Refine
Document Type
- Bachelor Thesis (1)
Year of publication
- 2015 (1) (remove)
Language
- English (1)
Keywords
Institute
- 01 Elektro- und Informationstechnik (1) (remove)
This Bachelor thesis deals with connected systems consisting of a multitude of similar electronic devices
(often referred to as agents) endowed with information processing abilities. It is required that these socalled multi-agent systems solve a certain task with a high reliability, while the individual components
are not able to solve the problem on their own in a satisfying manner. A central control unit can not or
shall not be used in such systems for a variety of different reasons: For example, a significant drawback of
a central control unit is the vulnerability of the system. If the central control unit fails, the whole system breaks down. Therefore, multi-agent systems require special algorithms enabling the agents to solve a
common, global problem in a suitable manner by local interaction only.
In this thesis distributed algorithms are investigated which can be used for distributed information pro-cessing and control of such multi-agent systems. In the first part of this work, it is assumed that each
agent posses a private information state about a common parameter of interest. The described consensus algorithm enables all agents to reach a system-wide identical information state by local information
exchanges only. Subsequently, it is considered the case that every agent has access to streaming data containing information about an a priori unknown parameter. The diffusion strategy described in the second
part enables the agents to estimate this parameter and to minimize a global cost function which depends
on it. Both algorithms are described in a general framework and can therefore be applied to a variety of
different problems. One application of these strategies, which is described in the third part of this work,
is the simulation of swarming behavior.