A wireless sensor network distributed elastic clock synchronization method based on dynamic weight value updating

Abstract

The application discloses a wireless sensor network distributed elastic clock synchronization method based on dynamic weight value updating, and comprises the following steps: step 1, each node of the wireless sensor network sends its current relevant parameter information to all neighbor nodes, and meanwhile, receives the information sent by the neighbor nodes; step 2, a distributed weighted clock synchronization algorithm is designed; step 3, a false data injection (FDI) network attack is identified; step 4, a time-varying reputation value rep is introduced ij (k) dynamically updating the weight value w ij , a security defense mechanism is designed to repair the communication channel disconnected due to the FDI attack, and elastic clock synchronization is realized; and step 5, the above steps 2-4 are repeated until the clock synchronization is finished. The application has good robustness and expandability, can be used for enriching the theory and application of the wireless sensor network clock synchronization, and provides a new idea for realizing clock synchronization.

Description

Technical Field

[0001] This invention relates to the field of wireless sensor network technology, and in particular to a distributed elastic clock synchronization method for wireless sensor networks based on dynamic weight updates. Background Technology

[0002] In recent years, with the rapid development of technologies such as computer networks and wireless communication, various application terminals have increasingly higher requirements for network latency. Wireless sensor networks, as one of the carriers of this technology, are widely used in many fields such as traffic control, environmental monitoring, smart homes, and the military. Among them, clock synchronization technology is the foundation and prerequisite for the implementation of many applications in wireless sensor networks, achieving a unified time standard by correcting the local clocks of nodes. To improve the scalability and robustness of clock synchronization protocols, many scholars have begun to study distributed clock synchronization algorithms in recent years. Each sensor node receives information from its neighboring nodes and uses the neighboring nodes' information and its local time to update the virtual clock parameters. Therefore, distributed clock synchronization algorithms have strong robustness and scalability, making them particularly suitable for large-scale wireless sensor networks.

[0003] Recently, a distributed weighted consensus clock synchronization (WCCS) can realize distributed clock synchronization between network nodes. In this distributed algorithm, each node periodically exchanges local clock readings with neighboring nodes and uses this time information to calculate its relative offset and tilt relative to neighboring nodes, such as technique [1] (see H Aissaoua, M Aliouat, A Bounceur, R Euler. A distributed consensus-based clock synchronization protocol for wireless sensor networks. Wireless Personal Communications 2017; 95: 4579-4600.). However, it is noted that in actual wireless sensor network connections, malicious network attacks may occur, and the communication information of the sensors may be stolen or tampered with, resulting in a decrease in sensor performance or even failure to complete the expected tasks. False data injection (FDI) is an attack that destroys the data integrity of the system by injecting different signals to replace real information or change the integrity of real information. This attack is one of the most threatening attack methods against state estimation in wireless sensor networks. Compared to denial-of-service attacks, FDI attacks are more covert and harder to detect. On the one hand, attackers can easily inject false data into the channel; on the other hand, the stealthy nature of FDI attacks allows them to remain hidden for extended periods, leaving the system in a critical, dangerous state. By the time it is discovered, the damage caused by the attack may have already led to the collapse of the entire system. Therefore, studying the clock synchronization problem of wireless sensors under FDI attacks has significant practical implications.In wireless sensor networks, some scholars have used statistical outlier detection methods to identify malicious nodes during the synchronization process to deal with false data injection attacks, and then used weighted consistency clock synchronization algorithm to achieve clock synchronization, such as technique [2] (see AHabib, A Laouid, M Kara. Secure Consensus Clock Synchronization in WirelessSensor Networks. 2021 International Conference on Artificial Intelligence for Cyber ​​Security Systems and Privacy (AI-CSP) 2021, 1-6.). However, it is noted that this technique cannot effectively distinguish the impact of occasional random disturbances and malicious attacks on transmitted data. It can only permanently isolate related nodes through one detection, lacking a certain robustness and resilience against network attacks. Reference [3] (LAPhan, T. Kim, T. Kim, Robustneighbor-aware time synchronization protocol for wireless sensor network in dynamic and hostile environments. IEEE Internet of Things Journal, 2020, 8(3), 1934-1945.) proposes a neighbor-aware time synchronization protocol. Based on the state of neighboring nodes, each node can choose an appropriate operation (such as deleting outdated messages and ignoring messages from unsynchronized nodes). However, in the implementation of this protocol, there is a limitation on the number of unsynchronized neighbors of a node. If a node has n unsynchronized neighbors, it needs at least 2n synchronized neighbors to maintain its synchronized state. That is, there is a certain limitation on the topology connection after an attack, which is often difficult to meet in actual wireless sensor networks. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a distributed elastic clock synchronization method for wireless sensor networks based on dynamic weight update, which has good robustness and scalability, can be used to enrich the theory and application of clock synchronization in wireless sensor networks, and provide new ideas for realizing clock synchronization.

[0005] To address the aforementioned technical problems, this invention provides a distributed resilient clock synchronization method for wireless sensor networks based on dynamic weight updates, comprising the following steps:

[0006] Step 1: Let the set of nodes in the wireless sensor network be V, the total number of nodes be n, the set of edges be E, and the local clock of node i at time t be τ.i (t), initialize the relative clock slope α of each node. ij Slope compensation parameters Each node sends its current relevant parameter information to all neighboring nodes, and at the same time, receives information sent by neighboring nodes;

[0007] Step 2: Design a distributed weighted clock synchronization algorithm;

[0008] Step 3: Identify FDI network attacks involving fake data injection.

[0009] Step 4: Introduce time-varying reputation values ​​(rep) ij (k) Dynamically update weights w ij Design a security defense mechanism to repair communication channels that have been disconnected due to FDI attacks and achieve resilient clock synchronization;

[0010] Step 5: Repeat steps 2-4 above until clock synchronization is complete.

[0011] Preferably, in step 2, designing the distributed weighted clock synchronization algorithm specifically involves estimating the relative clock slope. And calculate offset compensation Let the virtual clock of node j at time t be... The local clock of node i is τ i (t), the offset between the two is Then the clock tilt error α ij The calculation formula is:

[0012]

[0013] Thus relative clock rate Clock tilt error α at node i i (k+1) is updated according to the following formula.

[0014]

[0015] Where k represents the number of iterations, w ij The weight of node i with respect to node j is defined as follows:

[0016]

[0017] Where d j Indicates the degree of node j;

[0018] Offset compensation of node i at time t1 The iterative update formula is:

[0019]

[0020] in The offset compensation of node i to node j at time t1 is calculated according to the following formula.

[0021]

[0022] Preferably, in step 3, identifying FDI (Fake Data Injection) network attacks specifically involves: Let δ(k) be the detection threshold function, G... ij (k) is the number of times node i receives verified correct information from its neighbor node j within cycle k, and is called a counter. If node i's neighbor node j does not transmit information to node i at a certain moment, or if the information transmitted by node j differs significantly from the data estimated by node i for that moment, i.e. The system then identifies that it is under FDI attack, G ij (k)=G ij (k-1); If node i's neighbor node j transmits information to node i at a certain time, and the difference between the transmitted data and the estimated value is within a certain threshold, i.e. The system then identifies that it has not been subjected to an FDI attack, G ij (k)=G ij (k-1)+1, specifically represented as

[0023]

[0024] Preferably, in step 4, a time-varying reputation value (rep) is introduced. ij (k) Dynamically update weights w ij The design of a security defense mechanism to repair communication channels lost due to FDI attacks and to achieve resilient clock synchronization specifically involves: designing a system for node i to cycle through the reputation value rep on node j k. ij The formula for calculating (k) is:

[0025]

[0026] Where η(k) is a monotonically decreasing function of k, and satisfies For example, it is advisable If G ij (k)=G ij (k-1)=k, then rep ij (k)=1. If G ij (k)=G ij (k-1)+1, then the reputation value rep ij (k) will decrease;

[0027] Let the reputation threshold be rep. th When the reputation value is below the threshold, i.e., rep ij (k)<rep thweight w ij =0, consider the network's priority connection characteristic to repair the network, that is, new nodes are more likely to connect to nodes with lower connectivity; if the communication channel between node m and node l is broken, node m cannot use information from node l. Let m be the set of neighboring nodes. The degree d of all neighboring nodes i Sort the data (i = 1, 2, ..., n) such that if...

[0028] d k <d1<d2<…<d n ,

[0029] Nodes k and m are selected to reconstruct the communication channel. For nodes with the same degree, the node with the smallest node number is selected to reconstruct the communication channel. Similarly, the communication channel corresponding to node l is reconstructed, and the new reputation value is calculated.

[0030] When the reputation score rises above the threshold, i.e., rep ij (k)≥rep th The weight update rule is defined by the proportion of the reputation value of the node's neighboring nodes in the sum of the reputation values ​​of all nodes at the current moment.

[0031]

[0032] Skip to step 2 and adjust the weights w ij (k) Update the virtual clock of neighboring nodes. reading.

[0033] The beneficial effects of this invention are as follows: This invention possesses good robustness and scalability. The security defense mechanism designed against FDI attacks repairs the network, ensuring network topology connectivity and improving network topology stability. By introducing a dynamically updated time-varying reputation value update rate and designing a distributed elastic clock synchronization algorithm with dynamically updated weights, relative clock slope, and offset compensation, this algorithm can dynamically update weights, relative clock slope, and offset compensation. This algorithm can improve the resilience and survivability of wireless sensor networks in a fully distributed manner by identifying and isolating false information injection attacks on nodes. It can be used to enrich the theory and application of clock synchronization in wireless sensor networks, providing new ideas for achieving secure clock synchronization. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0035] Figure 2 This is a schematic diagram of the network node topology of the present invention.

[0036] Figure 3This is an error curve between virtual clocks under the FDI attack of this invention, without dynamic weight updates and topology repair.

[0037] Figure 4 This is a graph showing the error curves between virtual clocks in the distributed elastic clock synchronization algorithm based on reputation value dynamic weight update under FDI attack of this invention.

[0038] Figure 5 This is a schematic diagram illustrating the trend of reputation value changes of node 1, the monitoring node of the present invention. Detailed Implementation

[0039] like Figure 1 As shown, a distributed resilient clock synchronization method for wireless sensor networks based on dynamic weight updates includes the following steps:

[0040] Step 1: Let the set of nodes in the wireless sensor network be V, the total number of nodes be n, the set of edges be E, and the local clock of node i at time t be τ. i (t), initialize the relative clock slope α of each node. ij Slope compensation parameters Each node sends its current relevant parameter information to all neighboring nodes, and at the same time, receives information sent by neighboring nodes;

[0041] Step 2: Design a distributed weighted clock synchronization algorithm;

[0042] Step 3: Identify FDI network attacks involving fake data injection.

[0043] Step 4: Introduce time-varying reputation values ​​(rep) ij (k) Dynamically update weights w ij Design a security defense mechanism to repair communication channels that have been disconnected due to FDI attacks and achieve resilient clock synchronization;

[0044] Step 5: Repeat steps 2-4 above until clock synchronization is complete.

[0045] Let τ i (t) is the local clock of the i-th sensor node at time t, typically modeled as

[0046] τ i (t)=a i t+b i ,

[0047] Among them, a i Indicates clock rate, b i This represents the clock offset. Assume τ i (t) and τ j (t) are the local clocks of nodes i and j, and their relative clocks can be defined as follows:

[0048]

[0049] in Indicates relative clock rate, Indicates relative clock offset. Defines a virtual clock. for

[0050]

[0051] in Let be the clock rate compensation amount for node i at time t. Let be the compensation amount for the clock reading deviation at node i at time t. Let the common clock at time t be τ. v (t) is

[0052] τ v (t)=α v t+β v ,

[0053] Where α v β represents the common clock rate. v This represents a common clock offset. For clock synchronization algorithms, the goal is to find, for each node, a common clock offset. and Make

[0054]

[0055] that is

[0056]

[0057] The objective of this invention is to design a clock synchronization algorithm that enables nodes in a sensor network to maintain stable clock synchronization even under FDI attacks, thus satisfying the aforementioned design objective. The specific implementation steps are as follows:

[0058] (1) Initialize the relative clock slope α of each node. ij Slope compensation parameters and virtual clock Parameters, etc. Any node in the network sends relevant parameter information to each neighboring node, and at the same time receives clock information sent by neighboring nodes.

[0059] (2) Design a distributed weighted clock synchronization algorithm to estimate the relative clock slope and calculate offset compensation. Let the virtual clock of node j at time t be... and the local clock τ of node i i The offset between (t) is Clock tilt error α ij The calculation formula is

[0060]

[0061] Then relative clock rate Clock tilt error α at node i i (k+1) is updated according to the following formula.

[0062]

[0063] Where k represents the number of iterations, w ij The weight of node i with respect to node j is defined as follows:

[0064]

[0065] Where d j This represents the degree of node j.

[0066] Similarly, the offset compensation of node i at time t1 The iterative update calculation formula is as follows:

[0067]

[0068] Where w ij Let represent the weight of node i with respect to node j, and The offset compensation of node i to node j at time t1 is calculated according to the following formula.

[0069]

[0070] in τ represents the virtual clock of node j at time t2. i (t2) represents the local clock of node i at time t2, τ i (t1) represents the local clock of node i at time t1. Indicates the relative clock rate.

[0071] (3) Identify FDI attacks. Let δ(k) be the detection threshold function, and G ij (k) represents the number of times node i receives verified correct information from its neighbor node j within cycle k, and can be called a counter. If node i's neighbor node j does not transmit information to node i at a certain moment, or if the information transmitted by node j differs significantly from the data estimated by node i for that moment, i.e. Then G ij (k)=G ij (k-1); If node i's neighbor node j transmits information to node i at a certain time, and the difference between the transmitted data and the estimated value is within a certain threshold, i.e. Then G ij (k)=Gij (k-1)+1, specifically represented as

[0072]

[0073] (4) This invention introduces a time-varying reputation value (rep). ij (k) Dynamically update weights w ij Design a security defense mechanism to repair communication channels lost due to FDI attacks and achieve distributed elastic clock synchronization. Design the reputation value rep of node i on node j cyclically on node k. ij The formula for calculating (k) is:

[0074]

[0075] Where η(k) is a monotonically decreasing function of k, and satisfies If acceptable

[0076] Let the reputation threshold be rep. th When rep ij (k)<rep th At that time, the weight w ij =0, considering the network's preferential connection characteristics to repair the network, meaning new nodes are more likely to connect to nodes with lower connectivity. Specifically: if the communication channel between node m and node l is broken, node m cannot use information from node l. Let m be the set of neighboring nodes. The degree d of all neighboring nodes i Sort the data (i = 1, 2, ..., n) such that if...

[0077] d k <d1<d2<…<d n ,

[0078] Nodes k and m are selected to reconstruct the communication channel. For nodes with the same degree, the node with the smallest node number is selected to reconstruct the communication channel. Similarly, the communication channel corresponding to node l is reconstructed, and a new reputation value is calculated.

[0079] When the reputation score rises above this threshold, i.e., rep ij (k)≥rep th The weight update rule is defined by the proportion of the reputation value of the node's neighboring nodes in the sum of the reputation values ​​of all nodes at the current moment:

[0080]

[0081] The weights are dynamically updated based on reputation value. ij (k) Update the virtual clock of neighboring nodes. reading.

[0082] (5) Repeat steps (2)-(4) above until clock synchronization is complete.

[0083] We provide the following Example 1 to illustrate the specific implementation steps and experimental results:

[0084] (1) Determine the topology of the sensor network. Figure 2 The diagram shows the network topology, representing the topology of the wireless sensor network used in the experiment. The network contains 5 nodes, and the set of nodes constituting the sensor network is V = {1, 2, 3, 4, 5}. Assume the local clock deviation range is [0.9999, 1.0001], and the clock offset range is [-0.001, 0.001]. Initialize the relevant parameters of the nodes: For any node i in the network, initialize its relative clock slope estimate α. ij (0) = 1 and slope compensation parameter and virtual clock

[0085] (2) Identify FDI attacks, set the threshold function δ(k) to 0.002, and the statistical counter G ij (k) information, calculate reputation value rep ij (k). Set the reputation threshold rep th It is 0.56 when rep ij (k)<rep th At that time, the weight w ij =0, the original communication channel is disconnected, isolating the impact of the FDI attack. Consider two attack scenarios: At 1025s, the communication link between nodes 4 and 5 is attacked by an FDI attack, injected with a uniformly distributed spoofed data segment [-30, 30]. After 10s, the communication channel between nodes 4 and 5 is disconnected. The security defense mechanism selects node 3 to rebuild the communication channel with node 5, completing network topology repair. At 1100s, the communication link between nodes 2 and 3 is attacked by an FDI attack, injected with a uniformly distributed spoofed data segment [-30, 30]. After 15s, the communication channel between nodes 2 and 3 is disconnected. Since node 5 has a degree of 1, which is the minimum, the security defense mechanism selects node 5 to rebuild the communication channel with node 3, completing network topology repair and calculating a new reputation value.

[0086] (3) Define the error of the virtual clock between node j and node i at time t as: The simulation results are given under an FDI attack, where the communication connection is broken and no topology repair is performed. The error e between virtual clocks is obtained. i The curve of (t), as shown Figure 3As shown, after being attacked by FDI, the error fluctuates until it converges after a period of time, approximately 1500 seconds, after the attack has subsided. A distributed elastic clock synchronization algorithm is designed based on dynamically updated weights according to reputation values. Experimental simulations are conducted on this algorithm under an FDI attack to obtain the error e between virtual clocks. i The curve of (t), as shown Figure 4 As shown, initially the error fluctuated within a small range, compared to Figure 3 After the attack disappears, the synchronization error quickly converges to 0, achieving clock synchronization in the wireless sensor network. Figure 4 The convergence speed is faster.

[0087] (4) Figure 5 It shows the trend of the reputation value of node 1 with respect to node 3, when G ij (k)=G ij When (k-1)=k, then rep ij (k) = 1. Consider two attack scenarios: at times 1025s and 1100s, inject a uniformly distributed spoofed data segment [-30, 30]. When the system recognizes that it is under FDI attack, the counter G... ij (k)=G ij When (k-1)+1, the corresponding reputation value rep ij (k) will decrease, and when the attack disappears, the reputation value will gradually increase again. The remaining nodes will gradually become synchronized, eventually achieving clock synchronization of the wireless sensor network. The reputation value will also converge and eventually reach a stable value.

[0088] (5) Repeat steps (2)-(4) above until clock synchronization is complete.

[0089] This invention proposes a resilient clock synchronization method for wireless sensor networks based on dynamic weight updates. It designs weights that are dynamically updated based on reputation values, dynamically adjusting the weights of communication edges between nodes. This effectively distinguishes the impact of occasional random disturbances and malicious attacks on transmitted data, achieving efficient clock synchronization. The reputation value update parameter η(k) is a dynamic function of k, effectively offsetting the insensitivity to attack isolation caused by increasing k. A security defense topology repair mechanism is designed to promptly replace communication channels disconnected due to FDI attacks, improving the connectivity of the network communication topology. This allows network nodes to achieve distributed and rapid clock synchronization, further enhancing the resilience of the entire network and its resistance to FDI attacks.

Claims

1. A distributed resilient clock synchronization method for wireless sensor networks based on dynamic weight updates, characterized in that, Includes the following steps: Step 1: Let V be the set of nodes in the wireless sensor network, n be the total number of nodes, E be the set of edges, and τ be the local clock of node i at time t. i (t), initialize the clock tilt error α of each node. ij Slope compensation parameters Each node sends its current relevant parameter information to all neighboring nodes, and at the same time, receives information sent by neighboring nodes; Step 2: Design a distributed weighted clock synchronization algorithm; Step 3: Identify FDI network attacks involving fake data injection. Step 4: Introduce time-varying reputation values ​​(rep) ij (k) Dynamically update weights w ij Design a security defense mechanism to repair communication channels that have been disconnected due to FDI attacks and achieve resilient clock synchronization; Design the reputation value rep of node i on the loop k of node j. ij The formula for calculating (k) is: Where η(k) is a monotonically decreasing function of k, and satisfies Pick If G ij (k)=G ij (k-1)=k, then rep ij (k)=1; if G ij (k)=G ij (k-1)+1, then the reputation value rep ij (k) will decrease; G ij (k) is the number of times node i receives verified correct information from its neighbor node j within loop k, and is called a counter; Let the reputation threshold be rep. th When the reputation value is below the threshold, i.e., rep ij (k) <rep th weight w ij =0, consider the network's priority connection characteristic to repair the network, that is, new nodes are more likely to connect to nodes with lower connectivity; if the communication channel between node m and node l is broken, node m cannot use information from node l. Let m be the set of neighboring nodes. The degree d of all neighboring nodes i Sort the data, where i = 1, 2, ..., n. d k <d1<d2<…<d n , Nodes k and m are selected to reconstruct the communication channel. For nodes with the same degree, the node with the smallest node number is selected to reconstruct the communication channel. The communication channel corresponding to node l is reconstructed, and the new reputation value is calculated. When the reputation score rises above the threshold, i.e., rep ij (k)≥rep th The weight update rule is defined by the proportion of the reputation value of the node's neighboring nodes in the sum of the reputation values ​​of all nodes at the current moment. Skip to step 2 and adjust the weights w ij (k) Update the virtual clock of neighboring nodes. reading; Step 5: Repeat steps 2-4 above until clock synchronization is complete.

2. The distributed elastic clock synchronization method for wireless sensor networks based on dynamic weight update as described in claim 1, characterized in that, In step 2, the specific steps of designing the distributed weighted clock synchronization algorithm are: estimating the relative clock rate. And calculate offset compensation Let the virtual clock of node j at time t be... The local clock of node i is τ i (t), the offset compensation between the two is Then the clock tilt error α ij The calculation formula is: Thus relative clock rate Clock tilt error α at node i i (k+1) is updated according to the following formula. Where k represents the number of iterations, w ij The weight of node i with respect to node j is defined as follows: Where d j Indicate the degree of node j; Offset compensation of node i at time t1 The iterative update formula is: in The offset compensation of node i to node j at time t1 is calculated according to the following formula.

3. The distributed elastic clock synchronization method for wireless sensor networks based on dynamic weight update as described in claim 1, characterized in that, In step 3, identifying FDI (Fake Data Injection) network attacks specifically involves: Let δ(k) be the detection threshold function, and G... ij (k) is the number of times node i receives verified correct information from its neighbor node j within cycle k, and is called a counter. If node i's neighbor node j does not transmit information to node i at a certain moment, or if the information transmitted by node j differs significantly from the data estimated by node i for that moment, i.e. The system then identifies that it is under FDI attack, G ij (k)=G ij (k-1); If node i's neighbor node j transmits information to node i at a certain time, and the difference between the transmitted data and the estimated value is within a certain threshold, i.e. The system then identifies that it has not been subjected to an FDI attack, G ij (k)=G ij (k-1)+1, specifically represented as

Images