Dr Stefano Tomasin,
University of Padova, IT
Prof. Xianbin Wang,
Western University, CA
£50 Tutorial only | Tutorial + Full Conference access from £120
Half-day Tutorial, Monday 31st August 14:00 (BST)
Further pricing details can be found on the registration page.
Many emerging services over wireless networks rely on the information of user location, and sometimes specific regulations are enforced according to the user position.
Device tampering prevents resorting to user-reported location and it becomes critical to obtain an independent verification of the position through the wireless network. This verification may also confirm the user identity, when their position is slowly time-varying and user identity-position has been authenticated earlier. The location can be verified by analyzing the wireless channel, whose some features (such as the path-loss) are immediately associated to the position, while others (such as multipath or shadowing) may be treated as noise.
About user authentication also the imperfections introduced by the transceivers may be exploited. Still, making decisions on physical layer features with traditional tools may be challenging, especially when feature statistics are unknown. Machine learning strategies instead directly learn the function relating the estimated features with the decisions.
This tutorial will give an overview of physical layer user authentication and location verification techniques, outlining potentials and shortcomings, and indicating practical solutions. An important part will focus on machine learning approaches for fusing multiple channel features, also establishing the connection with optimal authentication when statistics are perfectly known.
Structure and content
At the centre of the tutorial is how to confirm the identity of the author of a message (user authentication) or how to ensure that the device of interest is in a given location or area (location verification), using the characteristics of the channels over which the communication occurs. In order to solve this problem a number of associated technical issues are addressed:
- Hypothesis testing when partial information is available;
- Architecture and training of machine learning solutions;
- Estimation of channel features for authentication purposes;
- Security of the physical layer authentication.
The tutorial structure will run as:
- Communication Security in 5G and Beyond
- Design & Implementation of Physical Layer Authentication
- Design & Implementation of In-Region Location Verification at the
- Physical Layer
- Machine Learning Techniques (classification problems, neural
- networks, support vector machines)
- Machine Learning based In-Region Location Verification, with
- Machine Learning based Physical Layer Authentication, with
- Conclusions and open research points