Laurent Schmalen

The new roadmap on optical communications @IOPPublishing is now out!! iopscience.iop.org/article/10.108…

IEEE JSAC has a Call for Papers on optical communication and networking and we have asked the team of Guest Editors to provide their perspective on the state of research in this area of communications. comsoc.org/publications/b…

@bayvel_p @AvHStiftung @TU_Muenchen How long will you stay?

Excited to start my @AvHStiftung visit @TU_Muenchen!

#Quantum #LDPC codes decoded with belief propagation don't need to perform poorly. If we use a modified ensemble decoder and an adapted #QLDPC code design, we can mitigate the effects of unavoidable short cycles. #QEC #ErrorCorrection arxiv.org/abs/2401.06874

@AndreasAtETH I have a couple invite codes for scientists. DM me if interested

I have two more invites to share with a #quantum person to let their inner butterfly loose in the blue skies.

@CraigGidney There is an interesting analogy in communications. Until the 90s, people were chasing for codes with good distance properties and good performance. Turbo codes from 1993 turned out to have amazing performance in the regime of interest with only logarithmic distance growth

IMO it's a strategic mistake for quantum LDPC codes to aim for linear scaling in distance. Distance 50 is already overkill. I don't need to super duper overkill with a distance of 100, then 200, then 300, etc. Take those skill points out of distance and put them into coding rate.

@taodream @ECOC_Conference What’s the difference between PhD and EngD?

Yesss! Come and find me at @ECOC_Conference #Glasgow if interested!

@ID_AA_Carmack This was done in the GSM full rate speech codec. Only the MSBs were protected against bit errors by a convolution code. The LSBs of the residual signal (after linear prediction, sort of lossy compression ) were not protected as bit errors didn’t cause too many distortions

Interesting to consider error correcting codes that spend the overhead bits in a precision aware manner, allowing the less significant bits to be unprotected, while protecting and monitoring the more significant. If the traffic is floating point for ML work, just protecting the sign and exponent might be sufficient. I can imagine dedicated future hardware networks operating as isochronous streams, just accepting rare bit errors without any attempt at retransmission.

Call for Papers: IEEE VCC 2023 This brand-new conference supports researchers, professionals, and scholars in communications who may be unable to travel with a virtual platform to present their work. Submit your paper by 1 Sept: bit.ly/3qbvsla

The first IEEE International Conference on Machine Learning in Communications and Networking will be held in Stockholm, Sweden, May 2024. The submission deadline this September! Please spread the word and submit great work! @ieeeorg @comsoc #IEEE #ICMLCN icmlcn2024.ieee-icmlcn.org

@DarkoZibar14 @VILLUMFONDEN @ERC_Research Amazing! Great topic! Looking forward for collaboration

Truly humbled and thankful to receive one of the most prestigious Danish personal grants. The best thing is that I did it by focusing on highly technical journals and not the journals where all the interesting stuff ends in the supplementary material @VILLUMFONDEN @ERC_Research

@taodream Congrats! What’s the conclusion of the book? It is rather thin?

One should get promoted every month...

It's quite something that the world's top scientists put out a synthesis report that essentially says civilization as we know it will end unless the world comes together to stop rapidly expanding the fossil fuel industry but to instead rapidly end it, and almost no one cares

@divbyzero I also like to think of Sudoku as an error correcting code, in particular a code on a graph. Solving a sudoku boils down to decoding a code word transmitted over an erasure channel.

Sudoku is a problem in graph coloring. Each cell is a vertex, and there are edges from each cell to every cell in its row, column, and 3x3 grid. The objective is to "color" the graph with the numbers/colors 1 to 9 so no two cells joined by an edge are the same color.

Very happy to have been elevated to the grade of #IEEE #Fellow by @IEEEorg . Thanks and appreciation go to all my collaborators and students from @BellLabs and @KITKarlsruhe without whom I couldn‘t have made these contributions!

@ShriramKMurthi In particular in Germany, you can graduate at any point during the year, there is no fixed schedule. Also funded projects can start at any point in time. PhD students are not treated as students but rather as research assistants that join/leave the lab or get a PhD at any time

Something I don't understand. European profs will post "I have a new PhD position, deadline in two weeks" at random times of year. How do they get any apps? Who's sitting around waiting for these? In the US everyone is sync'ed around a standard calendar. #AcademicTwitter

@GerryMcBride The Siemens mobile Xelibri line was completely out of line. Here are just a few examples

After the early bricks but before smartphones, there was a stretch where phone design went absolutely catshit bananas (1/?)

Nach langer Pause gibt es wieder eine IEEE SB KA Signal Intelligence Challenge! #isic Am 02.11. geht es los! Alle Infos findet ihr wie immer auf ieee-ka.de ieee-ka.de/announcements/…