Title: Analog Computing for Communications and Signal Processing

Abstract: Modern systems rely on digital circuits, computing, and signal processors that operate on binary values, offering advantages such as precision, versatility, robustness. Yet, digital processors face major limitations as high power consumption and limited speed. Despite digital signal processing being widely adopted today, analog computing is attracting a renewed interest thanks to its ability to perform energy-efficient and massively parallelized computations.

The talk shows some promising avenues to devise new, faster, and more sustainable communication and signal processing architectures that marry the communication theoretic principles of modern digital communications with analog domain processing and computing paradigms such that information transmission, processing and computing are conducted faster with much lower computational complexity.

We introduce the concept of Microwave Linear Analog Computer (MiLAC) as a very general model of a computer exploiting the propagation of analog signals in a microwave network, offering exceptionally low computational complexity - unimaginable with conventional digital computers. We show that MiLAC can perform computation, e.g. matrix inversion and linear minimum mean square error estimation, with low complexity directly in the analog domain and enable new future MIMO communications with orders of magnitude lower computational complexity than digital processing.

Title: Programmable wave-domain computing for wireless communications

Abstract: Programmable wave-domain computing (pWDC) offloads processing burden from electronic processors to the wave domain, where signals are directly processed through reconfigurable wave–matter interactions. In this talk, I will discuss three contemporary challenges in the area of pWDC. First, I will address the prototype-aware optimization of real-life pWDC hardware. Second, I will discuss recent efforts to maximize the wave-domain flexibility of pWDC hardware. Third, I will discuss recent progress toward non-linear pWDC. Central to the entire talk will be an electromagnetically consistent system model of pWDC hardware based on multiport network theory. To ground my discussion in the experimental reality of real-life pWDC prototypes, I will describe how the model parameters can be estimated for real-life pWDC prototypes; examples will include various contemporary types of programmable metasurfaces. Moreover, I will emphasize how commonly neglected features of the model (mutual coupling between tunable elements; encoding of control vector in scattering characteristics of tunable elements), turn out to be valuable resources to tackle the discussed challenges.