A high-dimensional sensing data dimension reduction method in a multi-hop delay-sensitive network

Abstract

The application relates to a high-dimensional perception data dimension reduction method in a multi-hop time delay sensitive network, which comprises the following steps: taking a certain specific time window as a benchmark, reading data from each device to form a high-dimensional data set; for the high-dimensional data of each device, a new feature set is formed; a PCA algorithm is used to reduce the dimension of the selected features; an adaptive feature selection (AFS) algorithm is used to further optimize the features after dimension reduction; and the feature data after adaptive feature selection is used for network transmission, so that the data dimension is reduced while the key features of the data are maintained, the transmission time delay and the network load are reduced, the PCA and the AFS are comprehensively used, the advantages of the PCA and the AFS in dimension reduction and feature selection are fully exerted, the data after dimension reduction retains important information and reduces the data volume, and an effective solution is provided for high-dimensional perception data transmission in a multi-hop time delay sensitive network.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular to a method for dimensionality reduction of high-dimensional sensing data in multi-hop delay-sensitive networks. Background Technology

[0002] With the widespread application of multi-hop delay-sensitive networks, the data generated by sensor devices often has high-dimensionality characteristics. In these networks, the high dimensionality of the data leads to latency issues in data transmission and processing, and also increases the energy consumption of the devices.

[0003] Traditional data dimensionality reduction methods, such as PCA, can reduce the dimensionality of data, but in multi-hop latency-sensitive networks, the real-time characteristics and performance metrics of devices are often ignored. Therefore, existing methods cannot provide efficient dimensionality reduction that meets network performance and real-time requirements.

[0004] Sensor devices in multi-hop latency-sensitive networks transmit data via multiple hops, and this data is typically generated in a high-dimensional format. Network performance is highly sensitive to latency, thus requiring efficient data dimensionality reduction methods to meet the network's real-time requirements.

[0005] Currently, data dimensionality reduction methods typically focus on reducing the complexity of data transmission and processing to decrease latency and energy consumption. However, traditional methods fail to fully consider the real-time operating characteristics and performance metrics of devices, which is particularly important in multi-hop latency-sensitive networks. Summary of the Invention

[0006] The purpose of this application is to provide a method for dimensionality reduction of high-dimensional sensing data in multi-hop delay-sensitive networks, so as to meet the dimensionality reduction requirements of high-dimensional sensing data in multi-hop delay-sensitive networks and improve network performance and real-time performance.

[0007] To achieve the above objectives, this application provides the following technical solution:

[0008] This application provides a method for dimensionality reduction of high-dimensional sensing data in multi-hop delay-sensitive networks, including the following steps:

[0009] Step 1: In a multi-hop delay-sensitive network, each device continuously generates multidimensional data. Using a specific time window T as a reference, data is read from each device to form a high-dimensional dataset X = [X1, X2, ..., X...]. n ], where X i This represents the multidimensional data of the i-th device;

[0010] Step 2: For the high-dimensional data X of each device i Calculate the information entropy H(X) for each feature dimension. ij Select the k feature dimensions with higher information entropy to form a new feature set F.i ;

[0011] Step 3: Use the PCA algorithm to reduce the dimensionality of the selected features, and reduce the feature set F. i Feature fusion is performed to obtain the fused feature vector V. i ;

[0012] Step 4: Further optimize the dimensionality-reduced features using the Adaptive Feature Selection (AFS) algorithm. The AFS algorithm is used to adaptively select the features after PCA dimensionality reduction, retaining the features that have a greater impact on network performance.

[0013] Step 5: Use the feature data after adaptive feature selection for network transmission. This reduces the data dimensionality while maintaining the key features of the data, thereby reducing transmission latency and network load.

[0014] The process of selecting features in step 2 can be represented by the following formula:

[0015]

[0016] Among them, X ij Let x represent the j-th feature of the i-th device, and let P(x) represent the probability of feature value x.

[0017] Select the top k features with the highest information entropy:

[0018] F i ={X i1 ,X i2 ,…,X ik}, k≤m

[0019] Where m is the dimension of the original data, X ij This represents the j-th feature of the i-th device.

[0020] In step 3, the selected feature set F is first... i Feature fusion is performed. For each device i, its selected features are linearly weighted and fused to obtain the fused feature vector V. i :

[0021]

[0022] Where K is the number of selected features, F ik It is the kth feature, w k The weights of the features are determined by the information entropy or importance of the features. Then, the PCA algorithm is used to process the fused feature vector V. i After dimensionality reduction, the mathematical expression for PCA is:

[0023] Z i=V i ×U

[0024] Among them, Z i Let U be the dimensionality-reduced feature vector, and U be the dimensionality-reducing mapping matrix of PCA, which reduces the feature dimension to d dimensions by preserving the principal components.

[0025] The objective function of the AFS algorithm in step 4 is:

[0026]

[0027] Among them, Corr(Z) ik Z ij ) represents the correlation between the k-th and j-th dimension features, Corr(f k (,Y) represents the correlation between the k-th dimension feature and the network performance indicator data transmission latency Y. The objective function J(F) is optimized using the particle swarm optimization algorithm. i This yields the results of adaptive feature selection, which retains features that have a greater impact on network performance.

[0028] Compared with existing technologies, the beneficial effects of this application are: optimized utilization of network resources, reduced data storage requirements, reduced consumption of computing resources, and improved overall network performance. AFS is not simply about removing features; it considers the relationships between features, thus performing better in high-dimensional data and better adapting to different types of data. By reducing data dimensionality, it alleviates the network transmission burden and improves data transmission efficiency. Combining PCA and AFS methods can both preserve the main features of the dataset (an advantage of PCA) and select the most important features according to actual needs (an advantage of AFS), thereby better preserving key information. It achieves good results in processing various types of high-dimensional data and has broad application value. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a flowchart of the method in this application. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. It should be noted that similar reference numerals and letters in the following drawings indicate similar items; therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0032] The terms “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0033] The following is combined with Figure 1 The present invention specifically describes a method for combined prediction of time series data of data center equipment operation, comprising the following specific steps:

[0034] Step 1: Initialize the time window length to T, read data from each device, and form a high-dimensional dataset X = [X1, X2, ..., X...]. n The data includes transmission latency, data packet size, data frequency, etc.

[0035] Step 2: For the high-dimensional data X of each device i Calculate the information entropy of each feature dimension. Select the k features with the highest information entropy to form a new feature set:

[0036] F i ={X i1 ,X i2 ,…,X ik}, k≤m;

[0037] Where m is the dimension of the original data, X ij This represents the j-th feature of the i-th device.

[0038] Step 3: Perform feature fusion on the feature set. For each device i, perform linear weighted fusion on its selected features to obtain the fused feature vector V. i :

[0039]

[0040] Where K is the number of selected features, F ik It is the kth feature, w kThe weights of the features are determined using their information entropy. Information entropy measures the contribution of each feature to the target variable, thus determining the feature weights. First, the information gain of each feature is calculated. Information gain represents the degree to which the uncertainty of the target variable is reduced after the feature is introduced. The formula for calculating information gain is as follows:

[0041]

[0042] Where IG(X,F) is the information gain of feature F with respect to dataset X, H(X) is the information entropy of dataset X, V is the number of values ​​of feature F, and X is the number of values ​​of feature F. v It is a subset of the data when the feature F takes the value v, |X v |Is X v The number of samples in the sample.

[0043] With K features selected, the information gain IG(X,F) can be calculated for each feature. i Then, the information gain is standardized to the weight w. i The calculation formula is as follows:

[0044]

[0045] Among them, F k It is the kth feature, w k This represents the weight of the k-th feature.

[0046] Using the PCA algorithm to analyze the fused feature vector V i Dimensionality reduction is performed. The mathematical expression for PCA is:

[0047] Z i =V i ×U

[0048] Among them, Z i Let U be the dimensionality-reduced feature vector, and U be the dimensionality-reducing mapping matrix of PCA, which reduces the feature dimension to d dimensions by preserving the principal components.

[0049] Step 4: Further optimize the dimensionality-reduced features using the Adaptive Feature Selection (AFS) algorithm. The AFS algorithm adaptively selects features from those reduced by PCA, retaining those with the greatest impact on network performance. The objective function of the AFS algorithm is:

[0050]

[0051] Among them, Corr(Z) ik Z ij ) represents the correlation between the k-th and j-th dimension features, Corr(f k(,Y) represents the correlation between the k-th dimension feature and the network performance indicator data transmission latency Y. The objective function J(F) is optimized using the particle swarm optimization algorithm. i This yields the results of adaptive feature selection, which retains features that have a greater impact on network performance.

[0052] Step 5: Use the feature data after adaptive feature selection for network transmission. This reduces the data dimensionality while maintaining the key features of the data, thereby reducing transmission latency and network load.

[0053] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for dimensionality reduction of high-dimensional sensing data in multi-hop delay-sensitive networks, characterized in that, Includes the following steps: Step 1: In a multi-hop delay-sensitive network, each device continuously generates multidimensional data. Using a specific time window T as a reference, data is read from each device to form a high-dimensional dataset X = [X1, X2, ..., X...]. n ], where X i This represents the multidimensional data of the i-th device; Step 2: For the high-dimensional data X of each device i Calculate the information entropy H(X) for each feature dimension. ij Select the k feature dimensions with higher information entropy to form a new feature set F. i ; Step 3: Use the PCA algorithm to reduce the dimensionality of the selected features, and reduce the feature set F. i Feature fusion is performed to obtain the fused feature vector V. i ; Step 4: Further optimize the dimensionality-reduced features using the Adaptive Feature Selection (AFS) algorithm. The AFS algorithm is used to adaptively select the features after PCA dimensionality reduction, retaining the features that have a greater impact on network performance. Step 5: Use the feature data after adaptive feature selection for network transmission. This reduces the data dimensionality while maintaining the key features of the data, thereby reducing transmission latency and network load.

2. The method for dimensionality reduction of high-dimensional sensing data in a multi-hop delay-sensitive network according to claim 1, characterized in that, The process of selecting features in step 2 can be represented by the following formula: Among them, X ij Let P(x) represent the j-th feature of the i-th device, and let P(x) represent the probability of feature value x. Select the k features with higher information entropy: F i ={X i1 ,X i2 ,…,X ik },k≤m Where m is the dimension of the original data, X ij This represents the j-th feature of the i-th device.

3. The method for dimensionality reduction of high-dimensional sensing data in a multi-hop delay-sensitive network according to claim 1, characterized in that, In step 3, the selected feature set F is first... i Feature fusion is performed. For each device i, its selected features are linearly weighted and fused to obtain the fused feature vector V. i : Where K is the number of selected features, F ik It is the kth feature, w k The weights of the features are determined by the information entropy or importance of the features. Then, the PCA algorithm is used to process the fused feature vector V. i After dimensionality reduction, the mathematical expression for PCA is: With i =V i ×U Among them, Z i Let U be the dimensionality-reduced feature vector, and U be the dimensionality-reducing mapping matrix of PCA, which reduces the feature dimension to d dimensions by preserving the principal components.

4. The method for dimensionality reduction of high-dimensional sensing data in a multi-hop delay-sensitive network according to claim 1, characterized in that, The objective function of the AFS algorithm in step 4 is: Among them, Corr(Z) ik Z ij ) represents the correlation between the k-th and j-th dimension features, Corr(f k (,Y) represents the correlation between the k-th dimension feature and the network performance indicator data transmission latency Y. The objective function J(F) is optimized using the particle swarm optimization algorithm. i This yields the results of adaptive feature selection, which retains features that have a greater impact on network performance.

Images