Mobility modeling and topology prediction in cognitive mobile networks

• Main Author: Alshehri, Abdullah
• Other Authors: Heydari, Shahram
• Language: en
• Published: 2012
• Subjects: Situational awareness; Localization; MANET; Filters
• Online Access: http://hdl.handle.net/10155/272

The purpose of this thesis is to analyze a non-intrusive connectivity visualization method for OLSR-based MANET topology in different mobility models. The visualization relies on the local topology databases (neighborhood database and topology database) available in OLSR nodes in the network. Two different views are considered in this method: central view and nodal view. In the central view, the network topology is viewed from a control center which has access to the databases of all nodes, while on the other hand, the nodal visualization provides a picture of the network topology from individual nodes point of view. In this thesis, the full view of the network has been compared to the nodal view to calculate the error rate for topology discovery, based on the total numbers of active and undiscovered links. The main contribution of this thesis is to analyze and improve the accuracy of coarse localization techniques under different mobility models, using the Force-directed algorithm to calculate the approximate location of the nodes. The localization information was gathered from layer-3 connectivity, utilizing anchor nodes that are equipped with GPS and other non-GPS nodes instead of using traditional methods that include received signal strength, time of arrival and angle of arrival. The approximate location information of the nodes derived from this technique has been compared with original node location in order to determine the accuracy of this technique. To improve the accuracy, several mobility prediction filters such as moving average filter, Kalman filter and low pass filter have been applied to the approximate location data. The simulation is done to calculate the error between the original location data and the coarse approximations, and the results shows that Moving Average provides the best results. === UOIT