Guoda Tian worked on machine learning for wireless networks

Published: 2024-11-12

Title of thesis: Navigating the future - Machine learning for wireless sensing and localization. Link to thesis in Lund University Research Portal. Defence: Monday November 18th, 09:15, room E:1406. Zoom Link: https://lu-se.zoom.us/j/62386045480...

Title of thesis: Navigating the future - Machine learning for wireless sensing and localization.

Link to thesis in Lund University Research Portal.

Defence: Monday November 18th, 09:15, room E:1406.
Zoom Link:
https://lu-se.zoom.us/j/62386045480
Zoom ID: 62386045480

Describe your research in a popular science way

My research topic is the application of machine learning algorithms to wireless sensing and localization to improve system performance. My thesis covers three main areas: ML(machine learning)-based channel estimation, ML-based wireless sensing, and ML-based wireless positioning.

What is the most fascinating or interesting with your thesis subject?

In my thesis, I develop theoretical algorithms and conduct practical measurements to validate the results. This dual approach allows us to directly assess the performance of the proposed algorithms in practical applications.

Do you believe some results from your research will be applied in practice eventually? And if so, how / how?

I am confident that some of the results from my research will be applied in practice. The combination of machine learning and traditional signal processing algorithms represents a promising direction for future wireless systems, and relevant standardization organizations have already begun planning and discussing this topic.

What are your plans?

Currently, I am working full-time in Ericsson Sweden. My career plan is to be a standardization engineer so that I can contribute to the standard and push the frontier of wireless technology even further.

Link to the article Guoda Tian worked on machine learning for wireless networks

Mohammad Attari designs novel application-specific instruction set processors

Published: 2024-10-29

Title of thesis: Application Specific Instruction-set Processors for Massive MIMO Systems Link to thesis in Lund University Research Portal. Defence: Monday November 11th, 09:15, room E:1406. Zoom Link: https://lu-se.zoom.us/j/68795368213 Zoom ID:...

Title of thesis: Application Specific Instruction-set Processors for Massive MIMO Systems

Link to thesis in Lund University Research Portal.

Defence: Monday November 11th, 09:15, room E:1406.
Zoom Link:
https://lu-se.zoom.us/j/68795368213
Zoom ID: 68795368213

Describe your research in a popular science way
It is easy to forget that digital processors form the substrate on which most of the amazing technological advances we enjoy today are built upon. The way we communicate nowadays is no exception. The cellphone you hold in your hand is a powerhouse in its own right, matching, and even sometimes dwarfing, the performance of what room-sized supercomputers were capable of 20 years ago. There are also wireless communication systems that work tirelessly, day and night, to bring what you want to life. For instance, communication towers employ hundreds of antennas in a setup that is called massive MIMO. Having these many antennas implies the need for a commensurately powerful processor to crunch the numbers. But performance is only one side of the coin. With the advent of every new standard, and as we crank up the numbers sitting next to their G neighbors (3G, 4G, 5G, ...), what seems to become more clear is that not only do we need more performance, we also have to deal with different scenarios. This requires processors that are powerful in their number crunching capabilities, as well as in their ability to cope with this varied nature. And that is what my research boils down to: creating application specific instruction set processors known as ASIPs that can take on the vagaries of the changing standards with programmability, and yet stand tall when it comes to processing power by seeking aid from customized accelerators.

What made you want to pursue a PhD?
I wasn?t really planning on running the PhD gauntlet, but my sister floated the idea one day, and, after some initial (and futile) resistance, I picked it up. I?m a more practical-oriented kind of dude, so I thought maybe looking at things from the lens of research by means of a PhD degree would be an interesting challenge, so I went for it.

Do you believe some results from your research will be applied in practice eventually? And if so, how / how?
It?s tough to opine on this with certainty, as human beings are not known for their oracular abilities. Having said that, I believe malleability of the system when it comes to handling disparate problems is going to be a key factor going forward, and as a result, ASIPs look like an appetizing choice to fill this role. So, are the ASIPs what the doctor ordered? Only time will tell.

What are your plans?
Now that I?ve had my daily dose of PhD for the past few years, I am going to move on to industry for a change of pace.

Link to the article Mohammad Attari designs novel application-specific instruction set processors

Mgeni Makambi Mashauri worked with spatial coupling in error control

Published: 2024-10-21

Title of thesis: Enhancing Iterative Algorithms with Spatial Coupling Link to thesis in Lund University Research Portal. Defence: Friday November 8th, 09:15, room E:1406. Zoom Link: https://lu-se.zoom.us/j/67878645197 Zoom ID: 67878645197 ...

Title of thesis: Enhancing Iterative Algorithms with Spatial Coupling

Link to thesis in Lund University Research Portal.

Defence: Friday November 8th, 09:15, room E:1406.
Zoom Link:
https://lu-se.zoom.us/j/67878645197
Zoom ID: 67878645197

Describe your research in a popular science way

To get good performances with feesible complexity in many systems especially those dealing with communication, two or more components have to exchange information iteratively. The choice of the components is a critical decision for such iterative systems since having the best components on each part does not guarantee good performance. Good performance comes when the components are matched. This matching comes with compromises on some aspects of the performance.

My research investigates the application of spatial coupling, a novel concepts in modern error control coding, to iterative systems ranging from communication channels with intersymbol interference to group testing (a scheme aimed at reducing the number of tests to identify objects of interest in a large population). With spatial coupling blocks of information are not processed in isolation but are linked to form a chain resulting into better and robust performance without the trade-offs involved in classical systems.  We also propose new and better approaches in the design and processing of iterative algorithms.

What made you want to pursue a PhD?

  I pursued a PhD because of the desire to dig deeper into the things. I saw this as an opportunity to push my limits and gain skills for my future pursuits.

What is the most fascinating or interesting with your thesis subject?

It is interesting to see how one concept of spatial coupling could be applied to different fields which on the surface are unrelated with interesting results.

What are your plans?

My plans are to use the skills I have gained to explore more areas of interest by continuing research in academia or solving more hands on problems in industry.

Link to the article Mgeni Makambi Mashauri worked with spatial coupling in error control