COVID era and its impact on my life

 

Like everyone else, 2020 was the weirdest year for me among the past years in my life.  It was too sad and heartbreaking to see and hear that people were losing their jobs and more painfully their lives due to COVID. In our family, we too, very sadly, lost one of our loveliest relative due COVID that was very tragic.

However, despite all the challenges that COVID has brought for humans, from the academic aspect, it gave me the opportunity to work and focus more on my Ph.D. project which was essential for me at the beginning of 2020. Thanks to remote working, I was able to save the time and energy I had been spending commuting to our campus located at Palaiseau, 20km south of Paris. Although emotionally this was hard, it gave me 4-5 more useful hours each day to work on my project. I extremely needed this time due to the high volume of the required studies and experiments to be done owing to the interdisciplinary nature of my project between AI and optical fiber communications. On the other side, at home, I was able to concentrate highly more on my experiments and studies. As I am rather a social person, when I was going to the campus, some amount of my time used to be spent talking with my colleagues on some non-academic topics that would cause my concentration to be lost. Definitely, the absence of these talks and interacting with other people in society brought psychological damage to me, but if I had not had this opportunity I would have faced trouble for my mid-term defense.

Thankfully my mid-term defense passed well on 16 Nov. 2020. Thanks to the more available working time, I had carried out various experiments and succeeded in proposing a more sophisticated neural network (approximating optical fiber channel) than what was expected by the jury in even six months prior to defense. The proposed neural network architecture caught the jury's attention and convinced the supervisory board to let me continue my research on it in the second half of my thesis during the secondment.

It was demonstrated that the proposed neural network architecture provides roughly the same BER performance as the state-of-the-art approach while requiring lower memory and computational resource consumption. It was also shown that the proposed architecture has a better generalizability power over distance when trained for a fixed fiber length.

Due to COVID, we have started the secondment virtually, but thankfully, owing to have remote access to the TPT and NBL servers, we have been able to relatively handle this situation and push forward our experiments. In our remote meetings, sometimes we feel that we need to write to each other on paper to explain a mathematical matter or to express something in person. This is the most challenging part of remote working. Many times we managed to come up with a workaround, but sometimes we were not able to do so. Finding a workaround is always pleasing as it takes our mind out of the conventional solutions for the interactions and it opens new doors to us for better communication, such that several times, this tempted us to extend our new interaction solution as an application for the world. From the other side, the cases that put us in challenge remind us that the world requires more internet bandwidth in order to have more advanced and better remote interaction tools, for instance, 3D virtual conferences. This encourages us to work harder to push this mission forward with more determination ever than before.