22 to 0.27 and 0.15 to 0.25, respectively. When the number of nodes is 5000, firstly, the retransmission rate of the coded packet of the Kadcast algorithm is very high, the retransmission rate is concentrated between 0.5 and 0.57, and the median and the mean are concentrated at 0.55, which is much larger than those of the Velocity and ILT algorithms. This is because Kadcast uses UDP for transmission, which is unreliable. Although the Kadcast algorithm also reduces the packet retransmission probability through the FEC algorithm, the packet retransmission probability of the node will rise greatly when the bandwidth of the transmission node is limited. The Velocity algorithm and the ILT algorithm are centralized in data comparison, because both adopt the packet transmission algorithm of the coded packets, and a receiving party can recover the lost coded packets after receiving a certain coded packet, so that the packet retransmission rate is lower than that of the Kadcast algorithm.

As shown in fig. 19, the retransmission rate of the coded packets in different transaction amounts continuously increases with the increase of the transaction amount. When the transaction amount is 500, kadcast is 0.1 higher than the Velocity algorithm and 0.2 higher than the ILT algorithm. And as the volume of transactions increases, the gap also expands. This is because the Kadcast algorithm increases blocks when the transaction amount increases, and requires a larger bandwidth when transmitting the blocks, and the transmission performance of the node is still unchanged, so the packet retransmission rate also increases. The modes that the Velocity algorithm and the ILT algorithm encode the blocks firstly and then transmit the blocks are adopted, the condition that the transmission is limited due to the increase of the blocks is reduced, and the reliability of each transmission is improved by adopting TCP (transmission control protocol) transmission in the two algorithms; however, the Velocity algorithm does not measure the transmission channel and does not reasonably divide the blocks, which causes most of the packet loss phenomena in transmission that the node cannot receive a certain coded packet in time to cause retransmission. The ILT algorithm of the present invention increases the transmission rate by decreasing the number of packets according to the channel state when the number of transactions increases. Thus, even with ever increasing traffic, the Bao Chongchuan rate of the ILT algorithm is low compared to the other two algorithms.

Examples

The experimental environment is that the operating system is Windows 10, the processor is i 7.0 GHz,8G running memory. An Omnet + +5.4 simulator which is a discrete event driven simulator is adopted to simulate the block transmission process in a block chain network, and mainly comprises the following components:

(h1) The block generates an event: in the experiment, the ore digging part controls the block generation through a timer, and the ore digging time generated in the ore digging event is obtained through a random value.

(h2) Block transmission event: the block transmission event is started when a block generation event on the node is finished.

To simulate the variation of block size during block transfer, the block size is determined by the number of transactions involved, and the transaction size is fixed to 500bytes during simulation. The operation times of the embodiment are 50 times, and the online node of each experiment meets the Markov chain, which accords with the instability of the node in the block chain network. Thus node engagement P at a time _n Satisfy the requirement ofThe following formula:

in the formula (10), P _n Representing node on-line probability, P _o Representing the last online probability of the node, and l representing a fixed online time coefficient; z _i Represents the probability of node i rejoining the network, as shown in equation (11):

equation (11) probability setting for node rejoining, C _e Represents the length of the current chain; c _m Is a random value of ore excavation, where C _m ∈{1,2,…,C _e }；Z _c Expressed as the probability of the current node joining.

P of current node _n If the node is larger than or equal to the set online threshold theta, the node is represented as online; otherwise, it means that the node is not online.

Claims

1. An ILT-based block coding transmission method in a block chain network is characterized by comprising the following steps:

s20, coding the block by adopting a block coding algorithm;

2. The method for ILT-based block coding transmission in a block chain network according to claim 1, wherein in step S10, the transmission status of the channel is predicted by calculating an average transmission bandwidth of the channel, and the calculation formula of the average transmission bandwidth is as follows:

wherein P is the average transmission bandwidth; q. q.s _max Representing the maximum amount of data transmitted during transmission, q _max ＝max{q ₁ ,q ₂ ,q ₃ ,…,q _i },1≤i≤N；q _min Represents a minimum amount of transmission data; 1200 denotes two unit times 1200 seconds;

the estimated time for a block to pass through the channel is as follows:

wherein D is the distance between nodes, and S/N is the signal-to-noise ratio;

3. The method according to claim 1, wherein the encoding of the block in step S20 is a process of linearly combining several packets to generate a new encoded packet, where each encoded packet is independent, and the specific steps are as follows:

s211, arranging all transactions in groups according to a sequence;

s212, generating a degree d according to the degree distribution function;

4. The ILT-based block coding transmission method of claim 3, wherein a degree d is generated by a degree distribution function, then d random numbers are generated, and d packets are selected according to the random numbers, and the size of the random numbers should be greater than 0 and smaller than the packet number w:

wherein M is _t Representing a total number of transactions in the block; h represents the Mi-Merkle Tree height formed by the exchanges that make up the grouping; 2 ^h-1 Representing the number of transactions divided into a group;

carrying out XOR on the selected groups and recording the original positions of the groups to form a coding packet;

wherein s is the number of packets with the degree of 1, W represents the total number of packets, and delta represents the decoding failure probability; e represents the total number of the encoding packets, and E is more than W and more than s;

5. The method according to claim 1, wherein in step S30, the block transmission protocol uses a modified PiChu block transmission protocol to transmit the encoded packet, the verification of the block is to verify a block header, and the verification of the block is to identify the block information.

6. The ILT-based block coding transmission method in a block chain network according to claim 1, wherein in step S40, the specific steps of decoding by the block decoding algorithm are as follows:

s232, finding out the code packet symbol with the degree d of 1, recovering the group S1 corresponding to the code packet symbol, deleting the code packet with the degree of 1, carrying out XOR on the group S1 and other code packets connected with the group S1, and then storing the group S1 in a block, so that the code packet degree connected with the group S1 can be reduced by 1;

s234, constructing the intra-packet transaction according to the Mi-Merkle Tree sequence to obtain a complete block.

7. The ILT-based block coding transmission method in a block chain network according to claim 6, wherein the Mi-Merkle Tree is hash-grouped according to the original Merkle Tree in the block according to a certain height transaction;

when the transaction number is M _t When the value is, the height of Merkel tree is