Multi-Objective Optimization for Unmanned Surveillance Networks Using Evolutionary Algorithms

Open Access
Author:
Williams, Jonathan
Graduate Program:
Electrical Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
April 09, 2008
Committee Members:
  • Kwang Yun Lee, Thesis Advisor
Keywords:
  • evolutionary strategies
  • differential evolution
  • evolutionary algorithms
  • underwater vehicles
  • sensor networks
  • optimization
  • search path generation
Abstract:
The existing literature contains many control systems for wireless sensor and surveillance networks. In the majority of these systems, sensors are densely distributed and energy conservation is a secondary concern. It is the view of this work that a densely distributed sensor network would have an unacceptably high cost and is not practical for a large-scale underwater surveillance network. This thesis proposes a sparsely distributed underwater surveillance network combining autonomous mobile vehicles and fixedlocation sensor platforms. It is the premise of this thesis that the limitations of the sparsely distributed network can be overcome through vehicle mobility, properly placed fixed sensors, and network-level coordination and control. The control of this network is divided into two optimization problems. The first is the problem of initial target detection, termed directed search. This problem consists of creating vehicle search paths which maximize the probability of target detection while minimizing energy consumption. A two-tiered solution approach is presented which uses Differential Evolution and Evolutionary Strategies. The second problem is that of maintaining target surveillance, termed asset allocation. The problem consists of choosing which vehicles will engage in target surveillance and how the surveillance task will be divided. An optimization approach using Evolutionary Strategies is presented which maximizes hold time while minimizing energy consumption. Simulation results show the approach to be superior to a number of deterministic allocation algorithms.