- This event has passed.
Seminar: Gaming on the Edge: A Performance Framework towards Edge-AI
Welcome to the second TECoSA internal seminar this fall!
The purpose of this seminar series is to keep each other informed about ongoing work at the partners and to have technical discussions. The plan is that the TECoSA postdocs will present at the first four seminars.
Abstract: Edge computing is revolutionizing gaming and streaming services by processing data closer to the user, which greatly reduces latency and enhances bandwidth efficiency. This talk delves into the integration of edge computing with online gaming, focusing on the Gaming as a Service (GaaS) paradigm. Through an analytical model we show that edge-based gaming architectures can outperform traditional cloud gaming solutions in terms of response times. While developed in the context of gaming, this performance model offers insights that may extend to a broader class of applications. The underlying challenges of distributed processing, low latency, and high-performance computation are not unique to gaming but they are also central to emerging Edge-AI systems, where machine learning applications meet edge infrastructures. We will reflect on the parallels and distinctions between these domains and consider how such frameworks might inform future analyses of AI applications operating on distributed infrastructures.
Speaker: Diletta Olliaro, KTH, TECoSA Postdoc
Bio: Diletta Olliaro is a Postdoctoral Fellow at KTH Royal Institute of Technology, working within the TECoSA research center. She received her PhD in Computer Science from Ca’ Foscari University of Venice in March 2025, with a thesis titled “Models for Throughput Maximisation in Distributed Systems”. Her research interests include queueing theory, product-form models, and the stochastic modelling of computer and communication networks, with a focus on performance and reliability evaluation. During her doctoral studies, she had the opportunity to work as a visiting PhD student at IMDEA Networks Institute, PUC-Rio, and Imperial College London.