Please post your questions here

Good morning Professor,

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?

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.

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

Hello Prof Kaushik, could you please briefly introduce the purpose of training set normalization?

Wonderful presentation! Thank you. I am wondering how can we extract the IQ samples from off-the-shelf hardware?

thank you for yourcexcellent présentation!

Great presentation. How reliable would be the proposed approach to fingerprint 1.9 trillion devices? How scalable and reliable would it be?

Also, could an adaptive learning solution that tracks the changing of the channel without retraining the CNN be possible ?

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?

Thank you for the presentation and the answer :)

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?

How hard would it be for an adversary to impersonate another device by adjusting their IQ imbalance or by using some other method?

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?