![]() | Only 14 pages are availabe for public view |
Abstract The field of Ad-hoc networks has seen a rapid expansion of visibility and work due to the rapid increase of inexpensive, widely available wireless devices and the network community’s interest in mobile computing. Such networks are taking special great attention from research community due to their wide range of applications. One of the most important requirements, in Ad-hoc networks, is to avoid the partitioning problem of the network that may happen due to dynamic movement or failure of special node(s) called ”Critical Node(s)”. A critical node refers to the node that its removal from the network will break the network into many partitions. Many researchers presented different algorithms to detect critical node(s) in ad hoc networks. Nevertheless, most of these algorithms did not take the situation where two nodes together may also cause network separation if they are moved or removed at the same time. Indeed, because of the expensive and difficulty of real experiments, simulation technique is the primary methodological framework for research and development of such networks. However, an important problem in simulation of ad hoc networks is how to generate a connected graph to represent the network. This thesis first presents a literature survey of the most recent methods that concern the generation of network graphs. Then, it introduces two novel and fast algorithms for generating topologies of Ad-hoc networks. The proposed approaches enable the user to generate various network topologies by deciding number of nodes, radio range and minimum distance between any two adjacent nodes in the graph. Finally, this thesis presents a new technique called Two Critical-Nodes Detection Algorithm (TCNDA) to detect both individual critical node(s) and each pair of nodes that are critical together in Ad-hoc networks. The proposed approaches are evaluated and compared with the most recent algorithms by simulation. Numerical results demonstrate that the proposed approaches speed up simulation of Ad-hoc networks and achieve an essential computational cost reduction in comparing with the most recent methods. |