Optimal Byzantine data attack method for multi-hop multi-relay wireless sensor network

Abstract

The invention relates to an optimal Byzantine data attack method for a multi-hop multi-relay wireless sensor network, and belongs to the technical field of wireless communication. According to the method, by analyzing the characteristics of the time-varying channel and the structure of the multi-hop multi-relay wireless sensor network, the condition that the average mutual information amount in the binary system channel is zero is found, that is, the sum of the overturning probabilities alpha i and beta i of the judgment result of the sensor node is 1, and the condition can enable the transmission quality of the channel to be worst. Therefore, by controlling the attack probability of the Byzantine node, the sum of the overturning probabilities alpha i and beta i of the judgment result of the sensor node is 1, the optimal Byzantine data attack is realized, and the method has important theoretical and practical significance for researching reliable transmission of sensor data.

Description

Technical Field

[0001] The invention relates to an optimal Byzantine data attack method for a multi-hop multi-relay wireless sensor network, belonging to the technical field of wireless communications. Background Art

[0002] With the rapid development of the Internet of Things, the manufacturing industry has ushered in unprecedented changes and opportunities. Modern manufacturing is moving towards a highly intelligent and automated direction, among which unmanned factories are gradually becoming an important trend. In the equipment manufacturing process of unmanned factories, sensors are distributed in various equipment and production lines to collect parameters such as temperature, pressure, vibration, and speed in real time. These sensor nodes transmit the collected data to the central control system through wireless sensor networks (WSNs) for data analysis and intelligent decision-making. As one of the core technologies of unmanned factories, sensor technology plays a key role, enabling various parameters in the equipment manufacturing process to be monitored and analyzed in real time, and optimizing the production process through the application of advanced automation equipment and intelligent systems, realizing the comprehensive digitization and networking of the production process, thereby greatly improving production efficiency and product quality.

[0003] In the context of "empowering everything with data", the application field of IoT communication technology continues to expand. The demand for data acquisition in wide-area coverage scenarios such as predictive maintenance of industrial production equipment and power transmission equipment, farmland environment monitoring and management, remote sensing monitoring of atmospheric environment and forest fires, and rural smart medical care continues to rise. Due to the nonlinear relationship between transmission power and communication distance, using a wireless multi-hop relay (Multi-hop Multi-relay) network to transmit sensor data to the fusion center is an inevitable choice to reduce the energy consumption of sensor nodes and increase service life. This technology gradually forwards data through multiple relay nodes, thereby extending the transmission distance, enhancing network coverage, and improving overall transmission efficiency. However, due to the time-varying characteristics of wireless channels, they are extremely vulnerable to attacks. Especially in multi-route multi-relay networks, Byzantine data attacks are a common form of malicious attacks. This attack method is to attack the network by capturing normal sensor nodes and maliciously tampering with the judgment results of the sensor nodes. However, in the implementation process of the existing optimal Byzantine data method, it is necessary to accurately know the state of the source, the channel state information of the sensor network, the quantization threshold of the sensor node and other prior information. However, if the channel between the information source and the sensor node, or the channel between the sensor node and the fusion center, is a mobile channel that changes dynamically at an extremely fast speed, or the statistical characteristics of the channel state are completely unavailable, it is impossible to accurately obtain the channel state information. Under this condition, the Byzantine attack node cannot obtain any prior information, and it is still unclear whether the optimal Byzantine data attack method exists. Summary of the invention

[0004] The purpose of the present invention is to provide an optimal Byzantine data attack method for a multi-hop multi-relay wireless sensor network to solve the problem that the attack effect of the optimal Byzantine data attack method is unclear when the Byzantine attack node cannot obtain any prior information.

[0005] In order to solve the above technical problems, the present invention provides an optimal Byzantine data attack method for a multi-hop multi-relay wireless sensor network, the method comprising the following steps:

[0006] Select some sensors from a multi-hop multi-relay wireless sensor network as Byzantine nodes;

[0007] Control the attack probability of Byzantine nodes when the fusion center is unaware of the attack strategy of Byzantine nodes and the Byzantine nodes do not know any channel state information and The flip probability α of the original hard decision result of the sensor node i With β i The following conditions are met;

[0008] α i +β i =1

[0009] in, ρ is the proportion of sensors attacked by Byzantine to all sensors, It represents the probability of flipping the original hard decision result 0 to 1 under the condition that the sensor node is attacked by Byzantine. represents the probability of flipping the original hard decision result 1 to 0 under the condition that the sensor node is attacked by Byzantine, α i Indicates the probability of being flipped to 0 when the original hard decision result is 1, β i It indicates the probability of being flipped to 1 when the original hard decision result is 0.

[0010] Furthermore, the Byzantine nodes are not aware of all channel state information from the source to the fusion center, that is, the Byzantine nodes do not have any prior information about the distributed detection system.

[0011] The beneficial effects of the present invention are as follows: as an improved invention, the present invention finds the condition that makes the average mutual information in the binary channel zero by analyzing the characteristics of the time-varying channel and the structure of the multi-hop multi-relay wireless sensor network, that is, the flip probability α of the decision result of the sensor node i With β iThe sum of is 1, which can make the channel transmission quality the worst. Therefore, the present invention controls the attack probability of the Byzantine node to make the flip probability α of the decision result of the sensor node i With β i The sum is 1, realizing the optimal Byzantine data attack, which has important theoretical and practical significance for studying the reliable transmission of sensor data and is easy to implement. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a working flow chart of a multi-hop multi-relay wireless sensor network of the present invention;

[0013] Figure 2 is a schematic diagram of a binary channel used in an embodiment of the present invention;

[0014] Figure 3 is a relationship diagram of the average mutual information and conditional entropy of the binary channel used in the embodiment of the present invention and the false alarm probability and detection probability;

[0015] Figure 4 It is the entire channel state information from the unknown information source of the Byzantine node to the fusion center in the embodiment of the present invention, that is, the channel model diagram corresponding to the case where the Byzantine node has no prior information about the distributed detection system;

[0016] Figure 5 This is a schematic diagram of the performance impact of the attack method in the first attack situation on the multi-hop multi-relay wireless sensor network;

[0017] Figure 6 This is a schematic diagram of the performance impact of the attack method in the second attack case on the multi-hop multi-relay wireless sensor network. DETAILED DESCRIPTION

[0018] The specific implementation of the present invention will be further described below in conjunction with the accompanying drawings.

[0019] By analyzing the characteristics of time-varying channels and the structure of multi-hop multi-relay wireless sensor networks, the present invention finds the condition that makes the average mutual information in the binary channel zero, that is, the flip probability α of the decision result of the sensor node i With β i The sum of the two conditions is 1, which can make the channel transmission quality the worst. On this basis, the present invention selects some sensors from a multi-hop multi-relay wireless sensor network as Byzantine nodes; controls the attack probability of the Byzantine nodes, so that the flip probability α of the judgment result of the sensor node i With β i The sum is 1, thus achieving the optimal attack.

[0020] The following is a study on the conditions that can make the channel transmission quality the worst and achieve the optimal Byzantine data attack.

[0021] 1. Determine the conditions under which the average mutual information of a binary channel is zero.

[0022] The working process of the multi-hop multi-relay wireless sensor network of the present invention is as follows: Figure 1 As shown in , each sensor node transmits the sensed data to the fusion center through the relay node, and the adjacent nodes use binary channels for transmission. For a binary channel, Figure 2 As shown, the binary data it sends is S=0 or S=1, and the output binary data is R=0 or R=1. D Indicates the case where the sending signal is S=1 and the output signal is R=1, with P F It represents the case where the transmitted signal is S = 0 and the output signal is R = 1. The average mutual information of the transmitted information S and the received information R is represented by:

[0023] I(S;R)=h(S)-h(S|R) (1)

[0024] Because the worse the channel condition, the smaller the average mutual information I(S; R), that is, the larger the conditional entropy h(S|R), if you want to get the minimum average mutual information, you only need to maximize the conditional entropy h(S|R). The conditional entropy h(S|R) is expressed as:

[0025]

[0026] in:

[0027]

[0028] Substituting (3) into (2), we can obtain the relationship between the average mutual information and conditional entropy and the false alarm probability and detection probability, as follows: Figure 3 As shown, the specific expression of conditional entropy h(S|R) is:

[0029]

[0030] The conditional entropy h(S|R) is F Find the first-order derivative and set it equal to zero, and then find the extreme point. The process is as follows:

[0031]

[0032] The extreme point can be obtained as: P D =P F

[0033] Will get P D =P FIt is the extreme point of conditional entropy h(S|R). It is necessary to further determine whether the extreme point is a maximum point or a minimum point. Therefore, the second-order derivative is further calculated. The process is as follows:

[0034]

[0035] The extreme point P D =P F Substituting into the conditional entropy:

[0036]

[0037] I(S;R) min =h(S)-h(S|R) max =h(S)-h(S)=0 (8)

[0038] So far, we have obtained that when the P of the entire transmission channel D =P F When , the average mutual information of the entire system reaches its minimum value, and the minimum value of the average mutual information is zero.

[0039] According to the knowledge of information theory, the average mutual information of the entire system is converted into the average mutual information of a single branch. Since the channel is a memoryless channel, according to the knowledge of information theory, we can get:

[0040]

[0041] Since the component v of the input vector V i Independent of each other, according to information theory knowledge, we can get:

[0042]

[0043] Therefore, combining (9) and (10), we can obtain:

[0044]

[0045] At this point, we have obtained that the average mutual information of the entire system is equal to the sum of the average mutual information of all paths.

[0046] 2. Determine the condition that the average mutual information is 0 when the Byzantine nodes have no prior information about the distributed detection system, that is, when all channel state information from the Byzantine node to the fusion center is unknown.

[0047] When the Byzantine nodes do not know the entire channel state information from the source to the fusion center, the corresponding channel model is as follows: Figure 4 As shown, at this time, the channel transfer probability matrix of the Byzantine node's known channel is:

[0048]

[0049] Using the condition P of zero average mutual information obtained in step 1 D =P F ,exist Figure 4 That is, it needs to meet:

[0050] P(u i =1|x i =1) = P(u i =1|x i =0) (13)

[0051] Right now:

[0052] α i +β i =1 (14)

[0053] Therefore, it can be obtained that the condition for the average mutual information to be zero in this case is α i +β i =1, at this time the channel transmission quality is the worst.

[0054] Taking various situations into consideration, the present invention provides an optimal Byzantine data attack method under the condition that the fusion center does not know the Byzantine attack strategy, the Byzantine attack node does not know any channel state information, that is, the Byzantine node does not have any prior information about the distributed detection system. The method selects some sensors from a multi-hop multi-relay wireless sensor network as Byzantine nodes; controls the attack probability of the Byzantine node, so that the flip probability α of the decision result of the sensor node i With β i The sum is 1, thus achieving the optimal attack. To further verify the feasibility of the solution of the present invention, a simulation experiment of the optimal Byzantine estimation method of the present invention is conducted below.

[0055] The impact of different Byzantine attack probabilities on the system bit error rate performance is as follows Figure 5 and Figure 6 As shown, Figure 5 and Figure 6 The horizontal axis is the attack probability (Crossover probability), and the vertical axis is the bit error rate (BER). i +β i When it approaches 1, the system bit error rate becomes larger and larger. i +β i =1 when the bit error rate reaches its maximum value. Figure 5 The simulation conditions are set as the number of routers N = 12, the number of relays J = 3, the Byzantine flip probabilities α = 0.2, β = 0, β = 0.2, β = 0.4, β = 0.6, β = 0.8, and β = 1; Figure 6The simulation conditions are set as N = 12 routers, J = 3 relays, α = 0.4, β = 0, β = 0.2, β = 0.4, β = 0.6, β = 0.8 and β = 1. It can be found that no matter how many sensors and relays there are, as long as the Byzantine flip probability satisfies α i +β i =1, the system performance will be completely ineffective.

[0056] In summary, by analyzing the characteristics of time-varying channels and the structure of multi-hop multi-relay wireless sensor networks, the present invention finds the condition that makes the average mutual information in the binary channel zero, that is, the flip probability α of the decision result of the sensor node i With β i The sum of is 1, which can make the channel transmission quality the worst. Based on the above conditions, the present invention can control the attack probability of the Byzantine node to make the flip probability α of the decision result of the sensor node i With β i The sum is 1, realizing the optimal Byzantine data attack, which has important theoretical and practical significance for studying the reliable transmission of sensor data. This method has the characteristics of strong applicability, high reliability and easy implementation.

Claims

1. An optimal Byzantine data attack method for a multi-hop multi-relay wireless sensor network, characterized in that: The method comprises the following steps: Select some sensors from a multi-hop multi-relay wireless sensor network as Byzantine nodes; Control the attack probability of Byzantine nodes when the fusion center is unaware of the attack strategy of Byzantine nodes and the Byzantine nodes do not know any channel state information and The flip probability α of the original hard decision result of the sensor node i With β i The following conditions are met; α i +β i ＝1 in, ρ is the proportion of sensors attacked by Byzantine to all sensors, It represents the probability of flipping the original hard decision result 0 to 1 under the condition that the sensor node is attacked by Byzantine. represents the probability of flipping the original hard decision result 1 to 0 under the condition that the sensor node is attacked by Byzantine, α i Indicates the probability of being flipped to 0 when the original hard decision result is 1, β i It indicates the probability of being flipped to 1 when the original hard decision result is 0.

2. The optimal Byzantine data attack method for a multi-hop multi-relay wireless sensor network according to claim 1, characterized in that: The Byzantine nodes are unknown to the entire channel state information from the information source to the fusion center, that is, the Byzantine nodes do not have any prior information about the distributed detection system.

Images