Distributed optimization on a wireless sensor network testbed

The focus of this thesis is to implement various distributed optimization algorithms on a physical wireless sensor network. Distributed optimization refers to optimization of some global function which is not completely known to any single node in a communication network. The global function is some combination of local functions that are available at each node. Therefore the objective is for all nodes to achieve consensus on the global optimum given only local information and communication with neighbors. Algorithms from the literature that address this problem in different set- tings are introduced, focusing on an incremental subgradient-based algorithm and a broadcast, gossip-based algorithm. These algorithms are applied to lo- calize a light source. This localization problem is formulated as a distributed optimization problem in which the global optimum is the true location of the source, and the local information is comprised of light intensity measurements at each node. Simulation results and results from physical implementations on the testbed are presented for the two different approaches. A modified version of the broadcast algorithm is also presented, and is shown to be supe- rior to the unaltered algorithm in certain settings via simulation and testbed results.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical & Computer Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Venkatesan, Neeraj

• sensor
• wireless sensor network (wsn)
• optimization
• distributed
• network
• localization
• least-squares

Statement dc:rights

• Copyright 2015 Neeraj Venkatesan

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/78664

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/78664