Decentralized workload balancing for edge-cloud computing networks
Abstract
This project concerns fundamental research on mathematical models for current and future distributed computing infrastructures. In recent years, the cloud computing paradigm has become prevalent, where computation becomes an external resource to be summoned in real-time from a remote server infrastructure supported by high bandwidth communication. This trend fundamentally changes the operations of companies and end-users, who may offload computation to a third party and contract dynamically to serve varying computation needs. However, remote processing at the cloud level has limits- many emerging applications (e.g., Internet of Things deployments) are delay-sensitive and require low latency responses from the computing infrastructure; this motivates the edge computing paradigm, which are smaller-scale computing deployments located near end-users on the edge of the network. Optimizing such edge-cloud computing infrastructures calls for new fundamental research. Cloud computing has already brought about new basic research in networking- in contrast with the traditional paradigm in which service capacity is fixed and centralized, here the computation may be dynamically right-sized and load balanced between a large number of servers residing in different computer clusters; task scheduling within each cluster also raises new challenges, due to the distributed localization of the data. Our group has made fundamental contributions in recent years through convex optimization, control, and queueing theories, partially supported by AFOSR. Theoretical results lead to resource allocation algorithms with provable performance. With edge-cloud infrastructures, new challenges appear- latency becomes a crucial metric; the communication network structure cannot be abstracted completely; the locality of data and devices calls for decentralized solutions. In this project, we aim to develop fundamental theories for solving these problems through our work and establishing new collaborations with the US research community in this very active field.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 06, 2024
- Source ID
- FA95502310350
Entities
People
- Fernando Paganini
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- Universidad ORT Uruguay