APT covert channel identification method and system based on multi-mode anomaly detection

Through the multimodal anomaly detection method, the variational autoencoder and adaptive multi-scale deep-coupled autoencoder are used to solve the problems of low accuracy, high false alarm rate and poor adaptability in the existing technology, and realize the problem of low APT detection accuracy, high false alarm rate and poor adaptability. High accuracy, low false alarm rate and fast response APT hidden channel detection meet the needs of network security protection.

CN119066464AActive Publication Date: 2024-12-03UNIV OF ELECTRONICS SCI & TECH OF CHINA

Patent Information

Application Number
CN202411044854.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-12-03
Estimated Expiration
2044-07-31

AI Technical Summary

Technical Problem

When facing complex and variable APT attacks, existing APT detection methods generally have problems such as low detection accuracy, high false alarm rate, extended detection time, and poor adaptability, especially in multimodal data fusion and long-term behavior. There are obvious shortcomings in analysis and real-time adaptation, which is difficult to meet the current needs of network security protection.

Method used

Using multimodal anomaly detection method, normal behavior modeling is achieved through variational autoencoder, adaptive multi-scale deep-coupled autoencoder performs joint representation learning, combining contrast learning and long-term memory network capture timing abnormal mode, and using reinforcement learning Perform adaptive sampling, and aggregate multi-party knowledge through federated learning, update the anomaly detection model, and ultimately realize the identification and detection of APT hidden channels.

Benefits of technology

Significantly improve the detection accuracy of APT hidden channels, reduce false alarm rates, shorten detection delays, enhance system adaptability and resource utilization efficiency, provide clear attack insights, and provide effective defense strategies for security analysts. Supported.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119066464A_ABST
    Figure CN119066464A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of APT (Advanced Persistent Threat) covert channel identification methods, provides an APT covert channel identification method and system based on multi-modal anomaly detection, and aims to solve the problem of detection of a covert channel in an advanced persistent threat (APT) attack. Through multi-modal anomaly detection, the method can identify and analyze APT attacks in a complex network environment, especially those attacks using covert channels. The invention provides an APT covert channel identification method based on multi-modal anomaly detection. The method comprises the steps of obtaining multi-modal data, performing normal behavior modeling, performing joint representation learning, enhancing anomaly sensitivity, capturing a time sequence anomaly mode, optimizing anomaly features, updating an anomaly detection model, identifying an APT covert channel and generating a detection report. The APT covert channel detection method is used for improving the detection accuracy of the APT covert channel, reducing the false alarm rate, improving the detection time delay, enhancing the adaptability of the model, reducing the consumption of computing resources and providing a powerful network security protection means for enterprises.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of APT covert channel identification methods, and more specifically, to an APT covert channel identification method and system for multi-modal anomaly detection. Background Art

[0002] With the rapid development of information technology and the continuous evolution of network security threats, Advanced Persistent Threat (APT) has become one of the most difficult problems in the current network security field. APT attacks are highly concealed, long-lasting, and highly targeted, and traditional network security protection measures are often difficult to effectively deal with.

[0003] In recent years, APT detection technology has evolved from a single-modality, rule-based approach to a multi-modality, machine learning-based approach. Currently, there are mainly the following APT detection methods:

[0004] 1. Rule-based detection method: This is the earliest APT detection method, which uses a predefined set of rules to identify suspicious behaviors. Although it is simple to implement and consumes low computing resources, it has the following problems:

[0005] a) It is difficult to maintain rules and cope with the rapidly changing APT attack techniques;

[0006] b) High false alarm rate and weak detection capability for unknown attack patterns;

[0007] c) Unable to capture complex attack patterns and long-term behavioral anomalies.

[0008] 2. Single-modality detection methods based on machine learning: This type of method uses machine learning algorithms, such as support vector machines (SVMs) and random forests, to analyze a single data source (such as network traffic) to detect APTs. Although this method is an improvement over rule-based methods, it still has the following problems:

[0009] a) Reliance on a single data source limits the comprehensiveness of detection;

[0010] b) It is difficult to capture complex attack patterns across multiple data sources;

[0011] c) Feature engineering relies too much on expert experience and is difficult to automatically adapt to new attack patterns.

[0012] 3. Single-modality detection methods based on deep learning: This type of method uses deep neural networks, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), to automatically learn features and detect APTs. Although there has been a significant improvement in feature learning capabilities, it still faces the following challenges:

[0013] a) The model is highly complex and consumes large amounts of computing resources;

[0014] b) It is still limited to a single data source, making it difficult to fully understand the full picture of APT attacks;

[0015] c) The model has poor interpretability and is difficult to provide meaningful insights for security analysts.

[0016] 4. Simple multi-modal fusion APT detection method: This method attempts to combine multiple data sources, such as network traffic, log data, terminal behavior, etc., to improve the comprehensiveness of detection. Although the detection performance has been improved, the following problems still exist:

[0017] a) The multimodal data fusion method is simple, often using simple feature splicing or weighted averaging, which makes it difficult to fully utilize the complementary information between different modalities;

[0018] b) There is a lack of dynamic evaluation mechanism for the importance of different modes, which makes it difficult to adapt to the characteristics of different stages of APT attacks;

[0019] c) The timing and long-term nature of APT attacks are not taken into account, making it difficult to capture the evolution of complex attack patterns.

[0020] In summary, existing APT detection methods generally have problems such as low detection accuracy, high false alarm rate, extended detection time, and poor adaptability when facing increasingly complex and changeable APT attacks. In particular, there are obvious deficiencies in multimodal data fusion, long-term behavior analysis, and real-time adaptation, which makes it difficult to meet the urgent needs of current network security protection. Summary of the invention

[0021] The purpose of the present invention is to solve the problem of detecting covert channels in advanced persistent threat (APT) attacks. Through multimodal anomaly detection, the method can identify and analyze APT attacks in complex network environments, especially those using covert channels.

[0022] In order to solve the above technical problems, the present invention provides a multimodal anomaly detection APT covert channel identification method and system thereof.

[0023] The present invention provides an APT covert channel identification method for multi-modal anomaly detection, comprising the following steps:

[0024] Obtain multimodal data, and based on the multimodal data, implement normal behavior modeling through a variational autoencoder to obtain a normal behavior model; use an adaptive multi-scale deep coupling autoencoder to perform joint representation learning on the multimodal data to obtain a joint representation; based on the joint representation, enhance abnormal sensitivity through contrastive learning to obtain an enhanced feature representation; use a long short-term memory network to process the enhanced feature representation, capture the timing abnormality pattern, and obtain the timing abnormality feature; use reinforcement learning to adaptively sample the timing abnormality feature to obtain an optimized abnormal feature; aggregate multi-party knowledge through federated learning and update the abnormality detection model; based on the updated abnormality detection model, analyze the optimized abnormality feature to realize the identification of APT covert channels; and generate an APT covert channel detection report according to the identification result.

[0025] Specifically, the adaptive multi-scale deep coupled autoencoder includes:

[0026] Multi-scale feature extraction module, used to extract multi-scale features of multi-modal data;

[0027] Dynamic modal importance assessment module, used to assess the importance of different modes;

[0028] A sparse anomaly feature enhancement module for enhancing potential anomaly features; and

[0029] Adaptive feature fusion module, used to fuse multimodal features.

[0030] Specifically, the multi-scale feature extraction module is implemented using a spiral pyramid network, where:

[0031] F i =SPN(X i )=Concat[S1(X i ), S2(Xi), ..., S K (X i )]

[0032]

[0033] Among them, F i is the extracted feature, X i is the input data, S k is the k-th layer spiral convolution operation, σ is the Mish activation function, W k is the convolution kernel, R θ is the rotation transformation, K is the total number of layers, D k It is a dilated convolution with a dilation rate of 2 k , Represents element-wise addition.

[0034] Specifically, the dynamic modal importance evaluation module is implemented based on quantum annealing, wherein:

[0035] α i =QA(Φ(F i ))Φ(x)=cos(πx)+i·sin(πx)

[0036] Among them, α i is the importance weight of mode i, QA represents the quantum annealing process, Φ is the complex mapping function, F i is the output of the multi-scale feature extraction module, and i is an imaginary unit.

[0037] Specifically, the sparse anomaly feature enhancement module is implemented based on fractional differentials, where:

[0038] E i =F i +λ·D α [SparseMask(F i )]

[0039]

[0040] Among them, E i is the enhanced feature, F i is the input feature, λ is a learnable parameter, D α represents the α-order fractional differential operator, α is the fractional order, Γ is the gamma function, SparseMask is the differentiable sparse mask based on Gumbel-Softmax, and t and τ are time variables.

[0041] Specifically, the adaptive feature fusion module is implemented by nonlinear harmonic averaging, where:

[0042]

[0043] Among them, Z is the final joint representation, α i is the modal importance weight, E i is the enhanced feature, p is a learnable parameter, and ∑ represents the sum of all modes.

[0044] Specifically, the long short-term memory network adopts an improved LSTM unit, including an attention mechanism and a residual connection, wherein:

[0045] f t =σ(W f ·[h t-1 , x t ]+b f )

[0046] i t=σ(W i ·[h t-1 ,x t ]+b i )

[0047] o t =σ(W o ·[h t-1 , x t ]+b o )

[0048]

[0049] a t =Attention(h t -1,x t )

[0050]

[0051] Among them, f t 、i t , o t They are forget gate, input gate and output gate respectively, c t is the unit state, h t is the hidden state, a t is the attention weight, x t is the input, σ is the sigmoid function, tanh is the hyperbolic tangent function, W f , W i ,W o , W c is the weight matrix, b f 、b i 、b o 、b c is the bias term, is element-by-element multiplication, and Attention is the attention calculation function. The improved LSTM unit is designed to solve the problem of forgetting the timing of abnormal data packet transmission intervals in a specific network environment. Among them, the input is the abnormal time interval of the data packet; the weight matrix is ​​formed by quantizing the statistical average of the average data packet transmission time interval in a specific network environment within a certain period of time, and has certain memory characteristics: the attention weight can be adjusted according to the actual network situation.

[0052] Specifically, the reinforcement learning adaptive sampling is implemented using DuelingDQN, where:

[0053]

[0054] Among them, Q(s,a) is the state-action value function, V(s) is the state value function, A(s,a) is the advantage function, s is the state, a is the current action, a′ is all possible actions, |A| is the size of the action space, and ∑ represents the sum of all actions.

[0055] Specifically, the federated learning multi-party knowledge aggregation is implemented using homomorphic encryption and secure multi-party computing, where:

[0056]

[0057] Among them, θ (t+1) are global model parameters, is the local model parameter of participant i, t represents the number of iterations, N represents the number of participants, HE represents the homomorphic encryption operation, η represents the learning rate, is the gradient of the loss function, D i is the local data of participant i, N(0,σ 2 ) has a mean of 0 and a variance of σ 2 Gaussian noise.

[0058] Multimodal anomaly detection APT covert channel identification system, including:

[0059] A multimodal data acquisition module, used for acquiring multimodal data;

[0060] A variational autoencoder module, used for implementing normal behavior modeling based on the multimodal data to obtain a normal behavior model;

[0061] An adaptive multi-scale deep coupled autoencoder module, used for performing joint representation learning on the multimodal data to obtain a joint representation;

[0062] A contrastive learning module, used for enhancing abnormal sensitivity based on the joint representation to obtain an enhanced feature representation;

[0063] A long short-term memory network module is used to process the enhanced feature representation, capture the time series abnormal pattern, and obtain the time series abnormal feature;

[0064] A reinforcement learning module, used for adaptively sampling the time series abnormal features to obtain optimized abnormal features;

[0065] Federated learning module, used to aggregate multi-party knowledge and update anomaly detection models;

[0066] An APT covert channel identification module, used to analyze the optimized abnormal features based on the updated abnormal detection model to identify the APT covert channel; and

[0067] The report generation module is used to generate an APT covert channel detection report based on the identification results.

[0068] The beneficial effects of the present invention are mainly reflected in the following aspects:

[0069] 1. High accuracy and low false alarm rate: The adaptive multi-scale deep coupled autoencoder realizes the effective fusion of multimodal data, significantly improving the detection accuracy of APT covert channels and greatly reducing the false alarm rate. Experimental results show that the detection accuracy of this method reaches 97.8% and the false alarm rate is only 1.2%, which is much better than existing methods.

[0070] 2. Rapid response capability: The reinforcement learning-based adaptive sampling strategy is adopted to greatly shorten the detection delay of APT attacks. Test results show that the average detection delay of this method is only 2.3 hours, which is more than 8 times faster than traditional methods.

[0071] 3. Excellent adaptability: By aggregating multi-party knowledge through federated learning methods and combining dynamic modal importance evaluation mechanisms, the system can quickly adapt to different network environments and evolving attack patterns. The performance variance is only 0.015, showing excellent stability.

[0072] 4. Comprehensive anomaly capture capability: Combining variational autoencoders, contrastive learning, and long short-term memory networks, this method can effectively capture short-term and long-term anomaly patterns, providing early detection of APT attacks.

[0073] Provided strong support.

[0074] 5. High resource utilization efficiency: Although a complex deep learning model is used, through careful algorithm design and optimization, this method maintains relatively moderate resource consumption while providing excellent performance, enabling it to operate stably for a long time in actual network environments.

[0075] 6. Strong explainability: By generating detailed APT covert channel detection reports, this method provides security analysts with clear attack insights and helps to quickly develop effective defense strategies.

[0076] In summary, the method provided by the present invention has achieved a significant technical breakthrough in APT covert channel detection, providing enterprises and organizations with an efficient, reliable, and adaptable network security protection solution, effectively responding to increasingly complex network security threats. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 This is a flow chart of the APT covert channel identification method for multi-modal anomaly detection of the present invention.

[0078] Figure 2This is a module composition diagram of the adaptive multi-scale deep coupled autoencoder module of the present invention.

[0079] Figure 3 This is a schematic diagram of the APT attack chain reconstruction and prediction system framework based on causal reasoning of the present invention. DETAILED DESCRIPTION

[0080] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects thereof are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0081] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0082] The present invention provides a multimodal anomaly detection APT covert channel identification method and system based on an adaptive multi-scale deep coupled autoencoder, which is used to identify covert channels used by advanced persistent threats (APT) in a complex network environment.

[0083] The present invention is further described in detail below in conjunction with specific examples. The following examples are only used to more clearly illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention.

[0084] Please refer to Figure 1 In a typical application scenario, the method and system of the present invention are deployed in a network environment of a large multinational enterprise. The enterprise has multiple offices and data centers around the world and faces persistent advanced network threats from different regions and organizations. The method and system of the present invention detects potential APT covert channels in real time by analyzing multiple data sources in the enterprise network, thereby protecting the enterprise's sensitive data and intellectual property rights.

[0085] Specifically, the method of the present invention comprises the following steps:

[0086] 1. Obtain multimodal data. In this step, the system collects data from multiple data sources on the enterprise network. These data sources include but are not limited to:

[0087] Network traffic data: Use high-performance network probes (such as Gigamon GigaVUE-HC2) deployed at key nodes of the enterprise network to capture data at each layer of the TCP / IP protocol stack.

[0088] DNS request data: obtained through the log of the enterprise's DNS server (such as Microsoft DNS Server or BIND), including request time, domain name, request type and other information.

[0089] Encrypted communication data: Use SSL / TLS proxy servers (such as Blue Coat SSL Visibility Appliance) to collect encrypted communication metadata, such as handshake information, cipher suites, etc.

[0090] These multimodal data are initially aggregated and preprocessed through the enterprise's security information and event management (SIEM) system (such as Splunk Enterprise Security).

[0091] 2. Based on the multimodal data, normal behavior modeling is implemented through a variational autoencoder to obtain a normal behavior model. In this step, the system uses a variational autoencoder (VAE) to model the normal network behavior of each modality.

[0092] The architecture design of VAE is as follows:

[0093] Encoder: 3-layer fully connected network, the number of neurons in each layer is 1024, 512, 256 respectively, and the activation function is LeakyReLU (α=0.2);

[0094] Decoder: 3-layer fully connected network, the number of neurons in each layer is 256, 512, 1024 respectively, the activation function is ELUVAE, the potential space dimension is dynamically adjusted by the following formula:

[0095] d=max(10,int(log2(n features )·k))

[0096] Among them, n features is the input feature number, k is an adjustable parameter, and its initial value is set to 2. In practice, we found that the value range of k is between 1.5 and 2.5, which is a good effect.

[0097] The loss function of VAE is designed as follows:

[0098] L VAE =MSE(X,X′)+β·KL(q(z|X)||p(z))

[0099] The MSE(X, X′) mean square error (MSE) loss function is used to evaluate the difference between the encoder reconstructed data X and the original data X;

[0100] KL(q(z|X)||p(z)) is the KL divergence loss function, which is used to measure the difference between the conditional distribution q(z|X) given by the encoder about the latent variable z and the prior distribution p(z). KL divergence is used to encourage the encoder to learn a distribution q(z|X) that is closer to the prior distribution p(z), thereby helping the model generate more realistic data;

[0101] β is a dynamically adjusted weight parameter: is the current training step number. This design allows the model to pay more attention to the reconstruction error in the early stage of training, and then gradually increase the weight of KL divergence. The anomaly score calculation formula is as follows:

[0102] AS=α·MSE(X,X′)+(1-α)·KL(q(z|X)||p(z))

[0103] α is a learnable parameter with an initial value of 0.5. In practical applications, we find that the value of α usually converges to between 0.3 and 0.7 during training.

[0104] 3. Using an adaptive multi-scale deep coupled autoencoder to perform joint representation learning on the multimodal data, the step of obtaining the joint representation is one of the core innovations of the present invention. The adaptive multi-scale deep coupled autoencoder includes the following modules:

[0105] 3.1 Multi-scale feature extraction module 31, which is implemented using an innovative spiral pyramid network (SPN):

[0106] F i =SPN(X i )=Concat[S1(X i ), S2(X i ), ..., S K (X i )]

[0107]

[0108] Among them, F i is the extracted feature, X i is the input data (such as network traffic feature vector), S k is the k-th layer spiral convolution operation, σ is the Mish activation function, W k is the convolution kernel, R θ is the rotation transformation, K is the total number of layers (in practice we use K = 5), D k It is a dilated convolution with a dilation rate of 2 k , This design allows the network to extract features at different scales and directions, and is particularly suitable for capturing multi-scale patterns in network traffic and DNS requests.

[0109] 3.2 Dynamic modal importance evaluation module 32, which is implemented based on quantum annealing:

[0110] α i =QA(Φ(F j ))

[0111] Φ(x)=Cos(πx)+i.sin(πx)

[0112] Among them, α i is the importance weight of mode i, QA represents the quantum annealing process, Φ is the complex mapping function, F i is the output of the multi-scale feature extraction module, and i is an imaginary unit. The quantum annealing process simulates the following Hamiltonian:

[0113]

[0114] Among them, J ij encodes the correlation between modes; Γ is the annealing parameter, represents the Pauli Z operator, acting on the i-th quantum bit; represents the Pauli Z operator, acting on the j-th quantum bit; It is represented by the Pauli X operator, which acts on the i-th quantum bit; H represents the Hamiltonian of the system, which describes the total energy of the system; ∑ is the summation symbol, which means the summation of all possible i and j; J ij is the coupling strength, which represents the correlation between the quantum bit i data modes. ij The value of can be positive or negative, with a positive value indicating ferromagnetic coupling (tending to the same state) and a negative value indicating antiferromagnetic coupling (tending to the opposite state). At the beginning of the annealing process, the value of Γ is large, and as the annealing process proceeds, v gradually decreases to close to zero. i, i is the index of the quantum bit. In our application, each i or j can represent a specific data mode. D-Wave quantum annealer, which is a quantum computing device specially designed to perform quantum annealing algorithms. It solves the optimization problem by physically implementing the above Hamiltonian. The annealing time is set to 20μs (microseconds) in this application. This is the duration of the entire quantum annealing process, during which the system evolves from the initial state to the final state.

[0115] In the APT detection application proposed in this patent, this quantum annealing process is used to dynamically evaluate the importance of different data modalities. ijThe value of is set according to the correlation between different data modalities (such as network traffic, DNS requests, encrypted communications, etc.). These values ​​can be obtained through correlation analysis in the preprocessing step. Each qubit represents a data modality. At the end of the annealing process, the final state of the qubit is used to determine the importance weight of the corresponding modality. The Γ parameter gradually decreases from a large initial value (such as 2.0) to close to 0 (such as 0.01) during the annealing process. This process simulates the transition from a quantum superposition state to a classical state. The annealing time of 20μs was experimentally optimized, and within this time range, the system can fully explore the solution space and converge to a good solution. In this way, we use the advantages of quantum computing to efficiently solve the complex combinatorial optimization problem of modal importance evaluation, thereby improving the accuracy and responsiveness of APT detection.

[0116] In practical applications, we use a D-Wave quantum annealer to implement this process, with the annealing time set to 20 μs.

[0117] 3.3 Sparse abnormal feature enhancement module 33, which is implemented based on fractional differential:

[0118] E i =F i +λ·D α [SparseMask(F i )]

[0119]

[0120] Among them, E i is the enhanced feature, F i is the input feature, λ is a learnable parameter (initial value is set to 0.1), D α represents the α-order fractional differential operator, α is the fractional order (we found that α = 0.5 works well), Γ is the gamma function, SparseMask is a differentiable sparse mask based on Gumbel-Softmax, and t and τ are time variables.

[0121] This design can effectively enhance sparse anomaly features and is particularly suitable for detecting tiny anomalies in APT attacks.

[0122] 3.4 Adaptive feature fusion module 34, which is implemented by nonlinear harmonic averaging:

[0123]

[0124] Among them, Z is the final joint representation, α i is the modal importance weight, E iis the enhanced feature, p is a learnable parameter (the initial value is set to -1) ∑ represents the sum of all modes i. In practice, we find that the value of p usually converges to between -2 and 2, which allows the model to dynamically adjust between the arithmetic mean and the geometric mean.

[0125] 4. Based on the joint representation, enhance the abnormal sensitivity through contrastive learning to obtain the enhanced feature representation. In this step, we use the improved InfoNCE loss function:

[0126]

[0127] Among them, s is the cosine similarity, τ is the temperature parameter, which is adjusted dynamically:

[0128]

[0129] The initial value of τ0 is 0.07, and T is the total number of training steps. This design allows the model to pay more attention to obvious differences in the early stages of training, and then gradually increase its sensitivity to subtle differences.

[0130] 5. Use the long short-term memory network to process the enhanced feature representation, capture the abnormal time series pattern, and obtain the abnormal time series features. This step uses an improved LSTM unit, including an attention mechanism and a residual connection:

[0131] f t =σ(W f ·[h t-1 , x t ]+b f )

[0132] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0133] o t =σ(W o ·[h t-1 , x t ]+b o )

[0134]

[0135] a t =Attention(h t -1,x t )

[0136]

[0137] Among them, f t 、it , o t They are forget gate, input gate and output gate respectively, c t is the unit state, h t is the hidden state, a t is the attention weight, x t is the input, σ is the sigmoid function, tanh is the hyperbolic tangent function, W f , W i , W o , W c is the weight matrix, b f 、b i 、b o 、b c is the bias term, is element-by-element multiplication, and Attention is the attention calculation function.

[0138] In practical applications, we use 3 layers of stacked LSTM, and the number of hidden units in each layer is 256. This design can effectively capture long-term dependencies and is particularly suitable for detecting long-term abnormal patterns in APT attacks.

[0139] 6. Use reinforcement learning to adaptively sample the time series anomaly features to obtain optimized anomaly features. This step is implemented using DuelingDQN:

[0140]

[0141] Among them, Q(s,α) is the state-action value function, V(s) is the state value function, A(s,a) is the advantage function, s is the state, a is the current action, a′ is all possible actions, |A| is the size of the action space, and ∑ represents the sum of all actions. The state space is designed as follows:

[0142] S t =[CPU t ,Mem t ,Net t , A.S. t-1 , A.S. t-2 ,…,AS t-k ]

[0143] Among them, CPU t ,Mem t ,Net t Respectively represent the current CPU usage, memory usage and network load, AS t-i represents the anomaly score of the past i time steps. In practice, we use k = 10, that is, consider the anomaly scores of the past 10 time steps, and the action space includes 5 different sampling frequencies: [1s, 5s, 10s, 30s, 60s].

[0144] The reward function is designed as follows:

[0145] R t =w1·Detection_Rate-W2·False_Alarm_Rate-W3·Resource_Usage

[0146] Among them, W1, w2, w3 are weight parameters, and we determine the optimal values ​​through grid search;

[0147] Detection Rate: This is the detection rate, that is, the proportion of APT attacks detected in a given period of time;

[0148] False_Alarm_Rate: This is the false alarm rate, which is the proportion of normal behaviors that are mistakenly identified as APT attacks within a given period of time;

[0149] Resource_Usage: This is the resource usage rate, that is, the ratio of resources (such as computing resources, storage resources, etc.) consumed by the system within a given time. 7. Aggregate multi-party knowledge through federated learning and update the anomaly detection model. This step is implemented using homomorphic encryption and secure multi-party computing:

[0150]

[0151] Among them, θ (t+1) are global model parameters, is the local model parameter of participant i, t represents the number of iterations, N represents the number of participants, HE represents the homomorphic encryption operation, η represents the learning rate (the initial value is set to 0.001 and dynamically adjusted using the Adam optimizer), is the gradient of the loss function, D i is the local data of participant i, N(0,σ 2 ) has a mean of 0 and a variance of σ 2 Gaussian noise.

[0152] In practical applications, we use the Paillier homomorphic encryption scheme with a key length of 2048 bits. 2 It is dynamically adjusted according to the differential privacy budget ∈, and the initial ∈ is set to 0.1.

[0153] 8. Based on the updated anomaly detection model, the optimized anomaly features are analyzed to identify APT hidden channels. In this step, we use Dempster-Shafer evidence theory to fuse multi-source information:

[0154]

[0155] in, m1,m2,……,m n , which is the support of hypothesis A from n sources of evidence, each source of evidence has its own support m i , X i : This is the set of hypotheses supported by evidence source i, m i (X i ) represents the i-th source of evidence for event X i The basic belief assignment of A represents a compound event. In this system, we use network traffic analysis, DNS request analysis, and encrypted communication analysis as the three main sources of evidence.

[0156] The decision threshold is dynamically adjusted using the Youden index based on the ROC curve:

[0157] threshold=argmax t (TPR(t)-FPR(t))

[0158] Among them, TPR is the true positive rate and FPR is the false positive rate. This method can achieve a good balance between detection rate and false alarm rate. t (*) The maximum value operation function.

[0159] 9. Generate APT covert channel detection report based on the recognition results In this last step, the system generates a detailed APT covert channel detection report. We use a sequence-to-sequence model based on the attention mechanism to generate a natural language report.

[0160] The model structure is as follows:

[0161] Encoder: Bidirectional LSTM with self-attention

[0162] Decoder: LSTM with attention over encoded states

[0163] Pointer-generatornetwork: used to process unregistered words

[0164] The loss function is designed as follows:

[0165] L=λ1·L ce +λ2·L coverage +λ3·L abstractive

[0166] Among them, L c e is the cross entropy loss, L c Overage is the coverage loss, L abstractive is the abstract generation loss. λ1, λ2, λ3 are weight parameters, and the optimal value is determined by cross-validation.

[0167] In actual applications, both the encoder and decoder use 3-layer LSTM with 512 hidden units. The attention mechanism uses scaled dot-product attention.

[0168] This design generates detection reports that are both accurate and easy to understand, including key information such as the type of APT covert channel detected, possible attack sources, affected systems, and recommended mitigation measures.

[0169] Please refer to Figure 3 Based on the above method, the present invention also provides a multi-modal anomaly detection APT covert channel identification system. The system includes the following modules:

[0170] Multimodal data acquisition module 1: This module is responsible for collecting data from multiple data sources in the enterprise network. It includes high-performance network probes (such as Gigamon GigaVUE-HC2), DNS server log collectors, and SSL / TLS proxy servers (such as BlueCoat SSL Visibility Appliance). All data is initially aggregated and pre-processed through the enterprise's SIEM system (such as Splunk Enterprise Security).

[0171] Variational Autoencoder Module 2: This module implements the VAE architecture described above to model the normal behavior of each data modality. It uses NVIDIA Tesla V100 GPU for accelerated computing to ensure that it can process large amounts of data in real time.

[0172] Adaptive multi-scale deep coupled autoencoder module 3: This is the core module of the system, which implements the spiral pyramid network (SPN), quantum annealing-based dynamic modal importance evaluation, fractional-order differential-based sparse abnormal feature enhancement, and nonlinear harmonic mean-based adaptive feature fusion described above. The quantum annealing process is implemented using a D-Wave2000Q quantum computer.

[0173] Contrastive Learning Module 4: This module implements the improved InfoNCE loss function for enhancing anomaly sensitivity. It is tightly integrated with the Adaptive Multi-Scale Deeply Coupled Autoencoder module and shares the same NVIDIA Tesla V100 GPU.

[0174] Long Short-Term Memory Network Module 5: This module implements an improved LSTM unit to capture abnormal patterns in timing. It uses NVIDIA Tesla T4 GPU for accelerated computing.

[0175] Reinforcement Learning Module 6: This module implements the Dueling DQN algorithm for adaptive sampling. It uses the Intel Xeon Platinum 8280 processor for computation to ensure real-time performance.

[0176] Federated Learning Module 7: This module implements a federated learning algorithm based on homomorphic encryption and secure multi-party computing. It uses Intel SGX (Software Guard Extensions) technology to protect sensitive data and computing processes.

[0177] APT Covert Channel Identification Module 8: This module implements a multi-source information fusion algorithm based on Dempster-Shafer evidence theory. It uses the Intel Xeon Gold 6248 processor for high-speed computing.

[0178] Report generation module 9: This module implements a sequence-to-sequence model based on the attention mechanism to generate natural language detection reports. It uses NVIDIA Tesla T4 GPU for accelerated computing.

[0179] All these modules are integrated into a high-performance server cluster, using Kubernetes for container orchestration and management to ensure high availability and scalability of the system. The system integrates with the enterprise's existing security infrastructure, such as firewalls, intrusion detection systems, and security information and event management (SIEM) systems, through RESTful APIs.

[0180] In actual deployment, the system is able to process massive amounts of multimodal data in enterprise networks in real time and accurately identify various complex APT covert channels. For example, in an actual detection, the system successfully identified an APT attack that used DNS tunneling technology, which bypassed traditional network security measures by encoding data in DNS requests. The system not only detected abnormal DNS request patterns, but also confirmed the existence of data leaks by analyzing network traffic and encrypted communication data, and generated detailed detection reports in a timely manner, enabling the security team to take quick action to stop the attack.

[0181] In summary, the method and system provided by the present invention greatly improve the accuracy and efficiency of APT covert channel detection through innovative algorithm design and advanced hardware implementation, providing enterprises with a powerful network security protection measure.

[0182] In order to verify the superiority of the present invention, we conducted a series of comparative experiments. The following is a detailed comparison of Example 1 of the present invention and three comparative examples:

[0183] Embodiment 1: Multimodal anomaly detection APT covert channel identification method based on adaptive multi-scale deep coupling autoencoder of the present invention

[0184] Comparative Example 1: Traditional single-modality rule-based APT detection method

[0185] Comparative Example 2: Single-modality APT detection method based on deep learning

[0186] Comparative Example 3: APT Detection Method Based on Simple Multimodal Fusion

[0187] Test indicators and detection methods:

[0188] 1. Detection accuracy: the number of correctly identified samples divided by the total number of samples.

[0189] 2. False Positive Rate (FPR): The number of normal samples that are misjudged as APT attacks divided by the total number of normal samples.

[0190] 3. False Negative Rate (FNR): The number of undetected APT attack samples divided by the total number of APT attack samples.

[0191] 4.F1 score: The harmonic mean of precision and recall, which comprehensively measures the detection performance.

[0192] 5. Detection delay: The average time from the start of an APT attack to its detection by the system.

[0193] 6. Computational resource consumption: average values ​​of CPU usage, memory usage, and GPU utilization.

[0194] 7. Model adaptability: Performance stability under different network environments, expressed as performance variance.

[0195] Test environment:

[0196] We conducted a 30-day test in a simulated large-scale enterprise network environment. The environment contains 100 workstations, 10 servers, and 5 network switches. We simulated various types of APT attacks, including DNS tunneling, encrypted C&C communications, data exfiltration, etc. At the same time, we also simulated normal network activities to test the system's false positive rate.

[0197] The test results are shown in the following table:

[0198] index Example 1 Comparative Example 1 Comparative Example 2 Comparative Example 3 Detection accuracy 97.8% 82.3% 89.5% 93.2% False Positive Rate 1.2% 8.7% 5.3% 3.8% False negative rate 2.1% 12.5% 7.8% 5.6% F1 score 0.976 0.845 0.912 0.945 Detection delay 2.3 hours 18.7 hours 9.5 hours 6.2 hours CPU usage 45% 25% 60% 55% Memory usage 16 GB 4GB 24GB 20GB GPU Utilization 70% N / A 85% 80% Model fitness (variance) 0.015 0.089 0.042 0.031

[0199] According to the test results in the above table, it can be seen that Example 1 is the best example. The above results are analyzed and explained as follows:

[0200] 1. Detection accuracy: The method of the present invention achieves a high accuracy of 97.8%, which is significantly better than other methods. This is mainly due to the powerful feature extraction and fusion capabilities of the adaptive multi-scale deep coupling autoencoder and the comprehensive utilization of multimodal data.

[0201] 2. False alarm rate and missed alarm rate: The method of the present invention performs well in both indicators, with a false alarm rate of only 1.2% and a missed alarm rate of 2.1%. This shows that the method can effectively distinguish normal behavior from APT attacks and reduce the risk of false alarms and missed alarms.

[0202] 3. F1 score: The high F1 score of 0.976 further demonstrates the balanced performance of the proposed method in terms of precision and recall, which is crucial for APT detection in practical applications.

[0203] 4. Detection delay: The average detection delay of the method of the present invention is only 2.3 hours, which is much lower than other methods. This is due to the adaptive sampling strategy and real-time processing capabilities, which enable the system to identify potential APT attacks more quickly.

[0204] 5. Computational resource consumption: Although the resource consumption of the method of the present invention is slightly higher than that of the rule-based method (Comparative Example 1), this is acceptable considering the performance improvement. Compared with other deep learning methods, the resource utilization of the present invention is more efficient.

[0205] 6. Model adaptability: The proposed method exhibits the best adaptability, with a performance variance of only 0.015. This shows that the proposed method can maintain stable performance in different network environments and attack modes, which is crucial for dealing with ever-changing APT threats.

[0206] The method of the present invention is significantly superior to the comparative method in all key indicators. In particular, the present invention shows obvious advantages in core indicators such as detection accuracy, false alarm rate, missed alarm rate and detection delay. These results fully demonstrate the effectiveness of the innovative points of the present invention, such as adaptive multi-scale deep coupling autoencoder, multimodal data fusion, and reinforcement learning adaptive sampling.

[0207] High accuracy and low false alarm rates indicate that the present invention can more accurately identify APT covert channels and reduce the workload of security teams in dealing with false alarms. Shorter detection delays mean that security teams can respond to potential threats more quickly and minimize the losses caused by attacks.

[0208] In addition, the excellent adaptability of the method of the present invention shows that it can effectively cope with the evolving APT attack techniques, which is crucial for long-term network security. Although the computing resource consumption of this method is relatively high, this is a completely acceptable trade-off considering the significant performance improvement it brings.

[0209] In general, these test results fully demonstrate the innovation and practicality of the present invention in the field of APT covert channel detection, and provide enterprises and organizations with a more effective and reliable network security protection solution.

[0210] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-modal anomaly detection APT covert channel identification method, characterized in that: The following steps are involved: Obtaining multimodal data, and based on the multimodal data, implementing normal behavior modeling through a variational autoencoder to obtain a normal behavior model; Using an adaptive multi-scale deep coupled autoencoder to perform joint representation learning on the multimodal data to obtain a joint representation; Based on the joint representation, enhancing abnormal sensitivity through contrastive learning to obtain enhanced feature representation; Using a long short-term memory network to process the enhanced feature representation, capture the time series abnormal pattern, and obtain the time series abnormal feature; Adaptively sampling the time series abnormal features using reinforcement learning to obtain optimized abnormal features; Aggregate multi-party knowledge through federated learning and update the anomaly detection model; Based on the updated anomaly detection model, the optimized anomaly features are analyzed to identify the APT covert channel; And based on the identification results, generate an APT covert channel detection report.

2. According to the APT covert channel identification method of multimodal anomaly detection according to claim 1, it is characterized in that: The adaptive multi-scale deep coupled autoencoder comprises: Multi-scale feature extraction module, used to extract multi-scale features of multi-modal data; Dynamic modal importance assessment module, used to assess the importance of different modes; Sparse abnormal feature enhancement module, used to enhance potential abnormal features; Adaptive feature fusion module, used to fuse multimodal features.

3. According to the APT covert channel identification method of multimodal anomaly detection according to claim 2, it is characterized in that: The multi-scale feature extraction module is implemented using a spiral pyramid network, where: F i =SPN(X i )=Concat[S1(X i ),S2(X i ),...,S K (X i )] Among them, F i is the extracted feature, X i is the input data, S k is the k-th layer spiral convolution operation, σ is the Mish activation function, W k is the convolution kernel, R θ is the rotation transformation, K is the total number of layers, D k It is a dilated convolution with a dilation rate of 2 k , Represents element-wise addition.

4. According to the APT covert channel identification method of multimodal anomaly detection according to claim 3, it is characterized in that: The dynamic modal importance evaluation module is implemented based on quantum annealing, wherein: α i =QA(Φ(F i )) Φ(x)=cos(πx)+i·sin(πx) Among them, α i is the importance weight of mode i, QA represents the quantum annealing process, Φ is the complex mapping function, F i is the feature extracted by the multi-scale feature extraction module, and i is an imaginary unit.

5. According to the APT covert channel identification method of multimodal anomaly detection according to claim 4, it is characterized in that: The sparse anomaly feature enhancement module is implemented based on fractional differentials, where: E i =F i +λ·D α [SparseMask(F i )] Among them, E i is the enhanced feature, F i is the feature extracted by the multi-scale feature extraction module, λ is a learnable parameter, and D α represents the α-order fractional differential operator, α is the fractional order, Γ is the gamma function, SparseMask is the differentiable sparse mask based on Gumbel-Softmax, and t and τ are time variables.

6. The APT covert channel identification method for multimodal anomaly detection according to claim 5 is characterized in that: The adaptive feature fusion module is implemented by nonlinear harmonic averaging, where: Among them, Z is the final joint representation, α i is the modal importance weight, E i is the enhanced feature, p is a learnable parameter, and ∑ represents the sum of all modes.

7. The APT covert channel identification method for multimodal anomaly detection according to claim 1 is characterized in that: The long short-term memory network adopts an improved LSTM unit, including an attention mechanism and a residual connection, wherein: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) the t =σ(W o ·[h t-1 ,x t ]+b o ) a t =Attention(h t -1,x t ) Among them, f t 、i t , o t They are forget gate, input gate and output gate respectively, c t is the unit state, h t is the hidden state, a t is the attention weight, x t is the input, σ is the sigmoid function, tanh is the hyperbolic tangent function, W f , W i ,W o , W c is the weight matrix, b f , b i , b o , b c is the bias term, is element-by-element multiplication, and Attention is the attention calculation function.

8. The APT covert channel identification method for multimodal anomaly detection according to claim 1 is characterized in that: The reinforcement learning adaptive sampling is implemented using DuelingDQN, where: Among them, Q(s,a) is the state-action value function, V(s) is the state value function, A(s,a) is the advantage function, s is the state, a is the current action, a is all possible actions, |A| is the size of the action space, and ∑ represents the sum of all actions.

9. The method according to claim 1, characterized in that: The federated learning multi-party knowledge aggregation is implemented by homomorphic encryption and secure multi-party computing, where: Among them, θ (t+1) is the global model parameter, is the local model parameter of participant i, t represents the number of iterations, N represents the number of participants, HE represents the homomorphic encryption operation, η represents the learning rate, is the gradient of the loss function, D i is the local data of participant i, N(0,σ 2 ) has a mean of 0 and a variance of σ 2 Gaussian noise.

10. Multimodal anomaly detection APT covert channel identification system, characterized by: include: A multimodal data acquisition module, used for acquiring multimodal data; A variational autoencoder module, used for implementing normal behavior modeling based on the multimodal data to obtain a normal behavior model; An adaptive multi-scale deep coupled autoencoder module, used for performing joint representation learning on the multimodal data to obtain a joint representation; A contrastive learning module, used for enhancing abnormal sensitivity based on the joint representation to obtain an enhanced feature representation; A long short-term memory network module is used to process the enhanced feature representation, capture the time series abnormal pattern, and obtain the time series abnormal feature; A reinforcement learning module, used for adaptively sampling the time series abnormal features to obtain optimized abnormal features; Federated learning module, used to aggregate multi-party knowledge and update anomaly detection models; An APT covert channel identification module, used to analyze the optimized abnormal features based on the updated abnormal detection model to identify the APT covert channel; and The report generation module is used to generate an APT covert channel detection report based on the identification results.

Citation Information

Patent Citations

  • APT attack detection method based on deep belief network-support vector data description

    CN108848068A

  • APT attack-oriented network security threat hidden target identification method

    CN112291260A

  • Threat detection method, device and equipment based on APT attack graph

    CN115484062A

  • APT attack detection method based on threat feature fusion and meta learning

    CN116248367A

  • Network attack detection method and device, storage medium and electronic equipment

    CN116827595A

Cited By

  • Petroleum pipeline abnormity early warning and monitoring method and system based on multi-source data fusion attention mechanism

    CN120067935A

  • Network space surveying and mapping threat detection method and system based on causal association privacy protection

    CN120337301A

  • Network security early warning method and system based on artificial intelligence

    CN120342671A

  • Computer network information security monitoring method, system, equipment and medium

    CN120498736A

  • Cluster collection communication system applied to AI data stream processing

    CN120499701A