search

DoHaiSon / ISDNN

Code for "ISDNN: A Deep Neural Network for Channel Estimation in Massive MIMO systems," Hanoi University of Industry Journal of Science and Technology, Sept. 2024.
GNU General Public License v3.0
4 stars 1 forks source link

DetNet missing within the given codes. #1

Open talefty123 opened 1 week ago

talefty123 commented 1 week ago

Hi, I have read your paper at ISDNN: A Deep Neural Network for Channel Estimation in Massive MIMO systems.

At Fig 3, you mentioned of using NMSE estimation for the DetNet architecture, I am a newbie of applying Machine Learning in Telecommunication at undergrate level, would you kindly give me more in details on how you used the algorithm for the comparision of the two proposed models ISDNN and S-ISDNN within the paper as I read about the DetNet, most of it comes from using an entirely differently library as TensorFlow whereas yours is Pytorch.

Thank you for reading, and I wish you have a great day!

DoHaiSon commented 1 day ago

Thx for ur asking, If we use the DetNet repo for comparison, it is not feasible because its purpose is data detection. Therefore, I rewrote a version of it based on the mathematical formulation described in the paper to create a version for result comparison. You can find it in the code I uploaded. The limitation is that DetNet only works with a batch size of 1, and I don't have the time to update this issue, but essentially, it's just a runtime problem.