A Quasi Stationary Service Architecture for Network Monitoring and Connectivity Prediction in Aeronautical Ad Hoc Network Using Fuzzy C Means Clustering

Description
Title: A Quasi Stationary Service Architecture for Network Monitoring and Connectivity Prediction in Aeronautical Ad Hoc Network Using Fuzzy C Means Clustering
Authors: Soumi, Ghosh
Date: 2014
Abstract: An Aeronautical Ad Hoc Network (AANET) of airborne elements is a high speed mobile network. The AANET has a 3D topology spread across the airspace. The high ground speed of the airborne elements changes the network topology rapidly. This makes AANET highly dynamic in nature. Upholding the connectivity in the network in such dynamic environment is a challenge. The connectivity in the network is primarily in uenced by proximity of airborne elements to each other and their relative velocities. Once an airborne element gets disconnected from the network, it becomes completely oblivious of the network scenario in its neighborhood. In the absence of a monitoring agent in the airspace, a disconnected member of the network largely depends on the ground infrastructure and satellite resources for immediate information regarding its surrounding region. Network monitoring in dynamic environment of AANET is a challenging task, mainly due to the mobility of the airborne elements. We propose an intelligent network monitoring system for AANET. Disconnected members of the network are directed towards region in the airspace with "higher" connectivity and "minimum" traffic under the monitoring system. Our monitoring system depends on a quasi stationary layer of Higher Altitude Platform (HAP) in the airspace. The primary focus of the monitoring system is to mitigate disconnectivity in the AANET. The monitoring of the network is achieved by a periodic monitoring scheme in every HAP. The proposed HAP monitoring system aims at making the AANET more independent of the ground infrastructure and satellite resources. We also reckon Fuzzy C Means (FCM) data clustering as a means to monitor changes in network topology and traffic in the AANET. The FCM clustering is an integral part of our monitoring scheme. Our simulations demonstrate that FCM clustering can efficiently track the network changes in the AANET and identify regions of connectivity in the network.
URL: http://hdl.handle.net/10393/31495
http://dx.doi.org/10.20381/ruor-6627
CollectionThèses, 2011 - // Theses, 2011 -
Files