Robust state estimation method and system for resisting composite jamming cluster multi-sensor network
By constructing a clustered multi-sensor network, using the AES-CTR encryption algorithm and Hamming distance grading, and combining the maximum correlation entropy criterion and information consistency algorithm, the noise model mismatch and secure transmission problems of multi-sensor systems in strong adversarial and high interference environments were solved, achieving high-precision state estimation and data protection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANKAI UNIV
- Filing Date
- 2026-05-29
- Publication Date
- 2026-08-04
AI Technical Summary
Existing technologies for state estimation of multi-sensor systems in environments with strong adversarial and high interference suffer from problems such as mismatch between noise models and actual environments, limitations in computing resources, and difficulty in balancing the requirements for secure transmission with estimation accuracy, thus failing to effectively resist complex interference.
A layered defense strategy is adopted, which involves constructing a clustered multi-sensor network containing multiple clusters, generating GMAC authentication tags using the AES-CTR encryption algorithm, performing security classification based on Hamming distance, and combining the maximum correlation entropy criterion and information consistency algorithm for state estimation to achieve data confidentiality and integrity protection.
While reducing the computational burden on sensors, it ensures the confidentiality and high accuracy of data transmission, improves the robustness and fault tolerance of the system, effectively resists complex interference, and achieves a two-way balance between security and estimation accuracy.
Smart Images
Figure CN122294103B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of state estimation technology for multi-sensor systems, and in particular to a robust state estimation method and system for multi-sensor networks that resists complex interference clusters. Background Technology
[0002] In practical applications such as industrial production, smart grids, and intelligent transportation, multi-sensor systems, composed of multiple sensors connected via communication links, have been widely adopted, significantly improving production efficiency and demonstrating promising application prospects. Classified by organizational structure, multi-sensor systems can be categorized into centralized and distributed types. Distributed multi-sensor systems are complex systems based on specific topologies, consisting of multiple geographically dispersed and logically autonomous network nodes. These nodes are interconnected via communication links and can autonomously and collaboratively complete data acquisition, transmission, and processing. Benefiting from the significant advantages of distributed network architecture in terms of scalability, robustness, resource efficiency, and decentralization, distributed multi-sensor systems are increasingly attracting widespread attention from researchers.
[0003] State estimation technology is one of the important techniques involved in multi-sensor system research. It aims to output an accurate estimate of the system state through the fusion processing of multi-source sensor data. Its core value lies in overcoming the limitations of single-sensor data types and measurement blind spots, achieving efficient utilization of data. However, in practical applications, existing technologies face the following challenges: 1. Performance degradation of estimation algorithms in complex environments. Traditional state estimation methods often assume that noise follows a specific distribution. As application scenarios expand to highly adversarial and high-interference environments, noise in measurement data often exhibits complex statistical characteristics such as heavy-tailed and multi-peaked patterns. This leads to a severe mismatch between the noise model and the actual environment. Under non-Gaussian noise interference, the accuracy of filter estimation decreases significantly, seriously affecting system reliability.
[0004] 2. The conflict between secure transmission requirements and node resources is prominent. In sensor networks, due to the limited computing resources of nodes, some existing complex encryption algorithms are difficult to apply directly and cannot meet the real-time and low-power requirements of encryption, decryption, and authentication processes.
[0005] 3. Balancing estimation accuracy and security requirements is difficult. Existing robust state estimation methods often focus on interference suppression while neglecting the need for secure state estimation; existing secure state estimation schemes mostly rely on passive, fault-tolerant elastic estimation methods. This not only fails to guarantee data confidentiality, but also allows contaminated data to spread throughout the network via network communication, creating a cascading contamination effect. The separation of state estimation and security protection methods makes it impossible to simultaneously address the ever-increasing demands for both estimation accuracy and security protection.
[0006] In summary, there is an urgent need to design a state estimation scheme suitable for distributed multi-sensor systems that can be robust to non-Gaussian noise and securely transmitted while ensuring data confidentiality and high accuracy of estimation results, thus filling the application gap of existing technologies in scenarios with strong adversarial and high interference. Summary of the Invention
[0007] This invention aims to at least solve one of the technical problems existing in related technologies. To this end, this invention provides a robust state estimation method and system for multi-sensor networks resisting complex interference clusters, thereby ensuring the confidentiality and integrity of data transmission.
[0008] This invention provides a robust state estimation method for multi-sensor networks resisting complex interference, comprising the following steps: S1: Construct a clustered multi-sensor network containing multiple estimators and multiple sensors, dividing the sensors and estimators into multiple clusters, each cluster containing one estimator and multiple sensors, and the estimators are interconnected through a distributed network; S2: Establish an encrypted authentication channel between the sensor and the corresponding cluster estimator, use the AES-CTR encryption algorithm to encrypt the sensor measurement data and generate a GMAC authentication tag, the estimator performs data integrity verification based on the authentication tag, if the integrity verification passes, proceed to step S4, if the integrity verification fails, proceed to step S3; S3: Calculate the sensor attack strength based on the Hamming distance of the authentication tag, classify the sensor into three levels of security (trustworthy, suspicious, and untrustworthy) according to the preset threshold, and then execute step S4. S4: Each estimator updates its local state based on the maximum correlation entropy criterion, using the measurement data of sensors with trusted security levels within the cluster and the sensor attack strength, and obtains the posterior state estimate. S5: Each estimator exchanges posterior state estimates with its neighbor estimators based on the information consensus algorithm, and achieves consistent fusion of state estimates among clusters through iterative communication.
[0009] According to the present invention, a robust state estimation method for a clustered multi-sensor network that resists complex interference is provided. The clustered multi-sensor network is a connected undirected weighted network, and the input data of the estimator and the sensor are position and velocity in the horizontal and vertical directions.
[0010] According to the present invention, a robust state estimation method for a multi-sensor network resistant to complex interference is provided, wherein step S2 includes: S21: Generate a 128-bit key based on the sensor ID and estimator ID and pre-distribute it; S22: Concatenate the sensor ID, timestamp and frame counter to generate a 128-bit counter, and use AES-128 encryption to generate a pseudo-random key stream; S23: The sensor XORs the measurement data with the pseudo-random key stream to obtain ciphertext, and uses the ciphertext, counter, sensor ID and timestamp as input to the GMAC algorithm to generate a 128-bit authentication tag; S24: After receiving the 128-bit authentication tag, the estimator reconstructs the authentication key stream for decryption and recalculates the GMAC tag for integrity verification.
[0011] According to the robust state estimation method for a multi-sensor network resisting complex interference provided by the present invention, step S24 is as follows: S241: The estimator generates an authentication counter based on the counter, sensor ID, and timestamp in the received 128-bit authentication tag; S242: The estimator generates an authentication key stream based on the authentication counter and the pseudo-random key stream; S243: XOR the ciphertext in the received 128-bit authentication tag with the authentication key stream to obtain the original measurement data; S244: Input the raw measurement data, authentication key stream, sensor ID and timestamp into the GMAC algorithm to generate a GMAC tag; S245: Calculate the Hamming distance between the GMAC tag and the 128-bit authentication tag; S246: If the Hamming distance is 0, the integrity verification passes and proceed to step S4; if the Hamming distance is not 0, the integrity verification fails and proceed to step S3.
[0012] According to the robust state estimation method for multi-sensor networks resisting complex interference provided by the present invention, step S3 includes: S31: Calculate average attack strength: in, Let T be the average attack strength of sensor i within frames t to tT, where T is a preset time window. Here, t represents the timestamp ordinal number, and t represents the frame sequence information. Hamming distance for GMAC label and 128-bit authentication label; S32: Calculate cumulative attack strength: in, Let be the cumulative attack strength of sensor i. The average attack strength of sensor i from frame 0 to frame k; S33: Set average attack alert threshold Compared with the cumulative attack alert threshold ,when When the sensor is considered to be a reliable safety grade with low risk, it is deemed to be safe. and When a sensor is deemed to have a suspicious security classification, indicating a potential for hijacking by an attacker, the kernel bandwidth of the subsequent state estimation algorithm is adjusted based on the suspicious classification result; when and At that time, it was considered that the sensor was an untrusted security classification, and the communication channel or the node itself had been hijacked by the attacker.
[0013] According to the robust state estimation method for multi-sensor networks resisting complex interference provided by the present invention, step S4 includes the following steps: S41: Calculate the prior estimate of the current state of the estimator: in, Let r be the prior target state vector of the current state of the estimator. Here is the state transition matrix. Let r be the posterior target state vector of the estimator at the previous time step. Let r be the covariance matrix of the prior state estimate of the current state of estimator r. Let be the covariance matrix of the posterior state estimate of the state of estimator r at the previous time step. for transpose, The covariance of the process noise vector at the previous time step; S42: The posterior state estimate is calculated based on the prior target state vector and the covariance matrix of the prior state estimate. : in, Let r be the posterior target state vector of the current state of the estimator r. Let r be the covariance matrix of the posterior state estimate of the current state of estimator r. For the set of estimator nodes, For sensor measurement matrix transpose, For scalar weights, To measure the covariance of the noise vector, Let be the measurement vector of the i-th sensor.
[0014] According to the robust state estimation method for multi-sensor networks resisting complex interference provided by the present invention, the scalar weights are calculated using the following formula: in, For A Gaussian kernel function with a kernel bandwidth. To modify the kernel bandwidth, Indicates Let be the elliptic norm of the weight matrix. For the initial kernel bandwidth, For modification coefficients, This represents the function that takes the maximum value. The average attack alert threshold, Let be the average attack strength of sensor i within the frame from t to tT.
[0015] According to the robust state estimation method for multi-sensor networks resisting complex interference provided by the present invention, step S5 includes: S51: Define the initial information vector and initial information matrix : in, Let r be the posterior target state vector of the current state of the estimator r. The covariance matrix of the posterior state estimate of the current state of estimator r; S52: The process of consistent information fusion in frame t+1 is represented as follows: in, Let r be the information vector of the (t+1)th frame of the estimator. Let r be the information matrix of the (t+1)th frame of the estimator. Here, represents the consistency weights between estimators r and j, where j is the estimator connected to estimator r. For the set of neighboring estimators of estimator r, Let j be the information vector of the t-th frame of the estimator. Let be the information matrix of the t-th frame of estimator j; S53: When the maximum number of iterations is reached Afterwards, the consensus fusion process is completed, and the estimated consensus fusion state is reconstructed from the information pairs. : in, Let r be the consistent fused state vector of the estimator r at the current time step. The information matrix for the maximum number of iterations. The information vector representing the maximum number of iterations. Let be the consistent fusion state covariance matrix of the current state of estimator r.
[0016] The present invention provides a robust state estimation method for multi-sensor networks resisting complex interference, when and At this time, the estimator will temporarily refuse to use information from the sensors and generate an alarm message to report to the control center for decision-making.
[0017] This invention also provides a robust state estimation system for clustered multi-sensor networks resistant to complex interference, comprising the following modules: Network building module: used to build a clustered multi-sensor network containing multiple estimators and multiple sensors, dividing the sensors and estimators into multiple clusters, each cluster containing one estimator and multiple sensors, and the estimators are interconnected through a distributed network; Data integrity verification module: used to establish an encrypted authentication channel between the sensor and the corresponding cluster estimator. The sensor measurement data is encrypted using the AES-CTR encryption algorithm and a GMAC authentication tag is generated. The estimator performs data integrity verification based on the authentication tag. If the integrity verification passes, the posterior state estimation module is executed. If the integrity verification fails, the security classification judgment module is executed. Security classification judgment module: used to calculate the sensor attack strength based on Hamming distance of authentication tags, classify the sensor into three levels of security: trustworthy, suspicious and untrustworthy according to preset thresholds, and then execute the posterior state estimation module; Posterior state estimation module: This module is used by each estimator to perform local state updates based on the maximum correlation entropy criterion, using measurement data from trusted sensors within the cluster and the intensity of sensor attacks, in order to obtain a posterior state estimate. Consistent fusion state estimation module: This module is used by each estimator to exchange posterior state estimates with its neighbor estimators based on the information consensus algorithm, and achieves consistent fusion state estimation between clusters through iterative communication.
[0018] The above-described one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: This invention addresses the combined interference of non-Gaussian noise at the measurement end and network attacks at the transmission end by employing a layered defense strategy: robust state estimation based on the maximum correlation entropy criterion suppresses non-Gaussian measurement noise, while lightweight encryption authentication using AES-CTR+GMAC defends against network attacks such as spoofed data injection and replay attacks at the transmission end. Compared to the active and passive protection methods commonly used in traditional security state estimation, this invention decouples noise and attacks, reducing the computational burden on the sensor end while ensuring the confidentiality of data transmission, and avoiding false alarms and decreased estimation accuracy caused by their coupling.
[0019] This invention utilizes dynamic seed generation and key isolation technology to achieve identity verification and integrity protection. A security assessment mechanism is constructed to implement sensor security classification, deeply linking the security assessment results with robust state estimation. Based on the attack strength quantification assessment results, the state estimation algorithm parameters are adaptively adjusted, avoiding the influence of contaminated data on the estimation results while maximizing the utilization of effective information, achieving a two-way balance between security protection and estimation accuracy. The rapid response mechanism for cluster alarms effectively reduces the negative impacts of cascading contamination and information leakage caused by single-point damage, facilitating management and decision-making at the control center.
[0020] This invention adopts a hybrid architecture of centralized processing within clusters and distributed fusion between clusters, eliminating the dependence on a central node. Even if some sensor nodes or communication links fail, the remaining network can still complete the state estimation task, significantly improving the system's fault tolerance and robustness. At the same time, it can flexibly adapt to different scales of cluster sensor network deployment scenarios and has scalability.
[0021] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0022] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0023] Figure 1 This is a flowchart illustrating a robust state estimation method for a clustered multi-sensor network that resists complex interference, provided by the present invention.
[0024] Figure 2 This is a schematic diagram of the network topology settings in an embodiment of the present invention.
[0025] Figure 3This is a comparison chart of the estimated average root mean square error values of the positions under four sets of experiments in the embodiments of the present invention.
[0026] Figure 4 This is a comparison chart of the estimated average root mean square error values of velocity under four sets of experiments in the embodiments of the present invention.
[0027] Figure 5 This is a schematic diagram of the structure of a robust state estimation system for a clustered multi-sensor network that resists complex interference, provided by the present invention. Detailed Implementation
[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. The following embodiments are used to illustrate this invention but cannot be used to limit the scope of this invention.
[0029] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0030] The following is combined with Figures 1 to 5 This invention is described.
[0031] Example like Figure 1 As shown, Figure 1 A flowchart illustrating a robust state estimation method for a clustered multi-sensor network to resist complex interference, provided by this invention, includes the following steps: S1: Construct a clustered multi-sensor network containing multiple estimators and multiple sensors. Divide the sensors and estimators into multiple clusters. Each cluster contains one estimator and multiple sensors. The estimators are interconnected through a distributed network. The input data of the estimators and sensors are position and velocity in the horizontal and vertical directions. S2: Establish an encrypted authentication channel between the sensor and the corresponding cluster estimator, use the AES-CTR encryption algorithm to encrypt the sensor measurement data and generate a GMAC authentication tag, the estimator performs data integrity verification based on the authentication tag, if the integrity verification passes, proceed to step S4, if the integrity verification fails, proceed to step S3; S3: Calculate the sensor attack strength based on the Hamming distance of the authentication tag, classify the sensor into three levels of security (trustworthy, suspicious, and untrustworthy) according to the preset threshold, and then execute step S4. S4: Each estimator updates its local state based on the maximum correlation entropy criterion, using the measurement data of sensors with trusted security levels within the cluster and the sensor attack strength, and obtains the posterior state estimate. S5: Each estimator exchanges posterior state estimates with its neighbor estimators based on the information consensus algorithm, and achieves consistent fusion of state estimates among clusters through iterative communication.
[0032] Specifically, step S1 involves establishing a hybrid network topology, where multiple resource-constrained sensor nodes belong to a resource-rich estimator, forming an edge-cluster network. These estimators are then connected via communication links to form a distributed network, enabling data transmission and sharing.
[0033] Construct a clustered multi-sensor network system that includes estimators and edge sensors. , ,in Represents the set of estimators. This represents a set of sensor nodes. Since sensors are geographically or logically distributed, the sensors and estimators in the network are divided into multiple clusters. Each cluster contains one estimator and one estimator. There are *r* sensors, where the set of sensor nodes in the *r*th cluster is defined as... , Within each cluster, sensors collect information, and the estimator, acting as the cluster head, connects and processes all measurement data within the cluster, calculates the local state estimate, and facilitates information transfer and sharing between clusters through communication with other estimators. It also has the capability to communicate with the upper-level control center. Within a cluster, sensors and estimators are directly connected; between clusters, estimators form a distributed network through communication links. An undirected graph is defined. , ,in Let represent the set of edges in the estimator network. If estimators r and j are connected, then: .definition Let represent the set of neighboring estimators directly connected to estimator r, i.e.: The network involved in this invention is a connected undirected weighted network, and a consensus weight matrix is defined. , M is the total number of estimators, and the consistency weights satisfy... : , .
[0034] This invention considers the state estimation problem of dynamic target tracking and models it as a state-space model as follows: in, Let be the target state vector. Let be the target state vector for the next time step. Let be the measurement vector of the i-th sensor. Here is the state transition matrix. This is the sensor measurement matrix. This is the process noise vector. The mean is 0 and the covariance is Gaussian white noise; For process measurement noise vector, The mean is 0, and the covariance is It is affected by external disturbances during measurement.
[0035] To address the secure transmission challenge under resource-constrained sensor node conditions, this invention employs a lightweight encryption and authentication method combining AES-CTR and GMAC authentication. Tags are generated concurrently with encryption, and attack detection during transmission is achieved by comparing tag differences. This completes the design of encryption / decryption and authentication functions between the sensor and estimator. Based on the differences in authentication results, a sensor security grading standard is defined, providing quantifiable security data to enhance the cluster network system's ability to measure attack strength, continuously monitor, and identify attacks, thus supporting system security decisions.
[0036] Specifically, step S2 includes: S21: Generate a 128-bit key based on the sensor ID and estimator ID and pre-distribute it; S22: Concatenate the sensor ID, timestamp and frame counter to generate a 128-bit counter, and use AES-128 encryption to generate a pseudo-random key stream; S23: The sensor XORs the measurement data with the pseudo-random key stream to obtain ciphertext, and uses the ciphertext, counter, sensor ID and timestamp as input to the GMAC algorithm to generate a 128-bit authentication tag; S24: After receiving the 128-bit authentication tag, the estimator reconstructs the authentication key stream, decrypts it, and recalculates the GMAC tag for integrity verification. Specifically: S241: The estimator generates an authentication counter based on the counter, sensor ID, and timestamp in the received 128-bit authentication tag; S242: The estimator generates an authentication key stream based on the authentication counter and the pseudo-random key stream; S243: XOR the ciphertext in the received 128-bit authentication tag with the authentication key stream to obtain the original measurement data; S244: Input the raw measurement data, authentication key stream, sensor ID and timestamp into the GMAC algorithm to generate a GMAC tag; S245: Calculate the Hamming distance between the GMAC tag and the 128-bit authentication tag; S246: If the Hamming distance is 0, the integrity verification passes and proceed to step S4; if the Hamming distance is not 0, the integrity verification fails and proceed to step S3.
[0037] Specifically, step S3 includes: S31: Calculate average attack strength: in, Let T be the average attack strength of sensor i within frames t to tT, where T is a preset time window. Here, t represents the timestamp ordinal number, and t represents the frame sequence information. Hamming distance for GMAC label and 128-bit authentication label; S32: Calculate cumulative attack strength: in, Let be the cumulative attack strength of sensor i. The average attack strength of sensor i from frame 0 to frame k; S33: Set average attack alert threshold Compared with the cumulative attack alert threshold ,when When the sensor is considered to be a reliable safety grade with low risk, it is deemed to be safe. and When a sensor is deemed to have a suspicious security classification, indicating a potential for hijacking by an attacker, the kernel bandwidth of the subsequent state estimation algorithm is adjusted based on the suspicious classification result; when and At that time, it was considered that the sensor was an untrusted security classification, and the communication channel or the node itself had been hijacked by the attacker.
[0038] In particular, when and At this time, the estimator will temporarily refuse to use information from the sensors and generate an alarm message to report to the control center for decision-making.
[0039] In this invention, the estimator is responsible for completing the entire state estimation and consensus fusion process. The estimator predicts the state based on the state information from the previous time step. After acquiring reliable measurement data, it designs a state estimation method based on the maximum correlation entropy criterion to complete local state updates. Based on this, it exchanges state information with neighboring estimators using an information consensus algorithm, achieving information fusion and ultimately achieving consensus on the state estimation across the entire network.
[0040] Step S4 includes the following steps: S41: Calculate the prior estimate of the current state of the estimator: in, Let r be the prior target state vector of the current state of the estimator. Let r be the posterior target state vector of the estimator at the previous time step. Let r be the covariance matrix of the prior state estimate of the current state of estimator r. Let be the covariance matrix of the posterior state estimate of the state of estimator r at the previous time step. for transpose, The covariance of the process noise vector at the previous time step; S42: The posterior state estimate is calculated based on the prior target state vector and the covariance matrix of the prior state estimate. : in, Let r be the posterior target state vector of the current state of the estimator r. Let r be the covariance matrix of the posterior state estimate of the current state of estimator r. For the set of estimator nodes, for transpose, Scalar weights.
[0041] The formula for calculating the scalar weight is: in, For A Gaussian kernel function with a kernel bandwidth. To modify the kernel bandwidth, Indicates Let be the elliptic norm of the weight matrix. For the initial kernel bandwidth, For modification coefficients, This represents the function that takes the maximum value.
[0042] In this invention, after obtaining the posterior state estimate, each estimator will communicate in a distributed manner through an inter-cluster network. Since the estimators are typically deployed on edge servers or gateway devices with strong computing power and guaranteed power supply, their computing resources are sufficient to support standard transport layer security protocols. Therefore, the inter-cluster network is considered a "trusted channel," ensuring the confidentiality and integrity of inter-cluster data transmission. To achieve consistency in state estimates within the network, iterative communication between clusters is required. Specifically, step S5 includes: S51: Define the initial information vector and initial information matrix : ; S52: The process of consistent information fusion in frame t+1 is represented as follows: in, Let r be the information vector of the (t+1)th frame of the estimator. Let r be the information matrix of the (t+1)th frame of the estimator. Here, represents the consistency weights between estimators r and j, where j is the estimator connected to estimator r. Let j be the information vector of the t-th frame of the estimator. Let be the information matrix of the t-th frame of estimator j; S53: When the maximum number of iterations is reached Afterwards, the consensus fusion process is completed, and the estimated consensus fusion state is reconstructed from the information pairs. : in, Let r be the consistent fused state vector of the estimator r at the current time step. The information matrix for the maximum number of iterations. The information vector representing the maximum number of iterations. Let be the consistent fusion state covariance matrix of the current state of estimator r.
[0043] The consistent fusion state estimation serves as the next step in the network's new input control system. It eliminates local observation biases, enabling distributed collaboration to improve robustness and fault tolerance. The physical essence of consistent fusion state estimation is to ensure that all nodes in the distributed network obtain identical, globally optimal results for the same physical objective.
[0044] To verify the effectiveness and superiority of the proposed method in resisting non-Gaussian noise, resisting network attacks, and dealing with complex interference, this embodiment constructs a target tracking scenario on the MATLAB R2023b simulation platform. A comparative experiment is conducted between a traditional security estimation method (a security filter based on a chi-square distribution detector) and a traditional robust estimation method (a filter based on a Huber function) to systematically verify the state estimation performance under complex interference. The core objective is to fully highlight the robustness, accuracy, and real-time advantages of the proposed clustered multi-sensor network state estimation method under different interference combinations through comparative experiments; to quantitatively verify the technical superiority of the proposed method compared to traditional methods in dealing with non-Gaussian measurement noise and mixed network attacks; to solve the technical bottlenecks of traditional methods such as poor adaptability, easy divergence, and computational complexity under complex interference; and to provide experimental support and engineering reference for reliable state estimation of clustered multi-sensor networks.
[0045] Network topology setup: Construct a cluster network containing 4 estimators and 16 edge wireless sensors, divided into 4 clusters based on geographical location. Each cluster contains 1 estimator and 4 sensors. All sensor and estimator nodes are deployed in a fixed manner according to a preset topology. The network topology setup is as follows: Figure 2 As shown.
[0046] The estimator nodes form an undirected graph, and Metropolis weights are used for communication to ensure the convergence of the estimation. Considering the resource constraints of sensor nodes in practical applications, a low-power wireless protocol is used for intra-cluster communication, and a high-reliability wireless protocol is used for inter-cluster communication.
[0047] In this embodiment, the tracked target is set to move at approximately uniform linear velocity within a two-dimensional plane. The state vector contains the target's position and velocity in the horizontal and vertical directions. The sensor nodes deployed in the network include two types: one type directly measures the target's position information, and the other type measures the distance and azimuth between the target and the sensor node. In each cluster of edge networks, there are two sensor nodes of each type.
[0048] This embodiment specifically sets parameters for non-Gaussian measurement noise and network attacks to ensure that the parameters are accurately adapted to the experimental scenario. The core parameter settings are as follows: 1. Non-Gaussian measurement noise. A Gaussian mixture distribution is used for modeling to simulate the irregular noise distribution at the sensor measurement end.
[0049] 2. Network Attacks. The most typical spoofing attack in target tracking applications is selected, where attackers directly add fake data to encrypted data packets. The attack vector applied to both types of sensors is set as a random variable.
[0050] In this embodiment, the average root mean square error of position (ARMSE) and the average root mean square error of velocity (ARMSE) in the encryption and decryption process are selected as evaluation indicators to verify the effectiveness of the proposed scheme. The average response time of the encryption and decryption process is the average of the time required for each sensor node in the entire network to complete the data encryption process and the corresponding decryption process. The average root mean square error measures the degree to which the state estimate deviates from the true state of the target. The smaller the value, the more accurate the estimation result.
[0051] This embodiment designs four different experimental scenarios: (1) no attack and no noise; (2) attack and no noise; (3) no attack and noise; (4) attack and noise. A comparison of the estimated root mean square error of position and root mean square error of velocity under the four experimental scenarios is shown in the figure below. Figure 3 and Figure 4 As shown. By Figure 3 and Figure 4 It is known that traditional security estimation algorithms perform well in the absence of noise, but their performance degrades under non-Gaussian noise interference; traditional robust estimation algorithms perform well in the absence of network attacks, but their performance degrades under the influence of network attacks. The algorithm proposed in this invention can guarantee good estimation accuracy under both single and combined interference conditions, can resist combined interference, and achieve high-precision state estimation.
[0052] like Figure 5 As shown, this embodiment of the invention also provides a robust state estimation system for multi-sensor networks resisting complex interference, comprising the following modules: Network building module: used to build a clustered multi-sensor network containing multiple estimators and multiple sensors, dividing the sensors and estimators into multiple clusters, each cluster containing one estimator and multiple sensors, and the estimators are interconnected through a distributed network; Data integrity verification module: used to establish an encrypted authentication channel between the sensor and the corresponding cluster estimator. The sensor measurement data is encrypted using the AES-CTR encryption algorithm and a GMAC authentication tag is generated. The estimator performs data integrity verification based on the authentication tag. If the integrity verification passes, the posterior state estimation module is executed. If the integrity verification fails, the security classification judgment module is executed. Security classification judgment module: used to calculate the sensor attack strength based on Hamming distance of authentication tags, classify the sensor into three levels of security: trustworthy, suspicious and untrustworthy according to preset thresholds, and then execute the posterior state estimation module; Posterior state estimation module: This module is used by each estimator to perform local state updates based on the maximum correlation entropy criterion, using measurement data from trusted sensors within the cluster and the intensity of sensor attacks, in order to obtain a posterior state estimate. Consistent fusion state estimation module: This module is used by each estimator to exchange posterior state estimates with its neighbor estimators based on the information consensus algorithm, and achieves consistent fusion state estimation between clusters through iterative communication.
[0053] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
[0055] It should be noted that the embodiments of this disclosure can be implemented using hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a programmable memory or a data carrier such as an optical or electronic signal carrier.
[0056] While this disclosure has been described with reference to several specific embodiments, it should be understood that this disclosure is not limited to the specific embodiments disclosed. This disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.
Claims
1. A robust state estimation method for clustered multi-sensor networks resistant to complex interference, characterized in that, Includes the following steps: S1: Construct a clustered multi-sensor network containing multiple estimators and multiple sensors, dividing the sensors and estimators into multiple clusters, each cluster containing one estimator and multiple sensors, and the estimators are interconnected through a distributed network; S2: Establish an encrypted authentication channel between the sensor and the corresponding cluster estimator, use the AES-CTR encryption algorithm to encrypt the sensor measurement data and generate a GMAC authentication tag, the estimator performs data integrity verification based on the authentication tag, if the integrity verification passes, proceed to step S4, if the integrity verification fails, proceed to step S3; S3: Calculate the sensor attack strength based on the Hamming distance of the authentication tag, classify the sensor into three levels of security (trustworthy, suspicious, and untrustworthy) according to the preset threshold, and then execute step S4. S4: Each estimator updates its local state based on the maximum correlation entropy criterion, using the measurement data of sensors with trusted security levels within the cluster and the sensor attack strength, and obtains the posterior state estimate. S5: Each estimator exchanges posterior state estimates with its neighbor estimators based on the information consensus algorithm, and achieves consistent fusion of state estimates among clusters through iterative communication.
2. The robust state estimation method for a clustered multi-sensor network to resist complex interference as described in claim 1, characterized in that, The clustered multi-sensor network is a connected undirected weighted network, and the input data of the estimator and the sensors are position and velocity in the horizontal and vertical directions.
3. The robust state estimation method for a clustered multi-sensor network to resist complex interference as described in claim 1, characterized in that, Step S2 includes: S21: Generate a 128-bit key based on the sensor ID and estimator ID and pre-distribute it; S22: Concatenate the sensor ID, timestamp and frame counter to generate a 128-bit counter, and use AES-128 encryption to generate a pseudo-random key stream; S23: The sensor XORs the measurement data with the pseudo-random key stream to obtain ciphertext, and uses the ciphertext, counter, sensor ID and timestamp as input to the GMAC algorithm to generate a 128-bit authentication tag; S24: After receiving the 128-bit authentication tag, the estimator reconstructs the authentication key stream for decryption and recalculates the GMAC tag for integrity verification.
4. The robust state estimation method for a clustered multi-sensor network to resist complex interference as described in claim 3, characterized in that, Step S24 is as follows: S241: The estimator generates an authentication counter based on the counter, sensor ID, and timestamp in the received 128-bit authentication tag; S242: The estimator generates an authentication key stream based on the authentication counter and the pseudo-random key stream; S243: XOR the ciphertext in the received 128-bit authentication tag with the authentication key stream to obtain the original measurement data; S244: Input the raw measurement data, authentication key stream, sensor ID and timestamp into the GMAC algorithm to generate a GMAC tag; S245: Calculate the Hamming distance between the GMAC tag and the 128-bit authentication tag; S246: If the Hamming distance is 0, the integrity verification passes and proceed to step S4; if the Hamming distance is not 0, the integrity verification fails and proceed to step S3.
5. The robust state estimation method for a clustered multi-sensor network resisting complex interference as described in claim 4, characterized in that, Step S3 includes: S31: Calculate average attack strength: in, Let T be the average attack strength of sensor i within frames t to tT, where T is a preset time window. Here, t represents the timestamp ordinal number, and t represents the frame sequence information. Hamming distance for GMAC label and 128-bit authentication label; S32: Calculate cumulative attack strength: in, Let be the cumulative attack strength of sensor i. The average attack strength of sensor i from frame 0 to frame k; S33: Set average attack alert threshold Compared with the cumulative attack alert threshold ,when When the sensor is considered to be a reliable safety grade with low risk, it is deemed to be safe. and When a sensor is deemed to have a suspicious security classification, indicating a potential for hijacking by an attacker, the kernel bandwidth of the subsequent state estimation algorithm is adjusted based on the suspicious classification result; when and At that time, it was considered that the sensor was an untrusted security classification, and the communication channel or the node itself had been hijacked by the attacker.
6. The robust state estimation method for a clustered multi-sensor network to resist complex interference as described in claim 1, characterized in that, Step S4 includes the following steps: S41: Calculate the prior estimate of the current state of the estimator: in, Let r be the prior target state vector of the current state of the estimator. Here is the state transition matrix. Let r be the posterior target state vector of the estimator at the previous time step. Let r be the covariance matrix of the prior state estimate of the current state of estimator r. Let be the covariance matrix of the posterior state estimate of the state of estimator r at the previous time step. for transpose, The covariance of the process noise vector at the previous time step; S42: The posterior state estimate is calculated based on the prior target state vector and the covariance matrix of the prior state estimate. : in, Let r be the posterior target state vector of the current state of the estimator r. Let r be the covariance matrix of the posterior state estimate of the current state of estimator r. For the set of estimator nodes, For sensor measurement matrix transpose, For scalar weights, To measure the covariance of the noise vector, Let be the measurement vector of the i-th sensor.
7. A robust state estimation method for a clustered multi-sensor network to resist complex interference as described in claim 6, characterized in that, The formula for calculating the scalar weight is: in, For A Gaussian kernel function with a kernel bandwidth. To modify the kernel bandwidth, Indicated by Let be the elliptic norm of the weight matrix. For the initial kernel bandwidth, For modification coefficients, This represents the function that takes the maximum value. The average attack alert threshold, Let be the average attack strength of sensor i within the frame from t to tT.
8. The robust state estimation method for a clustered multi-sensor network to resist complex interference as described in claim 1, characterized in that, Step S5 includes: S51: Define the initial information vector and initial information matrix : in, Let r be the posterior target state vector of the current state of the estimator r. The covariance matrix of the posterior state estimate of the current state of estimator r; S52: The process of consistent information fusion in frame t+1 is represented as follows: in, Let r be the information vector of the (t+1)th frame of the estimator. Let r be the information matrix of the (t+1)th frame of the estimator. Here, represents the consistency weights between estimators r and j, where j is the estimator connected to estimator r. For the set of neighboring estimators of estimator r, Let j be the information vector of the t-th frame of the estimator. Let be the information matrix of the t-th frame of estimator j; S53: When the maximum number of iterations is reached Afterwards, the consensus fusion process is completed, and the estimated consensus fusion state is reconstructed from the information pairs. : in, Let r be the consistent fused state vector of the estimator r at the current time step. The information matrix for the maximum number of iterations. The information vector representing the maximum number of iterations. Let be the consistent fusion state covariance matrix of the current state of estimator r.
9. A robust state estimation method for a clustered multi-sensor network to resist complex interference as described in claim 5, characterized in that, when and At this time, the estimator will temporarily refuse to use information from the sensors and generate an alarm message to report to the control center for decision-making.
10. A robust state estimation system for a clustered multi-sensor network resistant to complex interference, characterized in that, Includes the following modules: Network building module: used to build a clustered multi-sensor network containing multiple estimators and multiple sensors, dividing the sensors and estimators into multiple clusters, each cluster containing one estimator and multiple sensors, and the estimators are interconnected through a distributed network; Data integrity verification module: used to establish an encrypted authentication channel between the sensor and the corresponding cluster estimator. The sensor measurement data is encrypted using the AES-CTR encryption algorithm and a GMAC authentication tag is generated. The estimator performs data integrity verification based on the authentication tag. If the integrity verification passes, the posterior state estimation module is executed. If the integrity verification fails, the security classification judgment module is executed. Security classification judgment module: used to calculate the sensor attack strength based on Hamming distance of authentication tags, classify the sensor into three levels of security: trustworthy, suspicious and untrustworthy according to preset thresholds, and then execute the posterior state estimation module; Posterior state estimation module: This module is used by each estimator to perform local state updates based on the maximum correlation entropy criterion, using measurement data from trusted sensors within the cluster and the intensity of sensor attacks, in order to obtain a posterior state estimate. Consistent fusion state estimation module: This module is used by each estimator to exchange posterior state estimates with its neighbor estimators based on the information consensus algorithm, and achieves consistent fusion state estimation between clusters through iterative communication.