Research On Handwritten Information Estimation Technology Based On Smartphone Motion Sensor

Handwriting input is one of the most important information input methods for smartphone users.A large amount of user sensitive information is input into the smartphone by handwriting,for example,chat history,password and meeting record.As personal privacy breaches continue to emerge,smartphone users are increasingly focusing on self-private data protection and are beginning to guard against privacy breaches based on conventional methods(malware).However,users are currently lacking sufficient awareness of a more subtle approach to privacy breaches–based on personal privacy information from mobile-side channel analysis.In order to make smartphone users aware of the serious threat of privacy information and its confusing behavior,this paper proposes a handwritten information estimation system based on smart phone motion sensor.The speculative system in this paper is based on the discovery that mobile phone motion sensor data will change regularly with the user's handwriting behavior on the screen of the mobile phone.Moreover,the handwriting trajectories of different characters are significantly different,which may lead to differences in data fluctuations caused by handwritten behavior of various characters.This paper establishes the internal relationship between different handwritten information and the change of sensor data through the method of machine learning,and realizes the process of speculating the user's handwriting input information from the data change.In order to improve the accuracy of information estimation system,this paper not only optimizes the feature set for machine learning for the sensor data change of handwriting behavior,but also incorporates the posture state of the user when using the mobile phone,and minimizes external factors.The impact of information restoration.Finally,based on the collected 43,000 characters of handwritten data,the performance evaluation experiments of the system from various angles are carried out,and the effects of various factors on the system are described and analyzed.The final experimental results show that the average speculative accuracy of digital guessing in the information speculation system reaches 86.5%.Therefore,the information speculation system proposed in this paper can confirm that the side channel analysis technology based on mobile phone sensor does pose a threat to the privacy information of mobile phone users,resulting in personal privacy leakage risk.