A High-Energy-Efficient Space-Time Correlation Distributed Sensing Node Security Anti-Attack Verification Method

Abstract

The present invention discloses a high-energy-efficiency spatio-temporal correlation distributed sensing node security anti-attack verification method, aiming at enhancing the low power consumption and high efficiency security of a large number of distributed sensing nodes. It includes units such as low-power data transmission, spatial neighborhood information fusion for sensing data correction, adaptive sensing information trend prediction for generating time series security copies, micro-power adaptive data acquisition, adaptive difference-triggered anomaly detection, adaptive data integrity verification and recovery, etc. When the system runs, the gateway forms a time series-neighborhood fusion security copy according to the change rules of multi-node weighted data, and based on this, determines whether the current sensing data has been tampered with and corrects the wrong data. At the same time, based on the characteristics of the change trend of the node sensing data contained in the security copy, the gateway optimizes the node working mode in real time, further reducing power consumption while improving data security, prolonging the battery life, and promoting large-scale applications.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things node control, and particularly to an energy-efficient spatio-temporal correlation distributed sensing node security anti-attack verification method. Background Art

[0002] The Internet of Things (IoT) system relies on a large number of distributed sensing nodes to achieve extensive perception and data collection in the physical world, supporting applications in multiple fields such as smart cities, smart homes, and industrial monitoring. However, with the rapid development of IoT technology and the continuous expansion of the deployment scale, sensing nodes face two key challenges in practical applications: low-power optimization and data security guarantee. The effective solution of these problems is related to the stability, operation efficiency of the IoT system and the privacy protection of user data.

[0003] Specifically, firstly, low-power optimization is the core requirement for the efficient operation of IoT sensing nodes. Since IoT nodes usually rely on battery power supply, extending the battery life directly affects the long-term reliability of the system and the operation and maintenance costs. In a widely distributed node network, low-power optimization can help sensing nodes significantly reduce energy consumption and reduce the battery replacement frequency, thereby reducing the operation and maintenance burden and subsequent costs in remote or difficult-to-maintain environments. However, the resources of sensing nodes are limited, and how to achieve low power while performing multiple tasks has become a technical difficulty.

[0004] Secondly, IoT sensing nodes will acquire and transmit a large amount of environmentally sensitive data during long-term operation, such as indoor temperature in smart homes, traffic conditions in smart cities, etc. This information contains user privacy and important environmental states. Since there are potential security risks in all aspects of data collection, transmission and processing of sensing nodes, malicious attackers may steal, tamper with or forge sensing data, resulting in information leakage or system misjudgment, and thus threatening user privacy and system stability. Therefore, data security guarantee is crucial for the normal operation of the IoT system.

[0005] Currently, the application of common security encryption technologies on sensing nodes faces many limitations. Traditional encryption technologies usually require complex computing and storage resources. Although they can improve data security, these technologies are usually executed as an additional task on nodes, inevitably increasing node power consumption and thus affecting the low-power optimization effect. On the one hand, the complexity of encryption calculation leads to an increase in node power consumption, which is contrary to the goal of low-power optimization; on the other hand, the computing power and storage resources of sensing nodes are limited, and too many encryption tasks may cause the node load to be too high, affecting the response speed to actual environmental data. How to minimize power consumption while ensuring data security is the key problem of current technology.

[0006] To this end, this patent proposes an energy-efficient spatio-temporal correlation distributed sensing node security anti-attack verification method, aiming to solve the contradiction between low power consumption and high security. This method ensures the authenticity and integrity of data by establishing a time series-neighborhood fusion security copy on the data aggregation gateway. Specifically, the gateway performs weighted processing based on the data change situations of multiple sensing nodes to form a security copy that integrates time series features and neighborhood distribution features. When the sensing data is maliciously tampered with, the security copy can assist the gateway in identifying abnormal data and correcting errors, enabling the Internet of Things system to monitor data security in real time and reducing the impact of malicious attacks.

[0007] In addition, this method makes full use of the time series and spatial distribution information in the security copy, which can not only enhance data security but also optimize the working mode of sensing nodes by analyzing the data change trends of nodes. When the environmental change is not obvious, the gateway can reduce the node working frequency according to the copy information, thereby reducing energy consumption; while when the environment fluctuates violently, the node can increase the monitoring frequency to ensure data real-time. Through this adaptive dynamic adjustment mechanism, the system significantly reduces node power consumption and extends battery life while ensuring data security, meeting the requirements of large-scale Internet of Things application scenarios. Summary of the Invention

[0008] The object of the present invention is to provide an energy-efficient spatio-temporal correlation distributed sensing node security anti-attack verification method for enhancing low power consumption, high efficiency and security of a large number of distributed sensing nodes. With the rapid development of Internet of Things technology, multiple sensing nodes are deployed more widely, forming a distributed network system composed of numerous sensor nodes. However, traditional sensors face two problems in real-time data acquisition and transmission: excessive energy consumption and data security risks. These problems not only affect the continuous operation time of the device but also may cause data tampering or loss during transmission, thus seriously affecting the reliability of the system. Therefore, there is an urgent need for a new method to reduce the energy consumption of the system and improve the overall performance of the sensor network while ensuring data security.

[0009] To solve the above technical problems, the technical solution adopted by the present invention is an energy-efficient spatio-temporal correlation distributed sensing node security anti-attack verification method, which specifically includes the following steps:

[0010] Real-time collect real-time environmental data in multiple dimensions through a multi-sensor data acquisition unit. The collected data not only includes basic environmental information such as temperature, humidity and air pressure, but may also cover parameters such as light intensity, noise level and other parameters that may affect the environmental state. By integrating different types of sensors, this unit can collect multi-dimensional environmental information at the same moment, laying a foundation for subsequent data processing and analysis. The effectiveness of this process will directly affect the accuracy and reliability of the subsequent prediction model.

[0011] After data collection, based on the memory registration technology RDMA acceleration system, the RDMA protocol is offloaded to the DPU hardware using memory registration technology, allowing the network card and the node to directly read and write data, eliminating the context switching and data copy overhead of the traditional TCP / IP protocol stack. In addition, the low-power data transmission unit supports congestion control for users based on software programming interfaces through programmable network lossless congestion control algorithms, compatible with common algorithms such as DCQCN, NDP, HPCC, and LDCP, enabling the system to dynamically adjust according to feedback signals such as RTT, queue depth, packet loss rate, and ECN during transmission, achieving lossless congestion control.

[0012] After the gateway receives the data, to avoid the problem that the security copy generation of the current node deviates due to being attacked, first, according to the inherent change law of the physical characteristics sensed by each sensor node, the gateway uses the spatial neighborhood information to determine whether the current sensor node is abnormal. The gateway receives and analyzes the data from different sensor nodes in real time, determines their weights by calculating the spatial correlation of the data of each node. When the data of a certain sensor node is abnormal, the data of other sensor nodes adjacent to this node are used, and the weight is calculated according to the weighted algorithm to correct the data of the current node. This process not only considers the spatial distribution characteristics of the data but also makes full use of the neighborhood information, thus realizing the accurate identification and effective correction of abnormal data.

[0013] Node weighting coefficient formula:

[0014]

[0015] Among them, α k represents the weight coefficient of each factor, satisfying

[0016]

[0017] Among them, w ij is the weighting coefficient between the sensor node i and the neighbor node j, is the distance attenuation rate, d ij is the distance, μ is the height difference attenuation rate, h ij is the height difference, θ ij is the wind direction angle between nodes, o ij is the terrain obstruction, γ is the power attenuation rate, p i is the node power, ε is the time attenuation rate, t ij is the time difference.

[0018] After determining whether the current node is abnormal and handling it, the gateway uses a one-dimensional neural network (1D-CNN) and a gated recurrent unit (GRU) to analyze the time series data. By combining the feature extraction ability of 1D-CNN and the time series data processing ability of GRU, it can accurately capture the inherent change characteristics of the physical features sensed by the sensor nodes, thereby effectively predicting the future data change trend and generating a time series-neighborhood fusion security copy.

[0019] Furthermore, after obtaining the security copy, the method dynamically adjusts the sampling frequency of the sensor nodes based on the weighted comprehensive analysis of the first-order, second-order, and third-order derivative changes of the predicted spatio-temporal sequence copy.

[0020] The first derivative function of the copy represents the data change speed and reflects the current data change trend; the second derivative function represents the data change acceleration trend and reflects the degree of change intensity; the third derivative function represents the change of the data change acceleration, that is, the mutation of the change intensity. If the first derivative is large, the data changes significantly at the current time point; if the second derivative is large, it indicates that the degree of data change intensifies and a higher sampling frequency is required to capture subtle changes; if the third derivative is large, it indicates that the acceleration of the data fluctuates violently, which may indicate the occurrence of certain emergencies or extreme conditions, and the system needs to respond quickly to capture dynamic changes.

[0021] Assume that the predicted time series data is x(t), and its values at discrete time points t = t 1 , t 2 ,..., t n are x(t i ).

[0022] Derivation of the first derivative of the predicted time series:

[0023]

[0024] If the t i interval is a fixed step size Δt, it is simplified to:

[0025]

[0026] Derivation of the second derivative of the predicted time series:

[0027]

[0028] Derivation of the third derivative of the predicted time series:

[0029]

[0030] Weighted formula for dynamic adjustment of the sampling rate:

[0031] S(t) = w 1 ·|v(t)|p1 +w 2 ·|a(t)| p2 +w 3 ·|j(t)| p3

[0032] Among them, S(t) is a comprehensive evaluation index, v(t), a(t), j(t) are the first-order, second-order, and third-order derivatives of the time series security copy, respectively, and w 1 to w 3 is the weight, which can be optimized according to the experiment and system requirements. p1, p2, and p3 are the powers of the corresponding items, which control the influence of different derivative items on the whole to optimize the overall sampling rate adjustment effect. When the derivative value is small, its weight in the comprehensive evaluation index can be further reduced by appropriately selecting the power number. (Specifically, this patent considers the first three derivative factors that have the greatest impact on its change)

[0033] Sampling frequency setting formula:

[0034]

[0035] Wake-up interval formula:

[0036]

[0037] Where, f(t) is the current sampling frequency, f max is the maximum sampling frequency of the sensor, to prevent the system from errors caused by excessively high sampling frequency; f min is the minimum sampling frequency of the sensor to prevent missing important information due to a large sampling interval caused by a low frequency; T 1 and T 2 are two thresholds of the comprehensive evaluation index S(t), which are used to define the range of frequency adjustment; It is a smooth adjustment parameter used to control the frequency growth rate. The larger the value, the slower the growth.

[0038] When S(t)≤T 1 When the data changes little, the sampling frequency is kept at the minimum value f min When T 1 <S(t)<T 2 When the sampling frequency is f min and f max Smoothly changing between It can be ensured that the frequency gradually increases with the increase of S(t), and the growth rate is determined by Control, larger To make the frequency increase smoother; when S(t)>T 1 When the data changes drastically or abnormally, the sampling frequency reaches f maxand will not continue to increase, thus avoiding the problem of infinite frequency increase.

[0039] Specifically, the adaptive difference-triggered anomaly detection unit always maintains a close monitoring of the difference between the real-time collected data and the predicted data. We innovatively incorporate an adaptive sliding window threshold mechanism into this unit to dynamically match the unique characteristics and data fluctuation patterns of different sensors. When the difference between the monitored actual data and the predicted data exceeds the threshold dynamically set by the adaptive sliding window, this unit will quickly trigger the anomaly anti-tampering security verification mechanism. Through precise algorithms and efficient logical judgments, this mechanism can immediately identify abnormal fluctuations or potential tampering behaviors in the data and immediately activate the adaptive data integrity guarantee unit.

[0040] The adaptive data integrity guarantee unit includes a set of precise anomaly handling mechanisms. This mechanism first carefully analyzes the type and severity of the anomaly to determine whether the anomaly stems from sensor failures, data transmission errors, data processing logic defects, or other potential factors. Based on the analysis results, the unit selects and executes appropriate handling strategies, including directly discarding the abnormal data, triggering the sensor to re-collect, or using mathematical methods for data repair. At the same time, the unit exhaustively records all the process information of anomaly handling, covering key details such as the anomaly type, the selected handling strategy, and the data status after processing, providing a reliable basis for subsequent in-depth data analysis and system performance optimization.

[0041] The beneficial effects of the present invention are as follows. Through the above-mentioned high-energy-efficiency spatio-temporal correlation distributed sensor node security anti-attack verification method, it is possible to significantly reduce the energy consumption of sensor nodes while ensuring security and extend the service life of the device. At the same time, the model combines a one-dimensional convolutional neural network (1D-CNN) and a gated recurrent unit (GRU) to effectively extract features in multi-sensor spatio-temporal sequence data and improve the accuracy of anomaly detection. By introducing an evaluation mechanism based on the weighted synthesis of multi-order derivatives of the predicted time series copy, the system is more flexible and robust in dealing with dynamic environmental changes and can accurately judge when to wake up the sensor for data collection. By dynamically adjusting the working state of the sensor, the system achieves an ideal balance between power consumption and security, ensuring efficient and secure operation in various Internet of Things application scenarios. This series of innovative designs not only overcomes the bottlenecks of the existing technology but also provides new ideas and technical directions for the sustainable development of future intelligent Internet of Things. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is the overall architecture diagram of the high-energy-efficiency spatio-temporal correlation distributed sensor node security anti-attack verification method in the present invention;

[0043] Figure 2 is the detailed flowchart of the high-energy-efficiency spatio-temporal correlation distributed sensor node security anti-attack verification method in the present invention;

[0044] Figure 3 This is a schematic diagram of the process of the spatial neighborhood information fusion sensing data correction unit in the present invention;

[0045] Figure 4 This is a schematic diagram of the process of the adaptive sensing information trend prediction time series secure copy generation unit in the present invention. Specific embodiments

[0046] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0047] The high-energy-efficiency spatio-temporal correlation distributed sensing node security anti-attack verification method proposed by the present invention has a system block diagram as shown in Figure 2 , wherein, the method includes the following steps: The gateway first obtains real-time data from multiple sensors, and performs noise filtering and data smoothing processing on the collected data. The neighborhood information weighting mechanism is used to determine whether the current data is abnormal and process it. The processed data is predicted by the system, and the severity of environmental changes is judged based on the predicted time series copy, and the sampling rate of the sensing node is dynamically adjusted. Then the system performs anomaly detection, monitors the difference between the actual data and the predicted time series secure copy data in real time, and triggers an alarm and takes corresponding measures when the difference exceeds the set threshold of the sliding window. The system ensures data integrity through a multi-level adaptive security verification mechanism and automatically starts anti-tampering measures to cope with potential anomalies or tampering risks.

[0048] The data acquisition unit is responsible for obtaining real-time data from multiple sensors (such as temperature sensors, humidity sensors, and pressure sensors, etc.). This unit adopts a low-power design concept, optimizes the working state of the sensors to reduce battery consumption, thereby extending the service life of the device. The data acquisition unit performs preliminary processing on the collected raw data, including noise filtering and data smoothing, to generate time series data. These processed data are then sent to the low-power data transmission unit for further data processing and transmission.

[0049] The low-power transmission unit accelerates the system using the memory registration technology RDMA. By using the memory registration technology to offload the RDMA protocol to the DPU hardware, direct data reading and writing between the network card and the application are realized, so as to achieve end-to-end microsecond-level low-latency transmission, and at the same time reduce the computing overhead of the CPU.

[0050] The structural block diagram of the spatial neighborhood information fusion sensing data correction unit is shown in Figure 3, by utilizing the spatial conduction effect, combining the data of neighboring sensing nodes, determining the weights of sensing nodes through a weighted processing formula, and using neighborhood information to ensure the consistency of sensing data at the spatial level. In actual operation, the spatial neighborhood information fusion sensing data correction unit first constructs a spatial data network that covers all relevant sensing nodes and their data. Subsequently, the unit applies the weighted algorithm described in the invention content section. This algorithm assigns a weight value to the data of each node based on the spatial relationship between sensing nodes and the mutual dependence of data. When the system detects that the data of a certain sensing node may be abnormal, the spatial neighborhood information fusion sensing data correction unit will intervene immediately. The unit will refer to the data of other sensing nodes adjacent to this node and the weight value obtained through the weighted algorithm to deeply analyze and accurately correct the data of the current node. This correction process not only ensures the accuracy of the data but also effectively identifies and repairs potential abnormal data.

[0051] The structural block diagram of the adaptive sensing information trend prediction time series security copy generation unit is shown in Figure 4 , this unit combines a one-dimensional convolutional neural network (1D-CNN) and a gated recurrent unit (GRU) to learn and analyze the historical data of sensing nodes. The unit inputs the real-time sensor data after preprocessing into the 1D-CNN, and uses the convolutional layer to extract local features and capture short-term patterns. The extracted features are input into the GRU, and the GRU captures the long-term dependencies in the data through its gating mechanism, thereby improving the prediction accuracy. Finally, the unit generates an adaptive time series security copy, which is used for predicting future data trends and serves as the basis for dynamically adjusting the sampling rate, enhancing the reliability and security of the system.

[0052] After predicting the future data change trend, the gateway conducts a weighted comprehensive evaluation based on the multi-order derivative changes (the first, second, and third order derivatives with greater influence on changes are considered in this patent) of the predicted time series security copy and smoothly adjusts the sampling interval and target frequency of the sensing node. Specifically, the first derivative function represents the speed of data change; the second derivative reflects the acceleration of data change, that is, the degree of change in the data change speed. When the value of the second derivative is large, it indicates that the change trend of the sensing data is relatively intense, and the gateway makes an intelligent judgment on the sensor sampling rate accordingly. Relatively speaking, when the value of the second derivative is small, it indicates that the data change is relatively gentle, and at this time, the sensor sampling rate can be selected to be reduced to reduce power consumption; the third derivative function represents the change in the acceleration of data change. If the third derivative is large, it indicates that there is a mutation or abnormality, and the sampling rate should be appropriately increased. In this process, the gateway effectively manages energy consumption by dynamically adjusting the working state of the sensor, and achieves a balance between energy conservation and data accuracy while ensuring data integrity.

[0053] The adaptive difference-triggered anomaly detection unit continuously monitors the difference between the actual data collected by the sensor and the spatio-temporal sequence security copy. By continuously comparing the actual data with the security copy, this unit analyzes the deviation between them. When the monitored difference exceeds the sliding window threshold, the unit immediately triggers an alarm to prompt the system user or administrator for further inspection and response. Meanwhile, the unit records the relevant abnormal data, including information such as the difference value, timestamp, and sensor status, for subsequent analysis and auditing. This real-time monitoring mechanism ensures a rapid response to potential abnormal events, improves the reliability and security of the system, and provides strong data support for troubleshooting.

[0054] After detecting an abnormal situation, the adaptive data integrity guarantee unit automatically activates the data recovery mechanism to handle data anomalies or tampering. This unit first identifies the type of anomaly. By analyzing data characteristics and change patterns, it determines the nature of the anomaly, such as data mutation, abnormal fluctuation, or time delay. According to different anomaly types, the unit flexibly adjusts the correction strategy and takes corresponding measures for specific abnormal situations. For data mutation, the unit performs smoothing processing to reduce the impact of the mutation on the overall data set; for abnormal fluctuations, the unit uses a filtering algorithm to identify and remove noise data and restore the true trend of the data; when encountering a time delay, the unit adjusts the data timestamp to ensure the correctness of the data order. Through this multi-level detection and protection mechanism, the adaptive data integrity guarantee unit can quickly respond to abnormal events and automatically recover data when necessary, ensuring data integrity and reliability. In addition, the unit records abnormal events and their processing processes for subsequent auditing and analysis, thereby continuously optimizing the verification mechanism and improving the security and stability of the system.

[0055] In practical applications, the system proposed by the present invention can be widely applied to multiple fields such as smart home, environmental monitoring, agricultural monitoring, and industrial automation. By continuously monitoring environmental changes and promptly responding to emergencies, this system can improve the application level of the intelligent Internet of Things and promote the development of related technologies. Users can comprehensively control various sensor data through this system, thereby making more scientific decisions and responses.

Claims

1. A high-efficiency time-space correlation distributed sensor node security anti-attack verification method, characterized in that: include: After receiving the data, the gateway determines the abnormal state of the node based on the changing rules of the physical characteristics perceived by the sensor node through the spatial neighborhood information, and constructs a multi-dimensional spatial weight model for the abnormal node, including node spacing, wind direction angle, height difference, terrain obstruction, power attenuation and time difference, and ensures the correction accuracy through dynamic weight allocation; After completing the node anomaly processing, a one-dimensional neural network (1D-CNN) and gated recurrent unit (GRU) hybrid model is used to analyze the time series data and generate a time series-neighborhood fusion safety copy; Start the micro-power adaptive data acquisition module, build a multi-order derivative change evaluation model based on the safety copy, and dynamically adjust the sampling frequency through first-order, second-order and third-order weighted analysis; finally, perform difference analysis between the real data and the safety copy and activate the adaptive difference trigger anomaly detection module.

2. The energy-efficient spatiotemporal correlation distributed sensor node security anti-attack verification method according to claim 1 is characterized in that: The micro-power adaptive data acquisition process dynamically adjusts the sampling frequency of the sensor node by constructing a derivative domain change rate evaluation model: a multi-dimensional feature vector is constructed based on the first to third order derivatives of the data sequence, wherein the first order feature characterizes the change rate, the second order feature quantifies the acceleration, and the third order feature captures the mutation intensity; the nonlinear control of the sampling frequency is achieved through a gradient threshold trigger mechanism, the minimum sampling frequency is maintained when the feature vector modulus is lower than the reference threshold, and an exponential incremental response is initiated when the feature vector modulus exceeds the threshold, and the stable operation of the system is ensured by the frequency gain coefficient constraint.

3. The energy-efficient time-space correlation distributed sensor node security anti-attack verification method according to claim 1 is characterized in that: The adaptive difference-triggered anomaly detection module dynamically matches sensor features and data fluctuation patterns through an adaptive sliding window threshold mechanism, and monitors the difference between actual and predicted data in real time; when the difference exceeds the dynamic threshold, an abnormal anti-tampering check is triggered to identify abnormal fluctuations or potential tampering behaviors.

Images