High-security quantum random number generator system based on wavelet transform

Abstract

The invention relates to the field of quantum random number generators, and provides a high-security quantum random number generator system based on wavelet transform, which comprises an original random sequence generation module, and a security monitoring module and a post-processing and randomness detection module which are connected with the original random sequence generation module, the original random sequence generation module is used for generating an original random sequence; the safety monitoring module is used for processing the original random sequence based on wavelet transform so as to judge whether the system is abnormal or not; and the post-processing and randomness detection module is used for performing post-processing and randomness detection on the original random sequence when the system is not abnormal. According to the invention, through real-time monitoring and anomaly detection, the identification capability of signal anomaly and potential attack is significantly improved, and time-frequency analysis is carried out by using wavelet transform, so that the system can accurately capture local changes of signals, and the security is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of quantum random number generators, and in particular, to a high-security quantum random number generator system based on wavelet transform. BACKGROUND

[0002] Random numbers play an indispensable role in modern information technology, especially in the fields of cryptography, statistics, and scientific simulation. For example, in the field of cryptography, secure communication protocols such as encryption, digital signature, and identity verification rely heavily on high-quality random numbers to ensure security. However, traditional pseudo-random number generators (PRNG) generate random numbers based on algorithms and initial seeds, which may be predictable in certain cases, resulting in potential security risks.

[0003] Quantum random number generators (QRNG) extract quantum random numbers by observing quantum physical processes. Unlike classical pseudo-random number generation methods, QRNG generates random numbers from the inherent uncertainty of quantum mechanics, which is theoretically unpredictable and is the closest to ideal true random number generation technology to date, thus having significant advantages in applications with high security requirements.

[0004] A typical QRNG consists of a quantum entropy source (QES), a sampling module, a post-processing module, and a randomness detection module. The QES consists of a quantum light source module and a detection module. The quantum light source module generates a quantum state signal containing randomness, which is then converted into a random electrical signal by the detection module. Due to the inevitable presence of classical noise in actual systems, such as the random electrical signal output by the detection module, which often contains electrical noise, especially in the low frequency band, the electrical noise is more obvious. Therefore, before sampling, a filtering circuit is usually designed to remove low-frequency classical noise to improve the quantum-classical noise signal-to-noise ratio. Then, the sampling module discretizes the filtered electrical signal to obtain the original random sequence, and the post-processing module processes the generated original random sequence to obtain the final random sequence. The randomness detection module performs statistical analysis on the generated final random numbers to verify whether they meet the randomness standard.

[0005] In actual QRNG systems, filter circuits are widely used to filter out low-frequency classical noise to improve the signal-to-noise ratio of quantum signals and the randomness quality of system output. However, as a key component in a QRNG system, the filter module can also become a potential attack surface, especially in the frequency domain, which is vulnerable to spectral attacks. A spectral attack generally refers to an attacker affecting the frequency components of the filtered output signal by interfering with the frequency selection characteristics of the filter, thereby affecting the security of the random numbers generated by the system.

[0006] Although the traditional model often assumes that the attacker can directly manipulate the internal parameters of the filter circuit, in most actual deployment scenarios, the feasibility of such physical access attacks is low. In contrast, a more realistic attack method is to indirectly manipulate the system's spectral response through non-contact means, such as applying a specific frequency of external electromagnetic signals or using power side-channel interference. The following are two typical forms of spectral attacks: The first scenario is to affect the effective frequency response range of the filter through external interference signals. The attacker can apply a spectrum designed precisely electromagnetic disturbance, making the filter exhibit a phenomenon of bandwidth expansion or narrowing: bandwidth expansion can allow more low-frequency classical noise to penetrate the filter into the system, reducing the proportion of quantum signals in the output sequence; bandwidth narrowing can truncate part of the effective quantum signal frequency components. Such disturbances can destroy the spectral structure of the system output sequence, thereby causing randomness degradation or pattern enhancement, reducing the credibility of the system in high-security fields such as cryptography.

[0007] The second scenario is the frequency domain injection of pseudo-random or modulated signals. Pseudo-random signals or artificial interference signals are injected into quantum signals. This attack adjusts the frequency response of the filter circuit, causing the pseudo-random signal to superimpose with the quantum signal, thereby masking the inherent randomness of the quantum signal. The injection of such pseudo-random components can cause the output random numbers to exhibit certain predictable patterns, destroying their inherent uniformity and unpredictability. This attack directly undermines the core characteristics of the quantum random number generator, especially in high-security applications, which can cause security vulnerabilities in random number generation, thereby threatening the security of the system.

[0008] In addition, abnormal states of the quantum entropy source itself can also cause the system to produce unreliable output, thereby destroying the security of the system. The quantum entropy source is the fundamental source of randomness for a QRNG system, so when the quantum entropy source is in an abnormal state, such as device aging failure or environmental temperature interference, the quality of the generated random numbers will decrease. SUMMARY

[0009] The application aims to provide a high-security quantum random number generator system based on wavelet transform to solve the security problem of the QRNG system caused by the attack on the filter circuit and the abnormality of the entropy source.

[0010] The application provides a high-security quantum random number generator system based on wavelet transform, which comprises an original random sequence generation module, a security monitoring module connected with the original random sequence generation module, and a post-processing and randomness detection module. The original random sequence generation module is used to generate an original random sequence. The security monitoring module is used to process the original random sequence based on wavelet transform to determine whether the system is abnormal. The post-processing and randomness detection module is used to process the original random sequence when the system is not abnormal. In a preferred embodiment, the security monitoring switch comprises a wavelet transform module, an abnormality detection module, and an alarm and response module connected in sequence. The wavelet transform module is used to perform multi-scale discrete wavelet decomposition on the original random sequence based on wavelet transform to obtain wavelet coefficients of each scale layer, including approximation coefficients and detail coefficients. The abnormality detection module is used to compare the wavelet coefficients obtained by wavelet transform with statistical characteristics of historical normal operation data to calculate an abnormality total score. The alarm and response module is used to determine whether the system is abnormal according to the calculated abnormality total score.

[0011] In a preferred embodiment, the wavelet transform module is implemented by an orthogonal filter bank.

[0012] In a preferred embodiment, the orthogonal filter bank is composed of a pair of low-pass filters and high-pass filters.

[0013] In a preferred embodiment, the abnormality detection module is specifically used to: calculate four statistical quantities, including the energy and variance of the approximation coefficients and the detail coefficients in the wavelet coefficients obtained by wavelet transform; calculate the historical mean and standard deviation of the corresponding four statistical quantities for the historical normal operation data; calculate the standardized deviation of the four statistical quantities by using the energy and variance and the historical mean and standard deviation; add all the standardized deviations to obtain the abnormality total score; set a threshold value for the normal state or the abnormal state of the system, if the abnormality total score exceeds the threshold value, the system is determined to be in an abnormal state, otherwise the system is in a normal state.

[0014] In a preferred embodiment, the alarm and response module is specifically used to: set the threshold range corresponding to the abnormal level of the system in the abnormal state; According to the threshold range where the total score of the abnormality is located, it is judged whether the system is in a normal state or an abnormal state.

[0015] In a preferred embodiment, the alarm and response module is further used for: According to the abnormal level, different response strategies are adopted.

[0016] In a preferred embodiment, the abnormal level includes: Minor abnormality, indicating that the signal has a small fluctuation, when the minor abnormality occurs, the early warning mechanism is triggered to inform the administrator to monitor and record the abnormal information; Moderate abnormality, indicating that the signal deviates from the normal range, when the moderate abnormality occurs, the system will suspend the quantum random number generation and perform detailed analysis, and manual intervention is needed when necessary; Serious abnormality, indicating that the signal has a significant deviation, when the serious abnormality occurs, the system will immediately stop the quantum random number generation, disconnect the connection with the external communication, start the emergency response mechanism, and inform the administrator to handle the emergency.

[0017] In a preferred embodiment, the safety monitoring switch further includes a monitoring switch, which connects the safety monitoring module with the original random sequence generation module; The monitoring switch is used to control the activation of the safety monitoring module.

[0018] In a preferred embodiment, the monitoring switch includes three working modes: (1) Start opening mode: in the start opening mode, the safety monitoring module will automatically start when the system starts; (2) On-demand opening mode: in the on-demand opening mode, the activation of the monitoring function is triggered by external conditions or internal state changes of the system; (3) Periodic opening mode: in the periodic opening mode, the monitoring function will be started periodically according to the set time interval.

[0019] As described above, due to the adoption of the above technical solutions, the beneficial effects of the present application are: 1. The present application significantly improves the identification ability of signal abnormality and potential attack through real-time monitoring and abnormality detection, and uses wavelet transform for time-frequency analysis, so that the system can accurately capture the local change of the signal, and the security is enhanced.

[0020] 2. The present application takes corresponding emergency response measures according to different levels of abnormality to ensure flexible response to various threats.

[0021] Therefore, the application improves the security, stability and random number quality of the system as a whole, and is suitable for applications with high security requirements. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 A signal with a frequency mutation at 0.5s and the results of Fourier transform and wavelet transform of the signal, respectively.

[0023] Figure 2 A structure diagram of a high-security quantum random number generator system based on wavelet transform provided by an embodiment of the application. DETAILED DESCRIPTION

[0024] To make the purpose, technical scheme and advantages of the embodiments of the application clearer, the technical scheme of the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments. The components of the embodiments of the application described and shown in the drawings can be arranged and designed in various different configurations.

[0025] Therefore, the following detailed description of the embodiments of the application provided in the drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the application.

[0026] EMBODIMENT In order to effectively prevent spectral attacks, the spectrum of the signal can be monitored and analyzed in real time. Accurately mastering the spectral characteristics of the signal and its dynamic changes is the basis for spectral attack protection. In the field of signal processing, Fourier transform (FT) is a commonly used spectral analysis method. Fourier transform can convert time-domain signals to frequency domain, revealing the frequency components and their strengths of the signal. For continuous signals, the mathematical expression of Fourier transform is as follows:

[0027] wherein, is a signal function, is a frequency, t is a time.

[0028] However, Fourier transform also has obvious limitations. It assumes that the signal is stationary, which means that Fourier transform cannot capture the changes in signal spectrum over time, so it is not suitable for analyzing instantaneous or non-stationary signals. For example, when facing dynamic spectrum attacks such as frequency hopping or signal mutation, Fourier transform cannot provide dynamic information of signal spectrum over time, which limits its application in real-time spectrum monitoring.

[0029] To solve this problem, Wavelet Transform (WT) emerged as a time-frequency analysis tool. Unlike FT, Wavelet Transform can simultaneously obtain time and frequency domain information of the signal, thus realizing fine observation of the change of signal spectrum over time. Wavelet Transform can flexibly analyze the characteristics of the signal at different time points and different frequency ranges by using base functions (wavelet functions) of different scales and positions. Wavelet Transform provides multi-resolution analysis of the signal, which can provide localized information of time and frequency simultaneously. Discrete Wavelet Transform (DWT) is a commonly used form, whose mathematical expression is as follows:

[0030] where, is the wavelet base function, is the scale, b is the translation. This feature is particularly suitable for analyzing non-stationary signals or signals with instantaneous characteristics.

[0031] As Figure 1 shown, a signal with a frequency mutation at 0.5s is shown, and Fourier transform and wavelet transform are performed on the signal respectively. From the results, we can only get the overall frequency domain information from the Fourier transform result, while the wavelet transform result can get the time-varying characteristics of the signal. Fourier transform is mainly suitable for stationary signals, providing global frequency information, while wavelet transform has an advantage in analyzing non-stationary signals by providing localized analysis of time and frequency simultaneously. Fourier transform provides global frequency representation, but cannot capture the time-varying characteristics or instantaneous changes of the signal, which makes it suitable for stationary signals, but not suitable for non-stationary signals or signals with local characteristics. On the contrary, wavelet transform provides flexible time-frequency representation, which can change resolution at different scales.

[0032] In view of this, in view of the security problems of QRNG system caused by the fact that the filter circuit is easy to be attacked and the quantum entropy source is abnormal. The embodiment of the application provides a high-security quantum random number generator system based on wavelet transform, which analyzes the signal by introducing wavelet transform technology to judge whether the QRNG system is attacked or the quantum entropy source is in an abnormal state.

[0033] Specifically, by introducing wavelet transform technology, time-frequency analysis can be performed on quantum signals to monitor mutations and abnormalities in the signals in real time. When the frequency response of the filter circuit changes due to external interference or attack or the quantum entropy source is in an abnormal state, the QRNG system can accurately capture these mutations through wavelet transform to timely discover potential attack behaviors or abnormalities. The time-frequency localization capability of wavelet transform can finely identify local frequency changes of the signals, thereby effectively reducing the risk of the filter circuit being manipulated by an enemy. The present application can significantly improve the anti-interference capability and security of the quantum random number generator, especially when facing potential attacks and external interference, it can effectively ensure that the generated random numbers meet the standards of high-security applications, and is widely applicable to high-security fields such as encrypted communication and digital signature.

[0034] As shown in Figure 2 The present application provides a high-security quantum random number generator system based on wavelet transform, which comprises an original random sequence generation module, a security monitoring module connected with the original random sequence generation module, and a post-processing and randomness detection module. The original random sequence generation module is used to generate an original random sequence. The security monitoring module is used to process the original random sequence based on wavelet transform to determine whether the system is abnormal. The post-processing and randomness detection module is used to post-process and detect the randomness of the original random sequence when the system is not abnormal.

[0035] The specific implementation and functional principle of each module in the system are as follows: Module 1, original random sequence generation module: The original random sequence generation module comprises a quantum entropy source, a filter circuit, and a discretization sampling module connected in sequence, wherein the quantum entropy source comprises a quantum light source and a detection module connected in sequence.

[0036] The working principle of the original random sequence generation module is as follows: First, the quantum light signal generated by the quantum light source is detected by the detection module, which converts the quantum light signal into a measurable electrical signal. Then, the electrical signal output by the detection module is processed by the filter circuit to filter out low-frequency classical noise.

[0037] Finally, the continuous electrical signal processed by the filter circuit is discretized and sampled by the discretization sampling module to convert the filtered continuous electrical signal into a discrete original random sequence.

[0038] Module 2, security monitoring module: The safety monitoring switch includes a monitoring switch, a wavelet transform module, an anomaly detection module, and a warning and response module connected in sequence.

[0039] 1. The monitoring switch can be configured as needed to control the activation of the safety monitoring module, ensuring that the system can flexibly activate the monitoring function according to actual needs, maximizing efficiency and safety.

[0040] In this embodiment, the monitoring switch includes three operating modes: startup triggered mode, on-demand triggered mode, and periodic triggered mode. Here is a detailed description of these three modes: (1) Startup Triggered: In the startup triggered mode, the safety monitoring module is automatically started when the system starts. That is, when the quantum random number generator (QRNG) system starts working, the monitoring switch will immediately activate the monitoring function. This startup triggered mode is suitable for scenarios where the entire system security needs to be continuously monitored, ensuring that any potential security threats can be discovered in real time during the initial running phase of the system. By starting the monitoring module at system startup, this startup triggered mode can provide immediate response to potential problems in the early stages of the system.

[0041] (2) On-Demand Triggered: In the on-demand triggered mode, the activation of the monitoring function is triggered by external conditions or internal state changes of the system. When the system needs to be checked for security, the monitoring module is activated. This on-demand triggered mode is more flexible, enabling monitoring to be activated when specific security needs arise, avoiding unnecessary resource consumption. It is suitable for situations where the system does not need to be highly monitored all the time, but can respond promptly when an anomaly occurs.

[0042] (3) Periodic Triggered: In the periodic triggered mode, the monitoring function is activated periodically according to a set time interval. This periodic triggered mode ensures that the system checks its security status periodically during operation, effectively capturing any potential anomalies or attack behavior. This periodic triggered mode is suitable for applications that require periodic detection of signal stability or security, enabling effective monitoring of randomness and security during long-term system operation.

[0043] After the monitoring switch is turned on, the system begins real-time analysis of quantum signals through the wavelet transform module and the anomaly detection module, ensuring the safety of the system and the quality of the signals.

[0044] 2. Wavelet Transform Module The wavelet transform module is configured to perform multi-scale discrete wavelet decomposition on the original random sequence based on wavelet transform to obtain wavelet coefficients at each scale, including approximation coefficients and detail coefficients.

[0045] The wavelet transform module performs time-frequency analysis on the input signal to capture transient changes and frequency anomalies therein. When the input signal is the original random sequence, a discrete wavelet transform (DWT) is applied to perform multi-scale discrete wavelet decomposition. The amount of data of the input data has a direct impact on the reliability and computational complexity of the results. To ensure that sufficient signal features can be captured and accurate anomaly detection can be performed, the amount of data should be large enough to ensure multi-scale signal decomposition, while not being too large to cause excessive consumption of computing resources.

[0046] For an original random sequence with a length of , the discrete wavelet transform decomposes it into low-frequency (approximation) components and high-frequency (detail) components at multiple scales. Specifically, the signal is decomposed into a low-frequency component (approximation part) and multiple high-frequency components (detail parts), each representing a different frequency range of the signal. The number of frequency components is related to the number of wavelet transform layers. The low-frequency component (approximation) represents the overall profile of the signal and generally contains slower-changing information, similar to the "trend" of the signal. The high-frequency components (details) represent rapid changes in the signal, which are further subdivided into multiple subbands, each representing the signal's changes within a certain frequency range. The more high-frequency components, the richer the signal's detail information.

[0047] In this embodiment, the original random sequence is decomposed into multiple scales using an orthogonal wavelet function family (such as Daubechies, Haar, Morlet), and the maximum decomposition level is determined (generally 6-8 layers). That is, the wavelet transform module is implemented using an orthogonal filter bank, which is a core tool for implementing wavelet transform. It consists of a pair of low-pass and high-pass filters, with the low-pass filter extracting low-frequency information and the high-pass filter extracting high-frequency information. The signal is divided into multiple frequency bands by the orthogonal filter bank, with each frequency band containing different frequency information. Through the orthogonal filter bank (low-pass filter coefficients and high-pass filter coefficients ), wavelet coefficients at each scale are obtained in turn. For the th layer, its approximation coefficient and detail coefficient are calculated as follows:

[0048]

[0049] Initially, there are By repeating the above calculation until the first wavelet coefficient set of the layer: approximation coefficients and detail coefficients at each scale wherein .

[0050] 3. Anomaly Detection module The anomaly detection module relies on the time-frequency analysis results provided by the wavelet transform to identify abnormal fluctuations in the signal. After the wavelet transform completes the time-frequency decomposition of the signal, the anomaly detection module identifies any abnormal behavior by analyzing the statistical characteristics of each frequency component. If the change of a certain frequency component exceeds the predetermined threshold range, the detection module identifies the signal as an abnormal signal.

[0051] The core of anomaly detection is to compare the wavelet coefficients obtained by wavelet transform with the statistical characteristics of historical normal operation data to calculate the total score of abnormality. In this embodiment, specifically: (1) Calculate four statistics, including the energy and variance of the approximation coefficients and detail coefficients obtained by wavelet transform: The energy and variance of the wavelet approximation coefficients of the first k layer and are calculated as follows:

[0052]

[0053] wherein, is the mean of the wavelet approximation coefficients of the first layer, is the sample number of the wavelet approximation coefficients of the first layer. The energy and variance of the wavelet detail coefficients are also calculated as follows:

[0054]

[0055] wherein, is the mean of the wavelet detail coefficients of the first layer, is the sample number of the wavelet detail coefficients of the first layer.

[0056] ​(2) For historical normal runtime data, calculate the historical mean and standard deviation of the corresponding four statistics; the past normal QRNG sequence can be used to establish a reference statistical model to calculate the historical mean and standard deviation of the above four statistics: , , , In particular, if there is a lack of historical data in the scenario of the initial start of the system, a set of reference statistical values can be constructed through simulation data in the design phase as the default standard in the initial stage of the system going online.

[0057] (3) Calculate the standardized deviation (z-score) of the four statistics using the energy and variance and the historical mean and standard deviation, and the calculation formula is as follows:

[0058] Wherein, represents the current energy or variance, is the corresponding historical mean, is the corresponding historical standard deviation.

[0059] (4) Accumulate all the standardized deviations to get the total score of anomalies, and the calculation formula is as follows:

[0060] Wherein, M represents the total score of anomalies, and the greater the total score of anomalies, the more abnormal the data.

[0061] (5) Set the threshold value a of the system in the normal state or the abnormal state, if the total score of anomalies exceeds the threshold value, it is determined that the system is in an abnormal state, otherwise the system is in a normal state.

[0062] 4. Alarm and response module: The alarm and response module is started after the abnormality checking module finds that the system is in an abnormal state, and is responsible for classifying, evaluating and taking corresponding emergency response measures for abnormal conditions. These abnormalities may be caused by various reasons, including system failure, environmental changes or enemy attack filter circuit, etc. Therefore, the alarm and response module is specifically used for: Setting the threshold range corresponding to a plurality of abnormal levels of the system in the abnormal state; According to the threshold range where the total score of anomalies is located, it is determined whether the system is in a normal state or an abnormal state; According to the abnormal level, different response strategies are taken.

[0063] In this embodiment, the abnormality level is divided into three categories according to the degree of deviation of the signal: slight abnormality, moderate abnormality and severe abnormality. In combination with the aforementioned determination of the normal state, the system divides the state of the system into the following four categories according to the total score M of the current data in actual operation: (1) If M , the system is determined to be in a normal state; (2) If M , the system is determined to be in a slight abnormality; slight abnormality usually indicates that the signal has a small fluctuation, which may be caused by natural noise, environmental changes or other non-malicious factors. This type of abnormality will trigger a warning mechanism to notify the administrator to monitor and record abnormal information.

[0064] (3) If M , the system is determined to be in a moderate abnormality; moderate abnormality indicates that the signal deviates from the normal range, which may be caused by interference of the enemy to the filtering circuit or other systematic problems. At this time, the system will suspend the generation of random numbers and perform detailed analysis, and manual intervention if necessary.

[0065] (4) If M , the system is determined to be in a severe abnormality. Severe abnormality indicates that the signal has a significant deviation, which may be caused by complex attacks of the enemy to the filtering circuit or other serious problems. At this time, the system will immediately stop the generation of random numbers, disconnect the connection with external communication, start the emergency response mechanism, and notify the administrator for emergency handling.

[0066] Wherein, the parameters a, b, c are the dividing points for dividing the abnormality level, satisfying a

[0067] Module three, post-processing and randomness detection module: The post-processing and randomness detection module is used to post-process and detect randomness of the original random sequence when the system does not have abnormality, to ensure that the final output random number has high quality randomness and security, which specifically includes a post-processing module and a randomness detection module.

[0068] Firstly, the post-processing module will post-process the original random sequence, mainly including removing possible deviations, correcting non-uniform distribution and removing correlation between data, etc. This process optimizes the statistical properties of the data to ensure that the final output random number meets strict randomness standards such as uniformity and independence. Common post-processing methods include: m-LSB method, XOR method, Toeplitz matrix method, etc.

[0069] Subsequently, the post-processed data needs to pass the strict test of the randomness detection module. Randomness detection usually uses a series of statistical tests (such as NIST-STS, TestU01, GM / T 0005-2021, etc.) to evaluate the quality of random numbers and verify whether the random numbers meet the high security requirements. If the data passes the randomness detection, it can be finally output; if it fails the detection, the data will be discarded and the random number generation process will be restarted.

[0070] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A high-security quantum random number generator system based on wavelet transform, characterized in that, It includes a raw random sequence generation module, a security monitoring module connected to the raw random sequence generation module, and a post-processing and randomness detection module; The original random sequence generation module is used to generate the original random sequence; The security monitoring module is used to process the original random sequence based on wavelet transform in order to determine whether there is an anomaly in the system; The post-processing and randomness detection module is used to perform post-processing and randomness detection on the original random sequence when there are no abnormalities in the system.

2. The high-security quantum random number generator system based on wavelet transform according to claim 1, characterized in that, The safety monitoring switch includes a wavelet transform module, an anomaly detection module, and an early warning and response module connected in sequence. The wavelet transform module is used to perform multi-scale discrete wavelet decomposition on the original random sequence based on wavelet transform to obtain wavelet coefficients at each scale level, including approximation coefficients and detail coefficients. The anomaly detection module is used to compare the wavelet coefficients obtained by wavelet transform with the statistical characteristics of historical normal operation data to calculate the total anomaly score. The alarm and response module is used to determine whether there is an anomaly in the system based on the calculated total anomaly score.

3. The high-security quantum random number generator system based on wavelet transform according to claim 2, characterized in that, The wavelet transform module is implemented using an orthogonal filter bank.

4. The high-security quantum random number generator system based on wavelet transform according to claim 3, characterized in that, The orthogonal filter bank is implemented by a pair of low-pass filters and a high-pass filter.

5. The high-security quantum random number generator system based on wavelet transform according to claim 1, characterized in that, The anomaly detection module is specifically used for: Calculate four statistics, including the energy and variance of the approximation coefficients and detail coefficients in the wavelet coefficients obtained by wavelet transform; For historical normal operating data, calculate the historical mean and standard deviation of the four corresponding statistics; Calculate the standardized deviation of the four statistics using energy, variance, historical mean, and standard deviation; The total anomaly score is obtained by summing all the standardized deviations. Set a threshold for whether the system is in a normal or abnormal state. If the total abnormal score exceeds the threshold, the system is determined to be in an abnormal state; otherwise, the system is in a normal state.

6. The high-security quantum random number generator system based on wavelet transform according to claim 5, characterized in that, The alarm and response module is specifically used for: Set threshold ranges for several abnormal levels when the system is in an abnormal state; Based on the threshold range of the total abnormal score, the system is determined to be in a normal or abnormal state.

7. The high-security quantum random number generator system based on wavelet transform according to claim 6, characterized in that, The alarm and response module is also used for: Different response strategies are adopted based on the level of anomaly.

8. The high-security quantum random number generator system based on wavelet transform according to claim 7, characterized in that, The anomaly levels include: A minor anomaly indicates a slight fluctuation in the signal. When a minor anomaly occurs, an early warning mechanism is triggered to notify the administrator to monitor and record the anomaly information. A moderate anomaly indicates that the signal deviates from the normal range. When a moderate anomaly occurs, the system will pause quantum random number generation and conduct detailed analysis, and will intervene manually if necessary. A serious anomaly indicates a significant deviation in the signal. When a serious anomaly occurs, the system will immediately stop generating quantum random numbers, disconnect from external communication, activate the emergency response mechanism, and notify the administrator for emergency handling.

9. The high-security quantum random number generator system based on wavelet transform according to any one of claims 2-8, characterized in that, The safety monitoring switch also includes a monitoring switch that connects the safety monitoring module to the original random sequence generation module. The monitoring switch is used to control the activation of the safety monitoring module.

10. The high-security quantum random number generator system based on wavelet transform according to claim 9, characterized in that, The monitoring switch includes three operating modes: (1) Startup Enabled Mode: In startup enabled mode, the security monitoring module will automatically start when the system starts; (2) On-demand activation mode: In on-demand activation mode, the monitoring function is activated by external conditions or changes in the internal state of the system; (3) Periodic activation mode: In periodic activation mode, the monitoring function will start periodically according to the set time interval.

Images