Pushed by ever increasing requirements in throughput, mobility and ubiquity, wireless communications are constantly evolving towards more and more complex systems, serving a huge number of human and non-human users.
The wireless networks of tomorrow (6G) will be even more complex than today’s (5G) networks, featuring advanced techniques such as cell-free MIMO, advanced/distributed beamforming, new waveforms, etc. New use cases will set very tough requirements on the networks in terms of throughput, latency, user density and interference.
As for every new network generation, signal processing, at both TX and RX sides, will undergo a steep increase in complexity that is expected to be largely absorbed by CMOS scaling. Yet, the evolving network architectures, the explosion of the number of devices and the increasing variety of verticals (application domains) poses significant challenges to the signal processing in terms of performance, optimization and complexity. The traditional model-based optimizations are less and less applicable as we enter the realm of big data.
We envision PHY signal processing to naturally evolve towards more machine learning based algorithms when model-based approaches will start to fail. This will typically be when statistical assumptions are violated, when non-linearities are present and in interference scenarios.
The purpose of this PhD research is to explore new signal processing methods relying on both model-based and machine learning approaches. Performance will be key but keeping complexity low will be equally important. This research will address most algorithms of the wireless PHY such as channel estimation, equalization, synchronization, tracking, beamforming, front-end non-ideality compensation and transceiver calibration.
The successful PhD candidate will be part of a large IMEC team working on the research, implementation and prototyping of future communications systems: experts in digital, analog and mm-wave ASIC design, wireless communications, PHY processing, MAC and higher layers, machine learning.
Supervisor : Ingrid Moerman
Daily Advisor: Andre Bourdoux
Type of work : 80% modeling/simulation, 10% experimental, 10% literature
Required background: Electrical Engineering, Signal Processing for Communications
The reference code for this position is 2020-117. Mention this reference code on your application form.