SLIDE 1
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
2
Machine Learning based fingerprinting of Wireless modulation schemes - - PowerPoint PPT Presentation
Machine Learning based fingerprinting of Wireless modulation schemes - - PowerPoint PPT Presentation
Machine Learning based fingerprinting of Wireless modulation schemes Pratik Satam, Gregory Ditzler and Salim Hariri Sponsor: 802 Secure Inc. Sponsor: MITSUBISHI CAC Semi-Annual Meeting October 5-6, 2016 Dallas, Texas Project Overview CAC
SLIDE 2
SLIDE 3
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
3
Team and Leverage
p Team
n 802 Secure
p Kurt Gurtzmacher p Albert Pham
n UA Faculty
p Salim Hariri p Gregory Ditzler
n UA Graduate Student
p Pratik Satam
p Leverage
n Techniques developed during Anomaly Behavior Analysis (ABA)
methodology.
SLIDE 4
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
4
Background
n
IoT devices depend heavily on wireless networks for communication.
n
Its difficult to identity spoofed wireless packets/frames.
n
Fingerprinting the wireless modulation schemes and then the hardware used to transmit these signals will give us an effective method to identify spoofed packets.
SLIDE 5
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
5
Project Tasks: Overview
p Task 1: Collect Wireless traffic for different modulation types.
n This step involves collecting data for different wireless modulations.
p Task 2: Design machine learning/ neural network-based
models to finger-print the modulation scheme.
n This step involves developing machine learning models and neural
networks that are able to identify different modulation schemes.
p Task 3: Integrate the neural networks/ machine learning
models into a complete system.
SLIDE 6
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
6
Task 1: Collect Wireless traffic for different modulation types.
p Collect traffic using
Software defined radios(SDR) like hackrf one and rtl-sdr.
SLIDE 7
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
7
Task 2-3 Building and integrating machine learning/ Neural network modules
p We plan to build machine
learning models and neural networks that are able to characterize modulation schemes.
SLIDE 8
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
8
Preliminary Analysis Results
SLIDE 9
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
9
Activities and outcomes
p The primary goal of this project is to design, develop and
implement a framework to fingerprint wireless modulations and use those fingerprints to identify the type of modulation used by the analyzed wireless signal.
SLIDE 10
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
10
Deliverables and benefits
Deliverables
- Midterm and final reports documenting research methods, progress, results, and
analysis
- One or two scholarly conference and/or journal publications
Benefits
- Capability to identify different modulation schemes using the system.
- Receive software tools developed to detect different modulation schemes.
SLIDE 11
CAC Semi-Annual Meeting
October 5-6, 2016 Dallas, Texas
11