Zoom Logo

TCCN Sec-IG, Virtual Seminar Series - Shared screen with speaker view
Dola Saha
01:04:06
Please post your questions here
Andrea Stomaci
01:04:29
Good morning Professor,
Jens Schlumberger
01:06:04
For the Slide "Learning the Channel": How do you select the right part of a signal out of a bigger trace? How to classify preamble / payload?
Junqing Zhang
01:06:09
Thanks for the excellent talk. I would like to ask two questions.(1) What could be the possible reasons for the performance degradation regarding training one day test on another day?(2) In practical applications, how to inject transmitter impairments or apply the FIR filter to the transmitter? As many IoT devices won't be able to be customized.
Andrea Stomaci
01:06:11
I’d like to thank you for this talk and I also would like to ask you if the resilience of this system has been tested again an attacker that could jam the channel by injecting noise
Shen
01:06:22
Hello Prof Kaushik, could you please briefly introduce the purpose of training set normalization?
Laxima Niure Kandel
01:06:43
Wonderful presentation! Thank you. I am wondering how can we extract the IQ samples from off-the-shelf hardware?
Lamya Fergani
01:07:30
thank you for yourcexcellent présentation!
Vijay Shah
01:07:44
Great presentation. How reliable would be the proposed approach to fingerprint 1.9 trillion devices? How scalable and reliable would it be?
Andrea Stomaci
01:08:16
Also, could an adaptive learning solution that tracks the changing of the channel without retraining the CNN be possible ?
Manolya ATALAY
01:10:09
Hello dear professor,Thank you for your great presentation. Have you ever considered comparing your works with Attention-based LSTM networks or similar variants? Do you think multiple devices/agents would make this experiment inefficient?
Jens Schlumberger
01:10:33
Thank you for the presentation and the answer :)
CT Huang
01:15:03
Following the previous question, how do we know whether the normalization step is needed? Do you first see whether the learning is effective to determine whether normalization is necessary?
Jake Perazzone
01:21:08
How hard would it be for an adversary to impersonate another device by adjusting their IQ imbalance or by using some other method?
Xuchen Zhu
01:22:14
That's an amazing presentation. I would like to ask the question. In your paper, you mentioned low-cost devices are deployed in IoT network. So, does the filter-based menthod cost more energy? Is that method energy efficient?