PhD - Leuven | More than two weeks ago
You will design wireless links targeting Tbps throughput to support interconnections in future systems.
Wireless communications are offering more and more throughput, connecting more and more devices of different types through cellular and point-to-point links. This growth has been supported by increasing carrier frequencies, such that wider bandwidths can be used, and by simultaneously increasing the density of devices and the number of spatial streams.
Pushing this reasoning further, wireless connections with throughputs above 100 Gbps at short range can also replace high-speed wired connections, offering equivalent throughput with easier installation, more flexibility, and lower cost and power consumption.
One example is particularly demanding: connecting equipment within data centers where multiple links offering Tbps throughput are needed. In order to compete with alternative solutions such as optical fibers, wireless links need to push throughput, latency and power consumption requirements to the extreme. One way to do so is to exploit mm-wave and sub-THz frequencies, such as bands around 140 GHz, and in the the future around 300 GHz.
The goal of this PhD is to design short-range wireless links targeting a throughput of the order of 1 Tbps. The main use case to consider is communication within data centers, between different boards/racks of high-performance computers. However, other systems requiring extreme-throughput interconnects are also investigated. Moreover, this exercise will include benchmarking of the proposed wireless links with relevant wired alternatives.
This PhD will include:
Multiple-antenna technologies will be of crucial importance, as carrier frequency increase goes together with smaller antenna elements and larger antenna arrays. This will be supported by hybrid architectures combining analog and digital processing.
This PhD will build on imec's experience of high-throughput mm-wave communication systems as well as high-performance computing systems. The candidate will explore state-of-the-art literature and extend (Matlab-based) internal tools for performance and power modeling, in order to analyze system trade-offs and propose novel solutions.
Required background: Electrical Engineering, Signal Processing for Communications
Type of work: 20% literature/theory, 60% modelling/simulation, 20% design/experimental
Supervisor: Sofie Pollin
Daily advisor: Claude Desset
The reference code for this position is 2022-120. Mention this reference code on your application form.