Data Synchronization Method among Isolated Servers Using Mobile Relays

Kazuya ANAZAWA  Toshiaki MIYAZAKI  Peng LI  

IEICE TRANSACTIONS on Communications   Vol.E101-B   No.10   pp.2239-2249
Publication Date: 2018/10/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.2017EBP3468
Type of Manuscript: PAPER
Category: Wireless Communication Technologies
mobile relays,  data synchronization,  disaster resilient network,  network isolated environments,  

Full Text: PDF(2MB)
>>Buy this Article

After large-scale disasters, information sharing among people becomes more important than usual. This, however, is extremely difficult to achieve in disaster zones due to serious damage to the existing network infrastructure, power outages, and high traffic congestion. For the quick provision of alternative networks to serve heavy communication demands after disasters, establishing local area networks (LANs) consisting of portable servers with data storage has been considered as one of the most promising solutions. Based on the established LAN and a data server in each area, people can share many kinds of disaster-related information such as emergency information and supply/demand information via deployed neighboring servers. However, due to the lack of stable Internet connection, these servers are isolated and cannot be synchronized in real time. To enable and guarantee more efficient information sharing across the whole disaster-hit area, data stored on each server should be synchronized without the Internet. Our solution is to propose an intermittent data synchronization scheme that uses moving vehicles as relays to exchange data between isolated servers after disasters. With the objective of maximizing the total number of synchronized high priority data under the capability constraints of mobile relays, we first propose a data allocation scheme (DAS) from a server to a mobile relay. After that, we propose a trajectory planning scheme for the relays which is formulated as a Mixed Integer Linear Fractional Programming (MILFP) problem, and an algorithm to solve it efficiently. Extensive simulations and comparisons with other methods show the superior performance of our proposals.