A wrist-worn device identity authentication method based on vibration feedback

Abstract

The present invention proposes a wrist-worn device identity authentication method based on vibration feedback. A vibration signal is generated by devices such as smart watches, and the vibration response data of the user is collected in real time using a built-in accelerometer. This method includes noise filtering, direction correction, and normalization preprocessing of the signal. Subsequently, vibration features are extracted through multi-order differences, and a multi-branch convolutional neural network (MCNN) is used for feature extraction and fusion to capture vibration features at different time scales. Finally, identity matching is performed through a Softmax classifier. This method exhibits good stability and recognition accuracy in various wearing postures, hand movement changes, and complex usage scenarios, and has a high anti-interference ability.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent wrist-worn devices and relates to a wrist-worn device identity authentication method based on vibration feedback. Background Art

[0002] With the rapid development of information technology and the increasing popularity of Internet of Things (IoT) technology, devices have become an indispensable part of modern life. In particular, wrist-worn devices such as smartwatches and fitness trackers, with their multifunctionality and portability in real-time health monitoring, message notification, mobile payment, navigation, and sports tracking, have gradually become the core carriers of personal digital life. These devices can monitor real-time health data of users such as heart rate, blood pressure, and sleep quality, provide accurate health management services for users, and greatly improve the quality of life and daily convenience of users by integrating various application functions.

[0003] However, with the continuous expansion and deepening of the functions of wrist-worn devices, their security issues have become increasingly prominent. A large amount of sensitive information is stored and transmitted in wrist-worn devices such as smartwatches, including but not limited to users' health data, real-time location information, payment records, etc. Once this information is maliciously attacked or illegally obtained, it will pose a serious threat to the personal privacy and property security of users. Especially in the field of identity authentication, traditional authentication methods such as passwords and PIN codes are difficult to meet the dual requirements of high security and convenience for modern smart devices due to their disadvantages such as being easy to crack and easy to forget.

[0004] Currently, the closest prior arts mainly focus on two identity authentication methods: biometric recognition and vibration feedback. Biometric recognition technologies, such as fingerprint recognition, facial recognition, and iris scanning, although widely used in various smart devices and providing relatively high security due to their characteristics based on users' unique physiological features, face significant environmental limitations in practical applications. For example, the recognition accuracy of fingerprint recognition will drop significantly when fingers are wet, dirty, or injured; the recognition effect of facial recognition will be significantly weakened when there is insufficient light, the user wears a mask, or facial expressions change. In addition, once biometric data is stolen or misused, it will bring irreversible security risks, and these technologies often rely on specific hardware support, increasing the complexity and manufacturing cost of devices.

[0005] Another type of existing technology, namely the vibration - feedback - based identity authentication method, although attempts to achieve identity authentication by detecting the user's response to vibration stimuli and has achieved certain research results, is still in the exploratory stage, and its recognition accuracy and stability need to be further improved. For example, existing vibration - feedback identity authentication systems may require users to perform specific actions or patterns, which not only increases the user's operation burden but also may affect the convenience of authentication and the user experience. Secondly, existing methods usually rely on the combination of multi - axis vibration data and complex two - factor models. Although identity verification can be achieved, it is easily affected by changes in the wearing position and noise interference in practical applications, and has poor adaptability to natural gestures and daily activities. Summary of the Invention

[0006] To solve the above - mentioned technical problems, this application provides a vibration - feedback - based identity authentication method for wrist - worn devices. Compared with existing methods, the present invention shows obvious advantages in feature extraction, anti - interference ability, and resource consumption.

[0007] To achieve the above - mentioned purpose, this application adopts the following technical solutions:

[0008] A vibration - feedback - based identity authentication method for wrist - worn devices, comprising the following steps:

[0009] Generation and acquisition of vibration signals: Use the built - in vibration motor of the wrist - worn device to generate vibration feedback signals, and use the built - in accelerometer of the wrist - worn device to collect the vibration feedback signals in real - time;

[0010] Signal pre - processing: Noise filtering: Use a Butterworth band - pass filter to filter the noise of the vibration feedback signals, with cut - off frequencies of 10 Hz to 49.8 Hz and filter out the vibration feedback signals outside this range; Direction correction: By correcting the signal directions on the three axes of the accelerometer, ensure that the collected vibration feedback signals are consistent regardless of the user's wearing posture; Normalization processing: Normalize the amplitude range of the vibration signals that have undergone noise filtering and direction correction, aiming to standardize the vibration data of different time periods and different devices, so that they have the same scale and distribution;

[0011] Differential feature extraction: Perform first - order, second - order, fourth - order, and eighth - order differential processing on the pre - processed vibration signals to extract the change features of the vibration signals at different time periods;

[0012] Processing of Multi-branch convolutional neural networks (MCNN): Input the extracted differential features into MCNN; Use convolutional kernels of different sizes to capture short-term and long-term vibration features; Each branch extracts features of different time scales through convolutional layers and uses 64 filters; Use the ReLU activation function for non-linear processing; Convert the convolutional output into a one-dimensional vector through a flattening layer; In the feature fusion stage, use the Concatenate layer to splice the feature vectors from different branches together to form a comprehensive feature vector containing multi-scale features;

[0013] Classification and recognition: Input the comprehensive feature vector generated by the multi-branch convolutional neural network into the classifier through the fully connected layer, use the Softmax function to classify the results, and determine whether the vibration response matches the registered user features; If it matches, the identity authentication is successful; If it does not match, the identity authentication fails.

[0014] In some embodiments, the vibration feedback signal is the signal after the original vibration signal of the vibration motor acts on the wrist.

[0015] In some embodiments, the response data of the vibration feedback signal collected in real time by the accelerometer under the condition of 100 Hz, and the duration of each sampling is 3 seconds.

[0016] In some embodiments, the formula for correcting the signal directions on the three axes of the accelerometer is as follows:

[0017] α′ i =(1-β)(α i -g i ),i=x,y,z#

[0018]

[0019] Where, α i is the original acceleration value, g i is the gravitational acceleration captured by the accelerometer along the i-th axis, and β is the filter factor determined by the time constant t of the filter and the event transfer rate dT.

[0020] In some embodiments, the formula for normalizing the amplitude range of the vibration signal is:

[0021]

[0022] Where, α' i is the single reading of the i-th axis after direction alignment, u i is the mean value of the data along the same axis, and δ iIs the mean standard deviation of the data along the same axis; α″ i Is the accelerometer data.

[0023] A readable storage medium, on which a computing program is stored, and when the program is executed by a processor, it can implement the vibration feedback-based wrist-worn device identity authentication method described in any one of the above.

[0024] An electronic device, including the above-readable storage medium.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] The vibration feedback-based wrist-worn device identity authentication method of the present invention collects the user's vibration signal in real time through a high-precision vibration sensor, and after preprocessing steps such as noise filtering, normalization, and direction correction, extracts distinguishable differential features. Subsequently, the designed MCNN model is used to train and identify the features, realizing high-precision identity authentication.

[0027] The core advantage of the method of the present invention lies in its multi-scale dynamic feature extraction ability, which can effectively cope with signal changes in complex usage scenarios, such as changes in the user's activity state and adjustments to the tightness of the worn device. This not only greatly improves the accuracy and stability of identity recognition, but also enhances the security and reliability of the system, while reducing the dependence on additional hardware, bringing significant progress to the identity authentication technology of wrist-worn devices.

[0028] The method of the present invention can access a pre-collected challenge-response library, where the challenge refers to the vibration stimulus applied by the watch to the user, and the response is the user's vibration collected. Due to the vibration characteristics of the user's arm, each response is unique for each challenge and user.

[0029] The present invention captures the subtle changes in the vibration signal through multi-order differential feature extraction and MCNN, effectively improving the authentication stability and anti-interference ability under different wrist activities and wearing tightness changes. Especially in the case of slight position changes, the present invention can still maintain an authentication accuracy rate of more than 97%, which is suitable for natural activities in daily life and does not require strict wearing conditions.

[0030] The present invention only needs to use three-axis acceleration data, greatly reducing the hardware dependence and consumption of computing resources, and facilitating implementation on common smart watch configurations. This design significantly optimizes the resource usage efficiency, making this method have obvious advantages in terms of convenience, stability, and feasibility of popularization and application.

[0031] The method of the present invention realizes precise identity matching by comparing the user's vibration characteristics with the vibration characteristics of multiple known users pre-registered in the system one by one. Description of the Drawings

[0032] To more clearly illustrate the technical solutions of the embodiments of the present application, the accompanying drawings required for the embodiments will be briefly introduced below. It should be understood that the following accompanying drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related accompanying drawings can also be obtained based on these drawings.

[0033] Figure 1 It is a schematic flow diagram of the identity authentication method;

[0034] Figure 2 It is a diagram of the MCNN structure and parameters;

[0035] Figure 3 It is the accuracy of the authentication model at different sampling times provided in Embodiment 1 of the present invention;

[0036] Figure 4 It is the loss value of the authentication model at different sampling times provided in Embodiment 1 of the present invention;

[0037] Figure 5 It is the Accuracy, Loss, Recall, and F1-score of the authentication model after the device provided in Embodiment 2 of the present invention randomly moves on the original position baseline;

[0038] Figure 6 It is the Accuracy, Loss, Recall, and F1-score of the authentication model after the user's hand makes different habitual actions provided in Embodiment 3 of the present invention. Detailed implementation manners

[0039] In order to make the purpose, technical solutions, design methods, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below through specific embodiments with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0040] In all the examples shown and discussed herein, any specific value should be construed as merely exemplary and not as a limitation. Therefore, other examples of the exemplary embodiments may have different values.

[0041] For technologies, methods, and devices known to those of ordinary skill in the relevant fields, they may not be discussed in detail, but where appropriate, the said technologies, methods, and devices should be regarded as part of the specification.

[0042] Reference Figure 1 - Figure 2, the method for authenticating the identity of a wrist - worn device based on vibration feedback according to the present invention includes the following steps:

[0043] S1. User wears the device: The user wears the smart watch on the wrist. The device is equipped with a built - in vibration motor and an accelerometer. The main function of the accelerometer is to collect the vibration feedback signal of the vibration motor. When the original vibration signal emitted by the vibration motor passes through the user's wrist, this signal will propagate according to the physical characteristics of the wrist, such as bone structure, muscle tissue, and skin. Given that the wrist structure of each person, including bone density and muscle distribution, is unique, the original vibration signal will experience different degrees of absorption, reflection, and attenuation during propagation, thereby forming a vibration feedback signal with individual specificity.

[0044] The present invention can capture these differences in vibration responses caused by wrist structure differences and convert them into data characterizing the unique physical and physiological characteristics of the user's wrist. By analyzing this data, the system can accurately verify the user's identity. To ensure the accuracy of identity verification, the user needs to keep the device in a stable position when wearing it so that the accelerometer can accurately and effectively capture the vibration feedback signal.

[0045] S2. Generation and collection of vibration signals: The device generates a vibration feedback signal through the built - in vibration motor, and at the same time sets the data sampling frequency to about 100Hz. The accelerometer then records the response data of the vibration feedback signal in real - time. The accelerometer details record the acceleration data including the timestamp and the X, Y, and Z axes. After analyzing these data, the system can identify the key features of each user. Each sampling process lasts about 1 to 5 seconds. This duration ensures that the system can collect enough data for effective identity authentication.

[0046] S3. Signal pre - processing:

[0047] Noise filtering: Use a Butterworth band - pass filter to filter the noise of the vibration feedback signal, removing the frequency bands below 10Hz and above 49.8Hz to exclude the interference of the external environment.

[0048] Direction correction: To solve the signal differences caused by the change of the wrist - wearing posture, the system corrects the signal directions on the three axes of the accelerometer to ensure that the collected vibration signals are consistent. The formula is as follows:

[0049] α′ i =(1 - β)(α i -g i ),i = x,y,z#

[0050]

[0051] where α iis the original acceleration value, g i is the gravitational acceleration captured by the accelerometer along the i-th axis, and β is the filter factor determined by the time constant t of the filter and the event transfer rate dT.

[0052] Normalization: Since the vibration signals collected by the accelerometer of the wrist-worn device are triaxial accelerations along the x, y, and z axes, the value ranges between the three axes vary greatly, and the value ranges vary even more between different device models. Therefore, the amplitude range of the vibration signals is normalized to ensure data consistency for different users, different devices, and different usage scenarios. The formula is as follows:

[0053]

[0054] where α' i is the single reading of the i-th axis after direction alignment, u i and δ i are the mean and standard deviation of the data along the same axis, respectively. After normalization, the accelerometer data (α″ i ) is centered at 0 and scaled proportionally to have a standard deviation of 1. Therefore, data from different sources will be comparable.

[0055] S4. Differential feature extraction: Perform first-order, second-order, fourth-order, and eighth-order difference processing on the preprocessed vibration signals to extract the change characteristics of the signals at different time periods. The difference at each order captures the dynamic changes of the signal, enhancing the system's ability to perceive the wrist vibration pattern. The formula is as follows:

[0056] Diff n (t) = x(t) - x(t - n)#

[0057] where Diff n (t) represents the n-th order difference, and x(t) is the signal value at time point t.

[0058] S5. Processing by multi-branch convolutional neural networks (MCNN): Input the differential features into the multi-branch convolutional neural network. Different convolutional kernels (sizes 1, 2, 4, 8, 16, 32) are used to capture short-term and long-term vibration features. Each branch extracts features at different time scales through the convolutional layer to ensure that the model can accurately understand the complex changes of the vibration signals. Each branch uses 64 filters and adopts the ReLU activation function for non-linear processing. Subsequently, the convolutional output is converted into a one-dimensional vector through the flattening layer. In the feature fusion stage, the Concatenate layer is used to splice the feature vectors from different branches together to form a comprehensive vector containing multi-scale features.

[0059] S6: Classification and Recognition: The comprehensive feature vector generated by the convolutional network is input into the classifier through the fully connected layer, and the Softmax function is used to classify the results to determine whether the vibration response matches the registered user features. If it matches, the authentication is successful, and the device allows the user to access or operate; if it does not match, the system prompts that the authentication fails.

[0060] Example 1: Verification of the Impact of Training Dataset Size on Model Performance

[0061] To verify the impact of different training dataset sizes on the performance of the identity recognition model, in this example, under the condition that the device sampling frequency is 100 Hz, the balance effect between the authentication accuracy rate of the system and the consumption of computing resources for different sampling times was tested. It should be noted that the identity authentication method of the present invention can be carried out on a variety of intelligent wrist-worn devices, such as smart watches, smart glasses, smart phones, etc., further expanding the applicability of the method.

[0062] Experimental Setup:

[0063] Device Selection and Wearing Position: In this example, tests were carried out on the xiaomi watch first-generation smart watch. The experiment required users to wear the smart watch on the left wrist, with the back of the hand facing up and keeping the wrist stable to ensure the consistency and comparability of data collection.

[0064] Data Collection Tool: The experiment used a self-developed Android application to extract data from the accelerometer of the smart watch to ensure the flexibility and accuracy of the data collection process. This application can capture the acceleration response values in the x, y, and z axes to generate vibration feature data for model training.

[0065] Dataset Setup: Four users were randomly selected, and their vibration feedback data at different sampling times were collected respectively to support different-sized datasets. The sampling times were set to 1 second, 2 seconds, 3 seconds, 4 seconds, and 5 seconds, aiming to observe the impact of the change in dataset scale on the model training effect, so as to find the optimal sampling time.

[0066] Vibration Signal Collection and Model Training:

[0067] Signal Collection: The built-in vibration motor of the smart watch was used to apply vibration stimulation to the user as a challenge signal; after the vibration stimulation, the acceleration data collection application obtained the vibration response data, and the sampling frequency was 100 Hz. The acceleration response values of the acceleration sensor on the x, y, and z axes for each sampling were recorded.

[0068] Model processing: Use MCNN for training, adopt a Softmax classifier to classify the feature vectors, compare the extracted vibration features with the features of multiple pre-registered known users, and adopt a 1:N matching method to achieve accurate identity matching.

[0069] Analysis of experimental results:

[0070] Authentication accuracy: According to Figure 3 the results, as the sampling time increases from 1 second to 3 seconds, the authentication accuracy gradually rises and reaches 97.5% at 3 seconds, indicating that the model has sufficient feature information for efficient authentication at this time; after extending the sampling time to 4 seconds and 5 seconds, the accuracy no longer improves but instead increases the computational burden. Therefore, 3 seconds is the optimal sampling time, indicating that a 3-second time segment contains sufficient feature information to achieve efficient and accurate identity authentication with less data volume.

[0071] Model loss value: According to Figure 4 the results, as the sampling time increases from 1 second to 3 seconds, the loss value gradually decreases and drops to the lowest point of 0.12 at 3 seconds, which indicates that this sampling time can provide sufficient and effective feature information for the model, thereby significantly reducing the error. When the sampling time is extended to 4 seconds and 5 seconds, the loss value rises slightly, probably due to the introduction of redundant information resulting in an increased computational burden. Thus, a 3-second sampling time achieves an ideal balance between feature extraction efficiency and model performance.

[0072] This experiment shows that a 3-second sampling time reaches a balance point between the authentication performance of the model and the consumption of computational resources, with a high authentication accuracy and a low loss value. This conclusion highlights the unique advantage of the identity authentication method for wrist-worn devices based on vibration feedback in the present invention in setting the data acquisition time, provides an efficient and low-resource-occupying solution for practical applications, helps to improve the overall performance of the model and optimize resource utilization.

[0073] Example 2: Anti-interference ability test in complex environments

[0074] This example aims to verify the anti-interference ability of the identity authentication method for wrist-worn devices based on vibration feedback in the present invention in complex environments that users may encounter in their daily lives, and compare the authentication accuracy and robustness under different conditions by introducing a control group. The specific test steps are as follows:

[0075] Test environment setting:

[0076] Experimental group (with interference): Simulate the wrist movements of users in daily life, such as walking, slight swinging, or adjusting the wearing position, etc., which cause small displacements of the smartwatch on the wrist. The specific operation is to randomly move the watch up and down along the wrist by about one centimeter to test the robustness of the authentication algorithm under unstable wearing positions.

[0077] Control group (without interference): The user keeps the wrist stationary, and the smartwatch is worn at a fixed position on the wrist to avoid position changes. The test conditions of this group are designed to observe the authentication accuracy rate of the algorithm in the non-interference state as a benchmark for comparison.

[0078] Data collection and authentication process:

[0079] Data collection: Collect the vibration feedback data of the user's wrist under two conditions respectively. Under the conditions of the experimental group, the watch stays at each moving position for several seconds to ensure that the data covers the vibration responses of different wearing positions. In the control group, the position of the watch remains unchanged.

[0080] Authentication process: Preprocess the collected data, including noise filtering, direction correction, and normalization. Add differential features to the processed data and input it into the MCNN model for training, and conduct user identity authentication tests under the two conditions.

[0081] Results and analysis: By comparing Figure 5 the authentication accuracy rates of the experimental group and the control group, the following conclusions can be drawn. Even if the watch moves randomly on the wrist, the authentication algorithm can still maintain an authentication accuracy rate of over 97%, indicating that the algorithm still has a high recognition stability and anti-interference ability under slight position changes. This result verifies the reliability of the present invention in practical applications, especially when the user moves or adjusts the device position, it can still effectively identify the user's identity.

[0082] Example 3: Adaptability test under different hand movements

[0083] This example aims to verify the adaptability of the wrist-worn device identity authentication method based on vibration feedback of the present invention under the condition of user hand movement changes, and evaluate its stability and accuracy under different gestures.

[0084] Test environment setting: Simulate the natural gesture changes of users in daily life, and design several typical hand movements, including:

[0085] Natural relaxation of the palm: The user's hand is completely relaxed without making any specific movements.

[0086] Half-fist: The palm is slightly closed in a half-fist state.

[0087] Thumb up: The palm is naturally relaxed and the thumb points upward.

[0088] Under each gesture, the smartwatch is worn on the user's left wrist to ensure the consistency of the data collection location.

[0089] Data collection and authentication process: Under the above gestures, the vibration motor of the smartwatch is used to generate vibration signals, and the vibration feedback data of the user is collected through the accelerometer. The collected signals are preprocessed (including noise filtering, direction correction, and normalization) to eliminate the data differences that may be brought by different gestures and ensure the consistency of the data.

[0090] Results and analysis: According to Figure 6 the result analysis, it is found that the authentication accuracy under different gestures remains above 97%, and the specific accuracy rate reaches 97.49%, and the loss value is 0.11. There are no significant differences in various evaluation indicators (such as accuracy rate and loss value) under different gestures. This result shows that the identity authentication method of the present invention can still maintain a high recognition accuracy in the case of natural movement changes of the user's hand, fully demonstrating good adaptability.

[0091] It should be noted that although the above steps are described in a specific order, it does not mean that the steps must be executed in the above specific order. In fact, some of these steps can be executed concurrently or even the order can be changed as long as the required functions can be achieved.

[0092] The present invention can be a system, a method, and / or an electronic product. The electronic product can include a readable storage medium, on which readable program instructions for enabling a processor to implement various aspects of the present invention are uploaded.

[0093] The readable storage medium can be a tangible device that holds and stores instructions for use by an instruction execution device. The readable storage medium can for example include but is not limited to an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical coding device, such as a punched card or raised structures in a groove storing instructions thereon, and any suitable combination of the above.

[0094] The above are only the preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for authenticating the identity of a wrist-worn device based on vibration feedback, characterized in that, Including the following steps: Generation and acquisition of vibration signals: Using the built-in vibration motor of the wrist-worn device to generate vibration feedback signals, and using the built-in accelerometer of the wrist-worn device to collect the vibration feedback signals in real time; Signal preprocessing: Noise filtering: Using a Butterworth band-pass filter to filter the noise of the vibration feedback signals, using a cut-off frequency of 10 Hz to 49.8 Hz and filtering the vibration feedback signals outside this range; Direction correction: Correcting the signal directions on the three axes of the accelerometer; Normalization processing: Normalizing the amplitude range of the vibration signals that have undergone noise filtering and direction correction; Differential feature extraction: Performing first-order, second-order, fourth-order, and eighth-order differential processing on the preprocessed vibration signals to extract the change characteristics of the vibration signals at different time periods; Multi-branch convolutional neural network processing: Inputting the extracted differential features into the MCNN; Using convolutional kernels of different sizes to capture short-term and long-term vibration features; Each branch extracts features of different time scales through the convolutional layer and uses 64 filters; Using the ReLU activation function for non-linear processing; Converting the convolutional output into a one-dimensional vector through the flattening layer; In the feature fusion stage, using the Concatenate layer to splice the feature vectors from different branches together to form a comprehensive feature vector containing multi-scale features; Classification and recognition: Inputting the comprehensive feature vector generated by the multi-branch convolutional neural network into the classifier through the fully connected layer, using the Softmax function to classify the results, and determining whether the vibration response matches the registered user features; If it matches, the identity authentication is successful; If it does not match, the identity authentication fails.

2. The method for authenticating the identity of a wrist-worn device based on vibration feedback according to claim 1, wherein The vibration feedback signal is the signal after the original vibration signal of the vibration motor acts on the wrist.

3. The method for authenticating the identity of a wrist-worn device based on vibration feedback according to claim 1, wherein, The response data of the vibration feedback signal collected in real time by the accelerometer under the condition of a frequency of 100 Hz, and each sampling duration is 3 seconds.

4. The method for authenticating the identity of a wrist-worn device based on vibration feedback according to claim 1, wherein The formula for correcting the signal directions on the three axes of the accelerometer is as follows: α′ i =(1 - β)(α i - g i ), i = x, y, z where α i is the original acceleration value, g i is the gravitational acceleration captured by the accelerometer along the i-th axis, and β is a filter factor determined by the time constant t of the filter and the event transfer rate dT.

5. The method for authenticating the identity of a wrist-worn device based on vibration feedback according to claim 1, wherein The formula for normalizing the amplitude range of the vibration signal is: where α′ i is the single reading of the i-th axis after alignment of directions, u i is the mean of the data along the same axis, and δ i is the standard deviation of the mean of the data along the same axis; α″ i is the accelerometer data.

6. A readable storage medium, on which a computing program is stored, characterized in that, When the program is executed by the processor, it can implement the vibration feedback-based wrist-worn device identity authentication method described in any one of claims 1 to 5.

7. An electronic device, characterized in that, Including the readable storage medium described in claim 6.

Images