Trajectory and Data Planning for Mobile Relay to Enable Efficient Internet Access after Disasters

Our experiences in East Japan Earthquake show that disasters will cause a large-scale network interruption due to serious damage of existing network infrastructures. To enable Internet connection before network service restoration, which is usually time-consuming, we propose to use a mobile relay to...

Full description

Saved in:
Bibliographic Details
Published in2015 IEEE Global Communications Conference (GLOBECOM) pp. 1 - 6
Main Authors Anazawa, Kazuya, Li, Peng, Miyazaki, Toshiaki, Guo, Song
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2015
Subjects
Biological cells
Genetic algorithms
Internet
Mobile communication
Mobile computing
Relays
Trajectory
Online AccessGet full text

Cover

More Information
Summary:Our experiences in East Japan Earthquake show that disasters will cause a large-scale network interruption due to serious damage of existing network infrastructures. To enable Internet connection before network service restoration, which is usually time-consuming, we propose to use a mobile relay to carry data for several isolated communities formed after disasters. Specifically, we consider that only one community has the Internet connection, and data from Internet need to be carried to other communities by the mobile relay. The data downloading performance of each community is evaluated by the utility of obtained data minus the penalty of corresponding latency. With the objective of maximizing the poorest performance among communities, we formulate a max-min problem to optimize the trajectory of the mobile relay and its carried data volume for each community. Due to the NP-hardness of this problem, we propose a genetic algorithm by representing the trajectories of mobile relay as chromosomes that evolve to approximate the optimal solution. The fitness of each chromosome is evaluated by optimizing the data volume carried for each community. Extensive simulations are conducted to show that our proposed algorithm significantly outperforms existing algorithms.
DOI:10.1109/GLOCOM.2015.7417170