Prof. Uysal gave a seminar at Chalmers University

The seminar, titled Goal Oriented Communication for the Scale-up of MTC, was a part of the CS Distinguished Seminar Series organized by the Chalmers University of Technology, Sweden. You can find more detailed information about the seminar below:

Prof. Dr. Elif UYSAL

Goal Oriented Communication for the Scale-up of MTC

SLIDES (you can click to view the slides)

One of the ways that 6G is envisioned to differ from previous  generations of communication standards is its native and massive  support for remote computations, intelligence and decision making. To  achieve the scalable growth of such applications we need communication  networks to send only the right pieces of information to the point of  computation at the right time. This is within the concept that we  define as Goal Oriented (GO) communication, also referred to loosely  as “Semantic” communication in certain contexts, as it calls for a  departure from classical communication theoretic problem formulations.  Moreover, a GO approach calls for a departure from traditional layer  functionalities. In particular, through removing the assumption on the  exogeneous arrival of data, this approach couples classical  application layer functionalities in the link layer operation. Toward  practical algorithms that will feasibly place GO communication into  products and standards that can revolutionize the network capacity in  MTC, a first order yet effective approach has been optimization for  data freshness metrics.  In this talk will highlight a number of  examples of using such metrics in recent research and implementations.  We will show the effectiveness of designing for basic timeliness  metrics such as the Age of Information (AoI), Query Age of Information  (QAoI) at the physical, link, medium access, transport and application  layers, and exhibit resulting innovations in scheduling, random  access, and congestion control. We will close by highlighting recent  work moving forward from freshness to more advanced GO performance  metrics such as end-to-end inference error.