Keystroke Recognition using Wi-Fi Signals

Slides:



Advertisements
Similar presentations
We Can Hear You with Wi-Fi !
Advertisements

Noise & Data Reduction. Paired Sample t Test Data Transformation - Overview From Covariance Matrix to PCA and Dimension Reduction Fourier Analysis - Spectrum.
Recognizing Human Actions by Attributes CVPR2011 Jingen Liu, Benjamin Kuipers, Silvio Savarese Dept. of Electrical Engineering and Computer Science University.
Principal Component Analysis Based on L1-Norm Maximization Nojun Kwak IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008.
1 A Spectral-Temporal Method for Pitch Tracking Stephen A. Zahorian*, Princy Dikshit, Hongbing Hu* Department of Electrical and Computer Engineering Old.
Evaluating Color Descriptors for Object and Scene Recognition Koen E.A. van de Sande, Student Member, IEEE, Theo Gevers, Member, IEEE, and Cees G.M. Snoek,
Whole-Home Gesture Recognition Using Wireless Signals —— MobiCom’13 Author: Qifan Pu et al. University of Washington Presenter: Yanyuan Qin & Zhitong Fei.
FM-BASED INDOOR LOCALIZATION TsungYun 1.
Face Recognition and Biometric Systems Elastic Bunch Graph Matching.
Proposed concepts illustrated well on sets of face images extracted from video: Face texture and surface are smooth, constraining them to a manifold Recognition.
Data preprocessing before classification In Kennedy et al.: “Solving data mining problems”
Secure Unlocking of Mobile Touch Screen Devices by Simple Gestures – You can see it but you can not do it Arjmand Samuel Microsoft Research Muhammad Shahzad.
A Comprehensive Study on Third Order Statistical Features for Image Splicing Detection Xudong Zhao, Shilin Wang, Shenghong Li and Jianhua Li Shanghai Jiao.
Fifth International Conference on Information
A Practical Approach to Recognizing Physical Activities Jonathan Lester Tanzeem Choudhury Gaetano Borriello.
Keystroke Biometric Studies Security Research at Pace Keystroke Biometric Drs. Charles Tappert and Allen Stix Seidenberg School of CSIS.
Applications of Wavelet Transform and Artificial Neural Network in Digital Signal Detection for Indoor Optical Wireless Communication Sujan Rajbhandari.
Multi-Scale Analysis for Network Traffic Prediction and Anomaly Detection Ling Huang Joint work with Anthony Joseph and Nina Taft January, 2005.
Keystroke Biometric Studies Keystroke Biometric Identification and Authentication on Long-Text Input Book chapter in Behavioral Biometrics for Human Identification.
Hand Movement Recognition By: Tokman Niv Levenbroun Guy Instructor: Todtfeld Ari.
DAISY Data Analysis and Information SecuritY Lab
Facial Recognition CSE 391 Kris Lord.
Extracting Places and Activities from GPS Traces Using Hierarchical Conditional Random Fields Yong-Joong Kim Dept. of Computer Science Yonsei.
Presented by Tienwei Tsai July, 2005
COMMON EVALUATION FINAL PROJECT Vira Oleksyuk ECE 8110: Introduction to machine Learning and Pattern Recognition.
Snooping Keystrokes with mm-level Audio Ranging on a Single Phone
Keystroke Recognition using WiFi Signals
Simulation Of A Cooperative Protocol For Common Control Channel Implementation Prepared by: Aishah Thaher Shymaa Khalaf Supervisor: Dr.Ahmed Al-Masri.
Video Based Palmprint Recognition Chhaya Methani and Anoop M. Namboodiri Center for Visual Information Technology International Institute of Information.
K. Selçuk Candan, Maria Luisa Sapino Xiaolan Wang, Rosaria Rossini
Experimental Results ■ Observations:  Overall detection accuracy increases as the length of observation window increases.  An observation window of 100.
Optimal Dimensionality of Metric Space for kNN Classification Wei Zhang, Xiangyang Xue, Zichen Sun Yuefei Guo, and Hong Lu Dept. of Computer Science &
A NOVEL METHOD FOR COLOR FACE RECOGNITION USING KNN CLASSIFIER
Secure Unlocking of Mobile Touch Screen Devices by Simple Gestures – You can see it but you can not do it Muhammad Shahzad, Alex X. Liu Michigan State.
Turning a Mobile Device into a Mouse in the Air
Identifying “Best Bet” Web Search Results by Mining Past User Behavior Author: Eugene Agichtein, Zijian Zheng (Microsoft Research) Source: KDD2006 Reporter:
Instance-Based Learning Evgueni Smirnov. Overview Instance-Based Learning Comparison of Eager and Instance-Based Learning Instance Distances for Instance-Based.
Acoustic Eavesdropping through Wireless Vibrometry University of Wisconsin – Madison, Chinese Academy of Sciences School of Electronic Information and.
SUPERVISED AND UNSUPERVISED LEARNING Presentation by Ege Saygıner CENG 784.
1 Yue Qiao, Ouyang Zhang, Wenjie Zhou, Kannan Srinivasan and Anish Arora Department of Computer Science and Engineering PhyCloak: Obfuscating Sensing from.
Feature Matching and Signal Recognition using Wavelet Analysis Dr. Robert Barsanti, Edwin Spencer, James Cares, Lucas Parobek.
WiFi Action Recognition via Vision-based Methods Jen-Yin Chang, Kuan-Ying Lee, Kate Ching-Ju Lin*, Winston Hsu Communication and Multimedia Lab National.
CSCI 631 – Foundations of Computer Vision March 15, 2016 Ashwini Imran Image Stitching Link: singhashwini.mesinghashwini.me.
Feature learning for multivariate time series classification Mustafa Gokce Baydogan * George Runger * Eugene Tuv † * Arizona State University † Intel Corporation.
TBAS: Enhancing Wi-Fi Authentication by Actively Eliciting Channel State Information Muye Liu, Avishek Mukherjee, Zhenghao Zhang, and Xiuwen Liu Florida.
CS : Designing, Visualizing and Understanding Deep Neural Networks
When CSI Meets Public WiFi: Inferring Your Mobile Phone Password via WiFi Signals Warren Yeu When CSI Meets Public Wifi.
When CSI Meets Public WiFi: Inferring Your Mobile Phone Password via WiFi Signals Adekemi Adedokun May 2, 2017.
Machine Learning with Spark MLlib
RF-based positioning.
Keystroke eavesdropping attacks with WiFi signals
Pat P. W. Chan,  Michael R. Lyu, Roland T. Chin*
Supervised Time Series Pattern Discovery through Local Importance
ABSTRACT FACE RECOGNITION RESULTS
Hybrid Features based Gender Classification
Vijay Srinivasan Thomas Phan
WiTrace: Centimeter-Level Passive Gesture Tracking Using WiFi Signals
Context-Free Fine-Grained Motion Sensing using WiFi
WiFinger: Talk to Your Smart Devices with Finger-grained Gesture
DAISY Friend or Foe? Your Wearable Devices Reveal Your Personal PIN
We Can Hear You with Wi-Fi !
Enhancing Diagnostic Quality of ECG in Mobile Environment
Anindya Maiti, Murtuza Jadliwala, Jibo He Igor Bilogrevic
Acoustic Eavesdropping through Wireless Vibrometry
Faculty of Science IT Department Lecturer: Raz Dara MA.
Kyoungwoo Lee, Minyoung Kim, Nikil Dutt, and Nalini Venkatasubramanian
QGesture: Quantifying Gesture Distance and Direction with WiFi Signals
Combating Replay Attacks Against Voice Assistants
Mole: Motion Leaks through Smartwatch Sensors
Advisor: Dr.vahidipour Zahra salimian Shaghayegh jalali Dec 2017
Presentation transcript:

Keystroke Recognition using Wi-Fi Signals ACM MOBICOM 2015 Kamran Ali Dept. of Computer Science & Engineering Michigan State University Alex Liu Wei Wang Muhammad Shahzad

Keystroke Recognition Good Bad Virtual Keyboards Keystroke Eavesdropping Kamran Ali

Previous keystroke recognition schemes Camera based Sound based EM Radiations based SDR based Kamran Ali

Can we recognize keystrokes using commodity WiFi ? WiKey Key observations: Keystrokes impact WiFi signals – multipath changes Different keystrokes impact WiFi signals differently Channel State Information (CSI) Letter Letter I O Kamran Ali

Challenges Keystrokes are small gestures Constitute small motions Closely placed on keyboard Closely spaced in time Key challenge Detection and extraction of clean CSI waveforms for different keystrokes Kamran Ali

Noise Reduction Noisy CSI in all subcarriers  Low pass filtering CSI variations in subcarriers are correlated  30 groups of subcarriers per TX-RX antenna pair  Contain redundant information Principal Component Analysis (PCA) on subcarriers  Select top few projections of CSI data  Remove the noisy projections of CSI data Kamran Ali

Adds robustness against unrelated noisy CSI variations Noise Reduction Example Noisy projection Adds robustness against unrelated noisy CSI variations Kamran Ali

Keystrokes Extraction Observation: Processes waveforms from all TX-RX antenna pairs Robustly estimates the start and end points  Combines results from all TX-RX antenna pairs Keystrokes extracted using start and end points Typical increasing and decreasing trends in rates of change in CSI time-series Kamran Ali

Feature Extraction Shapes of keystroke waveforms used as features Discrete Wavelet Transform Compressed shape features from CSI waveforms Applied 3 times consecutively to reduce computational complexity Kamran Ali

Feature Extraction: Examples Some DWT Features of keystroke I Some DWT Features of keystroke O Kamran Ali

Extracted Keystroke Waveforms Classifier Training Dynamic Time Warping Comparison metric for shape features of keystrokes k-Nearest Neighbor (kNN) Classifiers Majority voting on decisions from all classifiers Total classifiers 3 x MT x MR = Extracted Keystroke Waveforms From all antenna pairs Kamran Ali

Laptop with Intel 5300 WiFi NIC Data Collection Experimental setup Intel 5300 NIC for CSI collection at receiver ICMP ping requests sent to router from laptop Collected data from 10 users For both separate keys & sentences More than 1480 samples collected from each user Inter-keystroke interval ~ 1 second 30 cm 4 m Laptop with Intel 5300 WiFi NIC TP-link router Kamran Ali

Keystroke Extraction Accuracy Keystroke extraction achieves average accuracy of 97.5% over all users Key misses occur due to: Inconsistencies in typing behavior Keys constituting smaller motions Kamran Ali

Classifier Accuracy: Single keys Experiment [1] Keys A-Z, 0-9 & Space Bar. Samples/key = 30 Slightly smaller accuracies in case of all keys Reason: Similarity of QWE row with digit keys User IDs 83% 10-fold cross validation accuracy averaged over all keys and all users Kamran Ali

Classifier Accuracy: Single keys Experiment [2] – Performed for user #10 Changing percentage of training set from 50% to 90% Keys tested A-Z. Samples/key = 80 Multifold cross validated accuracies stayed >= 80% Accuracies for keys like ‘j’, ‘k’, ‘v’, ‘e’ dropped < 60% Kamran Ali

Classifier Accuracy: Sentences Experiment [1] - Users typed 1 sentence with 2 repetitions - 30 training samples per key User IDs Average accuracy of 77.43% over all users Kamran Ali

Classifier Accuracy: Sentences Experiment [2] – Performed for user #10 80 training samples, 5 sentences, 5 repetitions Average accuracy increased from 80% to 93.47% Kamran Ali

Limitations Tested in interference free surroundings Affected by change in the positions of Wi-Fi devices Supports relatively slower typing speeds Approximately 15 words/minute Requires high CSI sampling rate Approximately 2500 samples/sec Requires many training keystroke samples per key Kamran Ali

Conclusions Wi-Fi based keystroke recognition scheme Correlations in Wi-Fi subcarriers can be leveraged to reduce noise Propose a robust algorithm for keystroke extraction Shapes of CSI waveforms  effective features for recognition of small gestures Wi-Key can achieve more than 90% keystroke recognition accuracy for reasonable typing speeds Kamran Ali

AirPass Biometrics using Wi-Fi signals Even if you know the password, you can not get into the system Variations in Wi-Fi signals caused when different people type same password is different Regular passwords are mainly prone to Video based attacks Shoulder surfing Limitations Designed for relatively stable environments (No major motion other than typing) E.g. personal office, home, library, etc. High intra-class variance [i.e. same user’s samples can vary a lot, as Wi-Fi signals are very sensitive] Kamran Ali

Questions ? Thank you! Kamran Ali