Experimental and theoretical study on physical layer authentication for IoT systems
| Project/Area Number |
20K19801
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 60070:Information security-related
|
|---|
| Research Institution | Toyo University |
|---|
Principal Investigator |
朱 金暁 東洋大学, 情報連携学部, 助教 (30754329)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Project Status |
Granted (Fiscal Year 2021)
|
| Budget Amount *help |
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2021: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2020: ¥2,600,000 (Direct Cost: ¥2,000,000、Indirect Cost: ¥600,000)
|
|---|
| Keywords | physical layer / radiometric feature / visibility graph / identification / authentication / fractal dimension / PLA / RF Feature / USRP / IoT / Authentication |
|---|
| Outline of Research at the Start |
This project will study a novel physical layer authentication (PLA) scheme to enhance the authentication ability of communication systems and protect IoT systems against security attacks, such as impersonation attack. The study will be conducted from both experimental and theoretical aspects. Particularly, in the first year, I will construct an experiment testbed and collect data of radio frequency feature from various wireless devices; in the second year, I will extract the raw data, then conduct statistical analysis to explore new features and design novel PLA schemes.
|
| Outline of Annual Research Achievements |
This project focuses on studying physical layer authentication schemes for device authentication/identification in wireless networks. In particular, we focused on analyzing radiometric features of wireless devices from both experimental and theoretical aspects.
In this year, we have expanded the experiment scale and proposed a new radiometric feature. Firstly, in our experiment platform, the number of wireless devices has been increased from 20 to 50 and those devices include four types, namely, USB network interface card, mobile phone, laptop, and wireless router. Secondly, after observing that the existing multi-feature based identification schemes explore features from signal's time, frequency, or phase domain, we are motivated to use visibility graph as a nonlinear characteristic analysis tool to mine intrinsic device characteristics from graph domain. Correspondingly, we proposed a new radiometric feature called normalized horizontal visibility graph Shannon entropy (HVGE) and illustrated that HVGE is independent of other existing features and that it improves the performance of identifying devices in the above experiment platform. Our related result has been published as a journal paper in (Elsevier) Ad Hoc Networks recently.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
Since I took four months maternity leave this academic year, I have extended the research period and rescheduled the research plan. According to the new plan, experiment scale will be expanded to improve the accuracy in performance measurement. Correspondingly, the number of wireless devices in our experiment platform has been increased from 20 to 50. Also we have proposed a new radiometric feature to identify wireless devices in the experiment platform, which meets the research schedule in my new research plan.
|
| Strategy for Future Research Activity |
By now, the major part of research works on physical layer identification focus on proposing new radiometric features together with designing multi-feature based identification schemes. Our previous research effort also falls in such direction. Note that calculating features from the received frames requires a careful design of feature extraction algorithm, and the number of available features effective for device identification is largely limited due to the lack of understanding of the underlying mechanisms of device-specific and channel-specific features.
Meanwhile, some research works try to apply deep learning-based methods, including fully connected neural network (FNN), convolutional neural network (CNN) to extract hidden features from wireless frames without using explicit feature calculation algorithms. Therefore, in the future, I plan to apply machine-learning methods to identify wireless devices based on raw data collected from the experiment platform.
|
Report
(2 results)
Research Products
(3 results)
