CE Distinguished Seminar by Prof.Aylin Yener

6G vision is quickly emerging as one of massive and complete connectivity, and of true convergence of communications, computing, and AI; enabling unprecedented applications in transportation, medicine, agriculture, and digital access for all. Wireless connectivity is at the heart of this vision: judicious design and utilization of the unique opportunities the wireless medium provides are becoming essential once again, much like in ahead of the first mobile revolution, but now fueled by a nexus of advances from across-the-board, electronics to machine learning. 6G, with its envisioned programmable open network architectures, is expected to be the first generation where fundamental innovations at the wireless edge can truly be integrated into the networked system design, shifting the network intelligence to the edge. In this talk, we will provide an overview of our activities and present our recent results in two directions relevant to this paradigm, inspired by the unique features and challenges of interacting in the wireless medium. First, we will consider semantic communications, an emerging paradigm that aims for transceiver designs to convey the meaning of information needed at the receiver. Implementing judicious designs that guarantees reliable communication of semantics, this paradigm promises efficient use and reuse of wireless resources and is particularly fitting for 6G where dynamic and programmable transceiver design and AI-guided strategies are expected to be the norm. As an example of this approach, we will present some of our recent results in semantic classification utilizing pre-trained models that have comparable performance in wireless channels to conventional communications while providing orders of magnitude savings in transmissions. Next, we will consider joint communications and edge learning by wireless edge devices. Focusing on over-the-air federated learning (OTA-FL) in realistic wireless environments, we will demonstrate that jointly optimizing communication (power) and learning (computation) resources of the edge devices in concert with programming the wireless environment (using reflecting intelligent surfaces) facilitates distributed learning effectively handling data heterogeneity. We will conclude by discussing open directions in 6G.