Game-theoretic and Bio-inspired Techniques for Self-positioning Autonomous Mobile Nodes
Item
-
Title
-
Game-theoretic and Bio-inspired Techniques for Self-positioning Autonomous Mobile Nodes
-
Identifier
-
d_2009_2013:299e79dd5fb9:11328
-
identifier
-
11663
-
Creator
-
Kusyk, Janusz,
-
Contributor
-
Umit M. Uyar
-
Date
-
2012
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Computer science | bio-inspired algorithms | game theory | MANETs | node-spreading | topology control
-
Abstract
-
Autonomous mobile nodes that position themselves over an unknown terrain can ameliorate many of the problems which mobile ad hoc networks (MANETs) face. Achieving good spatial placement of mobile agents leads to superior network topology with improved area coverage, reduced power consumption, enhanced spectrum utilization, and simplified routing procedures. However, autonomous decision-making by nodes may also increase uncooperative and selfish behavior by these independent agents. Since it is impractical in MANETs to sustain complete and accurate information at each node about the entire network layout and decision of an individual node about its position should only be based on local information with limited coordination among agents. These characteristics recommend game theory (GT) as a tool for modeling, analyzing, and designing many MANET applications. At the same time, biologically inspired computation techniques such as genetic algorithms (GAs) can be used for finding desirable solutions in a prohibitively large search space and, in the case of MANETs, reduce the computational complexity needed for a node to determine its next location.;We introduce several novel approaches for autonomous MANET nodes to distribute themselves uniformly over a dynamically changing environment without a centralized controller or a priori information about the deployment terrain or the state of other mobile agents. Our methods combine concepts from traditional GT, evolutionary game theory, and bio-inspired algorithms to effectively and efficiently guide autonomous MANET nodes in finding the best positions. We show that myopic actions of each individual node lead the entire network towards a stable and uniform distribution. We present formal analysis of our methods, and prove their important properties including convergence, area coverage, and uniformity characteristics.;We developed a Java-based modeling platform to simulate performances of mobile networks, which allows for real-time visualization of ongoing network dynamics and collection of all critical data for evaluation purposes. Our analysis and experimental results demonstrate that GT and GA can be successfully combined for autonomous node placement to provide useful and resilient methods for optimizing network topology.
-
Type
-
dissertation
-
Source
-
2009_2013.csv
-
degree
-
Ph.D.
-
Program
-
Computer Science
