Estudio de modelos de movilidad realistas usando métricas de teoría de grafos y su impacto en redes inalámbricas de malla

Abstract

Wireless mesh networks, WMNs, are networks with the ability to self-form, self-configure, and self-repair. These features were inherited from the ad-hoc mobile networks, MANETs. However, WMNs appear as an alternative to the use of MANETs due to their differences in architecture and radio technologies used. Within WMNs, three paradigms of architecture stand out: WMNs with infrastructure, WMNs clients and WMNs hybrid (as the combination of the two previous architectures). Furthermore, WMNs, having radio technologies based on the 802.11 standard, are more convenient than the MANETs for routing information to the Internet. Both MANETs and WMNs make use of mesh routers, MRs, for routing information. The large number of mobile devices introduced in recent years has forced the creation of new topological architectures based on WMNs. A clear example of modern architectures is the so-called WMN spontaneous. This type of architecture is characterized by being made up only of mobile devices or end user nodes. Within this type of architecture. The main characteristic is that all nodes act simultaneously both as user interface devices as well as traffic routers for their peers. However, in order to optimize and make more efficient the operation of the network. through topology control techniques, only a certain number of devices can be selected to act as the network routers. This selection process classifies the nodes into two groups: the MRs and the client nodes of the mesh network MCs. Thus, the routing of the traffic will be controlled only by the MRs. While the MCs will only act as end-user devices Studies such as [1, 2] have shown improvements in the performance of spontaneous WMNs networks, through the use of topology control. This technique was carried out by combining two graph theory metrics, betweenness centrality, and modularity. These two metrics were used to abstract the information from the topology of the client mesh networks. In order to select only the most important nodes in the network as routers. Through this topology control significant improvements in overall network efficiency are achieved over a network in which all devices act as routers. On the other hand, it is important to emphasize that being made up of end user devices, spontaneous WMNs generate dynamic scenarios within their topology. This topology is constantly changing as users move from one location to another. For this reason, mathematical models have been established in the literature that attempt to emulate, in the most realistic way possible, the dynamic behavior of nodes on the network. These models have been called mobility models. In a previous study [2], these realistic mobility models have been evaluated from the perspective of efficient topology control. Under these premises, the main objective of this work is to find and characterize the topological differences within six realistic mobility models available in the literature and applied to the evaluation of wireless mesh networks. These differences were established using six different graph theory metrics. Using the concept of spontaneous WMNs, networks were simulated and evaluated in which the topologies changed according to the chosen mobility model. In the end, conclusions were established that indicated which mobility models presented the most relevant results within each metric. Additionally, to illustrate an example of application of the analyzes carried out to mobility models. The concepts of network integrity and robustness were used. For this, two mobility models were analyzed on each metric. In each model, nodes were eliminated simulating a network attack. In order to evaluate the impact on the packet delivery rate, PDR, resulting each time a node was eliminated. It could be observed that not all the metrics delivered satisfactory results, so the application of network robustness based on these metrics specifically was discarded. Complementarily, for these metrics an analysis of PDR and delay was made simulating networks with a topological structure based on communities. Finally, conclusions were established on each mobility model and each metric. Also, it was indicated the most relevant results obtained on each one. This work has been carried out using different open access software tools widely used in the scientific community. For the mobility traces, BM and SUMO have been used. For the simulation of the WMNs with information traffic, the discrete event network simulator NS-3 has been used. The graph theory metrics have been obtained using Gephi software. Lastly, the comparisons between the mobility models and the metrics have been obtained using the Python platform