- 1 -
1. INTRODUCTION
Simulation modelling is an important tool for the study and evaluation of the behaviour of wireless networks. The use of simulations provides several benefits, for example, compared to real tests, it allows an easier what-if evaluation, with the possibility of easy choice and change of parameters.
Furthermore, simulations provide a compression/expansion of time for a more detailed analysis of the behaviour of the system. These and other factors have contributed to simulations becoming a very powerful tool for the study of networks and the development of new technologies.
A network simulator requires the user to define the topography of the network that has to be studied, creating the nodes and the links between them. In the evaluation of wireless networks, where a node is free to move around the simulation area, the simulator requires also a mobility model to apply to the nodes.
A mobility model is simply a series of algorithms that manage the movement of the nodes during the simulation. Even if this issue can appear “secondary” in the study of a network, it has been proved that simulation results are strongly influenced by the mobility model chosen, as shown in [2], [3] and [13].
- 2 -
In the real world, the movement of a node is influenced by several factors. Usually, a mobile node has to face the presence of obstacles and the path chosen to reach the destination is typically not random. In fact the node decides its path thinking about avoiding obstacles and, usually, covering the shortest way to the chosen destination. For instance, in indoor environment, when a node wants to move from a room to another, it just follows the shortest path to the chosen room along the corridors without stopping.
In simulation, unrealistic and too simplified models will likely produce unrealistic results. A number of mobility models have been already proposed in the past. Some of these create a random uncorrelated movement in a simulation area without obstacles. They are not accurate, but they are usually preferred because they are easy to implement and they also produce shorter simulation times. Some other models consider the presence of obstacles and the use of pathways graphs. They are more accurate, but also more complex and they take more time to run.
This thesis proposes an improvement of the mobility model introduced in [4], that offers the chance to incorporate obstacles and pathways in the simulation area in order to create
- 3 -
more realistic movements. The new model proposed, called Combined Mobility Model, has been designed for indoor environments, but it can easily adapted to mimic outdoor scenarios as well. It first acquires information about the simulation environment loading an image file with the map of the obstacles and another file for the pathway graph. Then it is possible for the user to decide if the nodes have to move: a) according to a random walk, b) along the pathway, or c) in a hybrid movement. In the first case, the movement is the same as the Random Waypoint Model (see Section 2.2.2), but the nodes have to avoid crossing the obstacles. In the second case, the nodes choose a destination vertex in the pathway graph, and they move towards that location using the shortest path. In the third case, the nodes switch from one of the two previous models to the other. That is, they use the Random Waypoint Model, but, sometimes, they decide to move to different areas using the pathway graph. Then they start moving in Random Waypoint again.
The presence of obstacles affects not only the movement of the nodes, but also the signal propagation. In our model the attenuation of a signal between source and destination depends
- 4 -
not only on the distance of the two nodes, but also on the material properties of the obstacles placed in between.
The evaluation of the Combined Mobility Model has been done calculating the parameters defined in [5]. The results of two of these parameters are also validated by mathematical analysis, studying a very simple scenario.
The remainder of the thesis is organized as follows: Chapter 2 shows briefly the most important mobility models already existing, explaining their advantages and drawbacks. Chapter 3 describes the mobility model presented in [4] and, motivating the choice of its improvement, introduces the Combined Mobility Model, explaining its functioning in details. Chapter 4 shows the evaluation of the model, studying the results of the simulation. Chapter 5 is dedicated to the mathematical demonstration of some of those results, and Chapter 6 presents future work and conclusions of the thesis. The C++ Source Code of the model can be found in Appendix A, while Appendix B contains the solving of some integral equations used in Chapter 5.
