GENETIC ALGORITHMS OPTIMIZE IMAGES TO MEET BANDWIDTH CONSTRAINTS FOR A SENSOR NETWORK

Abstract

Wireless Sensor Network (WSN) transmitting large amounts of data are constrained by bandwidth and the number of nodes in a system. This thesis demonstrates that WSNs can optimize images by using genetic algorithms in a closed-loop system to meet bandwidth constraints for a WSN. To meet prescribed bandwidths, genetic algorithms were designed to generate a solution set of optimal images based on compression and down-sample values. Multi-criteria decision Analysis (MCDA) was used to select a specific resultant image which meets prescribed bandwidths when transmitted over the network. A working system was modeled and simulated to host a central server and four drone nodes. The model was designed in a closed-loop configuration as drones navigate an operational area over various terrain types. Results from the simulation demonstrated the integrated genetic algorithms were able to calculate a solution set, select a unique solution, and transmit the image in real time for each drone at predefined rates to meet bandwidth requirements.