Smartphone authentication enhancement method, system and product based on hand vibration response
By collecting and analyzing the propagation of user hand vibration signals and muscle response characteristics, and using twin neural networks for identity verification, the system solves the security deficiencies of existing possession authentication methods when devices are lost or stolen, achieving enhanced authentication with high security and convenience.
Patent Information
- Application Number
- CN202511685908.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-18
- Publication Date
- 2026-02-03
- Estimated Expiration
- 2045-11-18
AI Technical Summary
Existing possession-based authentication methods lack further verification of the operator's identity when the device is lost or stolen, posing a high security risk and making user privacy data vulnerable to unauthorized access.
The system incorporates individualized physiological response signals from the user's hand as an auxiliary authentication basis. It collects hand vibration signals through a smart terminal, extracts vibration propagation features and muscle response features, and uses a twin neural network to determine identity.
Without altering the original authentication process and user experience, the system effectively enhances its resistance to attacks, prevents unauthorized users from bypassing identity authentication, and improves the accuracy and security of identity verification.
Smart Images
Figure CN121145189B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of information security technology and relates to a method, system and product for enhancing authentication of smart terminals. Specifically, it relates to a method, system and product for enhancing smartphone authentication based on hand vibration response, which is applicable to application needs to improve authentication security in scenarios where authentication is based on held objects (such as smartwatches). Background Technology
[0002] With the rapid development of mobile internet, smartphones have become indispensable devices in people's daily lives, widely used in communication, payment, health management, and other scenarios, carrying a large amount of personal privacy data, such as communication records, financial information, and health data. To prevent unauthorized access to sensitive information, smartphones are generally equipped with identity authentication mechanisms to ensure the security and privacy of user data.
[0003] Current mainstream identity authentication mechanisms can be divided into three categories: knowledge-based authentication, such as passwords; biometric-based authentication, such as fingerprint recognition and facial recognition; and object-based authentication, such as device-assisted unlocking via smartwatches. Among these, object-based authentication has been widely adopted in smartphone operating systems in recent years due to its ease of use and lack of user intervention; examples include Apple Watch unlocking iPhones and Android devices automatically unlocking via smartwatches. This method uses communication protocols such as Bluetooth to determine the proximity of the paired device, thereby determining the device owner's presence and automatically completing identity authentication.
[0004] However, this authentication mechanism has security vulnerabilities. Its core assumption is that "possessing the device automatically makes one a legitimate user." This means that if the authenticated device (such as a watch) is stolen or lost, the system will still assume the current operator is the device owner, allowing bypassing the authentication process. This single-factor authentication method, relying on the existence of the device, lacks further verification of the operator's identity, making it vulnerable to unauthorized access to user privacy data and posing a high security risk. Therefore, there is an urgent need for a technical solution that enhances the security of object-based identity authentication without affecting the original user experience and convenience, effectively improving the authentication system's resistance to attacks and the accuracy of identity verification. Summary of the Invention
[0005] The purpose of this invention is to propose a smartphone authentication enhancement method, system, and product based on hand vibration response. Addressing the security shortcomings of existing authentication methods based on possession (such as smartwatches) in situations where the device is lost or stolen, this invention proposes introducing individualized physiological response signals from the user's hand as auxiliary authentication criteria. This allows for implicit identification of the device holder's identity without altering the original authentication process or user experience, thereby effectively improving the authentication system's resistance to attacks and preventing unauthorized users from bypassing identity authentication by using the possessed device.
[0006] The technical solution adopted by the method of the present invention is: a smartphone authentication enhancement method based on hand vibration response, comprising the following steps:
[0007] Step 1: The user wears a smart terminal and holds a smartphone. When the user raises their hand, it triggers the smart terminal to generate a vibration signal. The smartphone collects the user's hand's response to the vibration signal and forms raw vibration response data.
[0008] Step 2: Preprocess the raw vibration response data to extract valid response data segments;
[0009] Step 3: Extract two types of features representing the user's hand vibration response from the preprocessed data, namely vibration propagation features and muscle response features, and merge the two types of features to obtain fused features;
[0010] Step 4: Input the fused features into the authentication model and determine whether the current operator is a legitimate user by comparison.
[0011] Preferably, in step 1, the raw vibration response data includes triaxial accelerometer data and triaxial gyroscope data.
[0012] As a preferred embodiment, step 2 includes the following sub-steps:
[0013] Step 2.1: Denoise the data from the triaxial accelerometer and triaxial gyroscope respectively, and calculate the net acceleration and net angular velocity;
[0014] Step 2.2: Perform vibration event detection and identify the signal time period containing the effective vibration response.
[0015] Preferably, in step 2.1, a Butterworth high-pass filter is first used to eliminate low-frequency interference caused by natural human movement; then the triaxial accelerometer data is processed. and three-axis gyroscope data Process the data and calculate the net acceleration separately. and net angular velocity ;
[0016] ;
[0017] .
[0018] in , , This represents the data collected by the triaxial accelerometer at time t. , , Axis data, , , This represents the data collected by the three-axis gyroscope at time t. , , Axis data.
[0019] Preferably, in step 2.2, the start and end points of the vibration event are determined by detecting high-frequency energy changes, resulting in an accelerometer and angular velocity response data segment containing N sample points, where N is a preset value.
[0020] Preferably, in step 3, vibration propagation features are extracted to reflect the propagation and attenuation patterns of vibration signals in the user's hand tissue. This involves applying a bandpass filter to the net acceleration and net angular velocity signals to extract high-frequency components, followed by extracting the Mel frequency cepstral coefficients (MFCCs) from the high-frequency components.
[0021] Preferably, in step 3, muscle response features are extracted to describe the contraction-relaxation feedback of hand muscles under vibration stimulation. This involves applying a bandpass filter to the net acceleration and net angular velocity signals to extract low-frequency components and calculating the cumulative distribution function (CDF). Based on the CDF, the signal is divided into 10 sub-bands, and 10 frequency domain statistical features are extracted for each sub-band, including peak frequency, peak-to-peak ratio, main band energy ratio, band center, band spread, band skewness, band kurtosis, band entropy, average power, and power spectral density.
[0022] Preferably, the authentication model is a Siamese neural network, comprising two subnetworks sharing weights;
[0023] Each sub-network consists of sequentially connected Conv2D+BN+ReLU layers, Conv2D+BN+ReLU layers, Pooling+Dropout layers, Flatten layers, FC+ReLU layers, and FC+Linear layers; where Conv2D is a two-dimensional convolution operation, BN is a batch normalization operation, ReLU is a linear rectified activation function, Pooling is a pooling operation, FC is a fully connected operation, and Linear is a linear activation function.
[0024] During authentication, the fused feature vector is input into a pre-trained Siamese neural network, and an embedding vector is generated through any sub-network. The similarity between the embedding vector and the pre-registered template vector is calculated using Euclidean distance. If the similarity distance is lower than a set threshold, the user is determined to be legitimate; otherwise, the user is considered an attacker.
[0025] The technical solution adopted by the system of the present invention is: a smartphone authentication enhancement system based on hand vibration response, comprising:
[0026] One or more processors;
[0027] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the smartphone authentication enhancement method based on hand vibration response.
[0028] The technical solution adopted by the product of the present invention is: a smartphone authentication enhancement product based on hand vibration response, including computer program instructions, which, when the computer program instructions are run on a computer, cause the computer to execute the smartphone authentication enhancement method based on hand vibration response.
[0029] Compared with the prior art, the beneficial effects of the present invention include:
[0030] (1) Enhance authentication security: Introduce the dimension of hand physiological characteristics on the basis of traditional authentication mechanism based on holding device, effectively preventing the security risk of bypassing identity authentication due to theft or loss of device.
[0031] (2) Seamless authentication experience: The authentication process is seamless for the user and is completed automatically without any additional operation, maintaining the original convenience of authentication.
[0032] (3) No new hardware required: Make full use of the vibrators and sensors built into existing smartwatches and smartphones to achieve low-cost deployment.
[0033] (4) High adaptability: It achieves high-accuracy user identification through multi-dimensional fusion features and twin network structure, and has good robustness. Attached Figure Description
[0034] The technical solutions of the present invention will be further illustrated below using embodiments and specific implementation methods. In addition, some accompanying drawings are used in the description of the technical solutions. Those skilled in the art can obtain other drawings and the intent of the present invention from these drawings without any creative effort.
[0035] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;
[0036] Figure 2 This is a structural diagram of the twin neural network model according to an embodiment of the present invention. Detailed Implementation
[0037] To facilitate understanding and implementation of the present invention by those skilled in the art, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0038] Please see Figure 1 This embodiment provides a smartphone authentication enhancement method based on hand vibration response, including the following steps:
[0039] Step 1: When a user wears a smart device (such as a smartwatch, smart bracelet, etc.) and holds a smartphone, the user raises their hand to trigger the smartwatch to generate a specific vibration signal. The smartphone collects the user's hand response to the vibration signal through its built-in motion sensors (including accelerometers and gyroscopes) to form raw vibration response data.
[0040] In one implementation, a user wears a smartwatch and holds a smartphone. Raising their hand triggers the smartwatch to generate a vibration signal lasting one second with a frequency range of 200Hz to 250Hz. This signal is transmitted to the smartphone through the muscles, bones, and skin of the user's hand. During this process, the smartphone, using its built-in three-axis accelerometer and three-axis gyroscope, synchronously collects the user's hand's response data to the vibration signal at a sampling rate of 500Hz, recording it as the raw hand vibration response signal.
[0041] Step 2: Preprocess the acquired raw vibration response signal, including signal denoising, calculation of net acceleration and net angular velocity, and vibration event detection, in order to extract effective response data segments;
[0042] In one implementation, step 2 specifically includes the following sub-steps:
[0043] Step 2.1: Denoise the accelerometer data and gyroscope data respectively, and calculate the net acceleration and net angular velocity;
[0044] In one implementation, the denoising process employs a Butterworth high-pass filter with a cutoff frequency of 30 Hz to eliminate low-frequency interference caused by natural human movement; then the triaxial accelerometer data is processed. = ( , , ) and three-axis gyroscope data = ( , , The net acceleration is calculated separately. and net angular velocity ;
[0045] ;
[0046] .
[0047] in , , This represents the data collected by the triaxial accelerometer at time t. , , Axis data, , , This represents the data collected by the three-axis gyroscope at time t. , , Axis data.
[0048] Step 2.2: Perform vibration event detection and identify data segments containing valid vibration responses.
[0049] In one implementation, because the energy of the acquired signal in the high-frequency range (200-250 Hz) is significantly enhanced during vibration, while the energy in this frequency band is close to background noise in the non-vibration state, the start and end points of the vibration event can be determined by the changes in high-frequency energy. This embodiment uses a bandpass filter to extract the high-frequency components (200Hz–250Hz) of net acceleration and net angular velocity, and calculates their energy envelope. Based on the significant changes in the envelope energy, the start and end times of the vibration event are detected, and effective acceleration and angular velocity response data segments containing 500 sampling points are extracted for subsequent feature extraction processing.
[0050] Step 3: Extract two types of features representing the user's hand vibration response from the preprocessed data, namely vibration propagation features and muscle response features, and merge the two types of features to obtain fused features.
[0051] In one implementation, vibration propagation characteristics are used to reflect the propagation and attenuation patterns of vibration signals in the user's hand tissue, primarily manifested in the high-frequency portion of the response signal. This embodiment first applies a bandpass filter to the net acceleration and net angular velocity signals to extract the high-frequency components in the 200-250 Hz band. Subsequently, Mel frequency cepstral coefficients (MFCCs) are extracted from the high-frequency signals. Specifically, each signal is divided into 20 frames with a 50% overlap between frames and a Hamming window is applied. A Fast Fourier Transform (FFT) is performed on each frame, and a Mel filter bank is applied for bandwidth compression. The logarithm of the Mel spectrum is taken, followed by a Discrete Cosine Transform (DCT), to extract the first 13 dimensions of the MFCC coefficients. Finally, two 13×20 feature matrices are obtained, representing the vibration propagation characteristics of net acceleration and net angular velocity, respectively.
[0052] In one implementation, muscle response features are used to describe the contraction-relaxation feedback of hand muscles under vibration stimulation, mainly reflected in the low-frequency components of the response signal. This embodiment employs a frequency domain statistical analysis method based on the cumulative distribution function (CDF) to extract low-frequency features. First, a bandpass filter is applied to the net acceleration and net angular velocity signals to extract the low-frequency components in the 30-100 Hz frequency band, and the cumulative distribution function (CDF) is calculated. Based on the CDF, the signal is divided into 10 sub-bands, and 10 frequency domain statistical features are extracted for each sub-band, including peak frequency, peak-to-peak ratio, main band energy ratio, band center, band spread, band skewness, band kurtosis, band entropy, average power, and power spectral density (see Table 1). Finally, two sets of 10×10 feature matrices are obtained, representing the muscle response features of net acceleration and net angular velocity, respectively.
[0053] In this embodiment, the two types of features are concatenated to obtain a 2×18×20 fused feature matrix.
[0054] Table 1
[0055]
[0056] Step 4: Input the fused features into the authentication model, and compare them to determine whether the current operator is a legitimate user, thereby enhancing the security of the item authentication.
[0057] In one implementation, the authentication model is a Siamese neural network, comprising two sub-networks sharing weights. The sub-networks, see [link to implementation details]. Figure 2As shown in Table 2, each sub-network consists of sequentially connected Conv2D+BN+ReLU layers, Conv2D+BN+ReLU layers, Pooling+Dropout layers, Flatten layers, FC+ReLU layers, and FC+Linear layers; where Conv2D is a two-dimensional convolution operation, BN is a batch normalization operation, ReLU is a linear rectified activation function, Pooling is a pooling operation, FC is a fully connected operation, and Linear is a linear activation function.
[0058] Table 2
[0059]
[0060] The pre-trained Siamese neural network is trained by inputting sample pairs into subnet 1 and subnet 2 respectively. It outputs two fixed-size embedding vectors, and further calculates their distance to quantify their similarity, thereby reducing the distance between positive sample pairs and increasing the distance between negative sample pairs. Training stops when the conditions are met, resulting in a trained Siamese neural network.
[0061] During authentication, the fused feature vector is input into a pre-trained Siamese neural network, and an embedding vector is generated through any sub-network. The similarity between the embedding vector and the pre-registered template vector is calculated using Euclidean distance. If the similarity distance is lower than a set threshold, the user is determined to be legitimate; otherwise, the user is considered an attacker.
[0062] This embodiment also provides a smartphone authentication enhancement system based on hand vibration response, including:
[0063] One or more processors;
[0064] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the smartphone authentication enhancement method based on hand vibration response.
[0065] This embodiment also provides a smartphone authentication enhancement product based on hand vibration response, including computer program instructions that, when the computer program instructions are run on a computer, cause the computer to execute the smartphone authentication enhancement method based on hand vibration response.
[0066] The following specific experiments further illustrate the solution of the present invention.
[0067] To verify the performance of this invention, tests were conducted on the Redmi K20 smartphone and the TicWatch Pro smartwatch. Subjects wore the TicWatch Pro smartwatch and held the Redmi K20 smartphone. Raising the hand triggered the smartwatch to actively emit a 1-second vibration signal. The Redmi K20 smartphone's built-in three-axis accelerometer and three-axis gyroscope collected the user's hand vibration response signal at a frequency of 500 Hz. This embodiment recruited 30 subjects aged 21-28 years to participate in the experiment, including 15 men and 15 women, collecting a total of 3000 samples.
[0068] This invention uses Balanced Accuracy (BAC) as an evaluation metric to measure the overall ability of the scheme to identify positive and negative samples. Experimental results show that this invention can identify legitimate users with an accuracy rate of 97.8%.
[0069] This invention incorporates the individualized physiological response characteristics of a user's hand to vibration signals during the authentication process, achieving implicit verification of the device holder's identity without altering the original authentication procedure. This effectively prevents authentication bypass issues caused by device loss or theft. This invention offers seamless, low-cost, and highly secure authentication enhancement capabilities.
[0070] It should be understood that the embodiments described above are only some, not all, of the embodiments of the present invention. Furthermore, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined to form feasible technical solutions. Such combinations are not constrained by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.
[0071] It should be understood that the above description of the preferred embodiments is quite detailed, but it should not be considered as a limitation on the scope of protection of this invention. Those skilled in the art, under the guidance of this invention, can make substitutions or modifications without departing from the scope of protection of the claims of this invention, and all such substitutions or modifications fall within the scope of protection of this invention. The scope of protection of this invention should be determined by the appended claims.
Claims
1. A smartphone authentication enhancement method based on hand vibration response, characterized in that, Includes the following steps: Step 1: The user wears a smart terminal and holds a smartphone. The user raises their hand to trigger the smart terminal to generate a vibration signal. The smartphone collects the user's hand's response to the vibration signal and forms raw vibration response data. Step 2: Preprocess the raw vibration response data to extract valid response data segments; Step 3: Extract two types of features representing the user's hand vibration response from the preprocessed data, namely vibration propagation features and muscle response features, and merge the two types of features to obtain fused features; Among them, the vibration propagation features are extracted to reflect the propagation and attenuation mode of the vibration signal in the user's hand tissue. This involves applying a bandpass filter to the net acceleration and net angular velocity signals to extract high-frequency components, and then extracting the Mel frequency cepstral coefficients (MFCCs) from the high-frequency components. Muscle response features are extracted to describe the contraction-relaxation feedback of hand muscles under vibration stimulation. This involves applying a bandpass filter to the net acceleration and net angular velocity signals to extract low-frequency components and calculating the cumulative distribution function (CDF). Based on the CDF, the signal is divided into 10 sub-bands, and 10 frequency domain statistical features are extracted for each sub-band, including peak frequency, peak-to-peak ratio, main band energy ratio, band center, band spread, band skewness, band kurtosis, band entropy, average power, and power spectral density. Step 4: Input the fused features into the authentication model and determine whether the current operator is a legitimate user by comparison.
2. The smartphone authentication enhancement method based on hand vibration response according to claim 1, characterized in that: In step 1, the raw vibration response data includes triaxial accelerometer data and triaxial gyroscope data.
3. The smartphone authentication enhancement method based on hand vibration response according to claim 1, characterized in that, Step 2 includes the following sub-steps: Step 2.1: Denoise the data from the triaxial accelerometer and triaxial gyroscope respectively, and calculate the net acceleration and net angular velocity; Step 2.2: Perform vibration event detection and identify the signal time period containing the effective vibration response.
4. The smartphone authentication enhancement method based on hand vibration response according to claim 3, characterized in that: In step 2.1, firstly, a Butterworth high-pass filter is used to eliminate low-frequency interference caused by natural human movement; then, the triaxial accelerometer data is processed. and three-axis gyroscope data Process the data and calculate the net acceleration separately. and net angular velocity ; ； ； in , , This represents the data collected by the triaxial accelerometer at time t. , , Axis data, , , This represents the data collected by the three-axis gyroscope at time t. , , Axis data.
5. The smartphone authentication enhancement method based on hand vibration response according to claim 3, characterized in that: In step 2.2, the start and end points of the vibration event are determined by detecting high-frequency energy changes, resulting in an accelerometer and angular velocity response data segment containing N sample points, where N is a preset value.
6. The smartphone authentication enhancement method based on hand vibration response according to any one of claims 1-5, characterized in that: The authentication model is a twin neural network, comprising two subnetworks sharing weights; Each sub-network consists of sequentially connected Conv2D+BN+ReLU layers, Conv2D+BN+ReLU layers, Pooling+Dropout layers, Flatten layers, FC+ReLU layers, and FC+Linear layers; where Conv2D is a two-dimensional convolution operation, BN is a batch normalization operation, ReLU is a linear rectified activation function, Pooling is a pooling operation, FC is a fully connected operation, and Linear is a linear activation function. During authentication, the fused feature vector is input into a pre-trained Siamese neural network, and an embedding vector is generated through any sub-network. The similarity between the embedding vector and the pre-registered template vector is calculated using Euclidean distance. If the similarity distance is lower than a set threshold, the user is determined to be legitimate; otherwise, the user is considered an attacker.
7. A smartphone authentication enhancement system based on hand vibration response, characterized in that, include: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the smartphone authentication enhancement method based on hand vibration response as described in any one of claims 1 to 6.
8. A smartphone authentication enhancement product based on hand vibration response, comprising computer program instructions, characterized in that: When the computer program instructions are executed on a computer, the computer performs the smartphone authentication enhancement method based on hand vibration response as described in any one of claims 1 to 6.