Files in this item



application/pdfECE499-Fa2016-zhang-Xuanyao.pdf (2MB)Restricted to U of Illinois
(no description provided)PDF


Title:Study on Distributed Optimization in a Lossy Network
Author(s):Zhang, XuanYao
Subject(s):distributed optimization
cost function optimization
lossy network
Abstract:This research is working toward evaluating an algorithm on distributed optimization. The environment is a synchronous multi-agent network with unstable connections. It is synchronous in the sense that there is a centralized time tracker that controls when a time segment, designated as “round”, begins and when it ends and the next segment begins. The links between agents, designated as nodes in the following, are unstable, preventing nodes from minimizing their cost function in a timely manner. We establish this network with a tool -- A Framework for Evaluating Iterative Algorithms on Distributed Systems implemented by a previous student that can generate multiple nodes represented by instances of a Java Class. They have their own variables and can communicate to each other in each round and the link is guaranteed to be stable. To simulate an unstable network, a random function is introduced to randomly drop packets, thus imitating link failures between one node and the others.
Issue Date:2016-12
Date Available in IDEALS:2017-08-17

This item appears in the following Collection(s)

Item Statistics