Abstract :
[en] Mobile ad hoc networks, also called MANETs, are self-organised networks spontaneously created between neighbouring devices that do not need any infrastructure. The possibility of deploying a communication network at any moment and at any place opens a wide range of potential scenarios where the use of an ad hoc network is extremely important and useful. For example, we can imagine an area with no infrastructure, due to a natural disaster (flooding, earthquake, etc.), a war damage or simply a remote location. In such situations, the deployment of an ad hoc network for rescuing or evacuating people is key. %Allowing the communication between groups located in distant areas may be very helpful in critical situations in order to coordinate and reorganise them. However, ad hoc networks are not restricted to safety of disaster situations, they can also be used for gathering sensed data and send it to a remote server, exchanging safety road messages between vehicles or simply for infotainment, etc.
Efficient communication protocols are very important for providing good services, but in ad hoc networks is very challenging. It is necessary to deal with packet loss, collisions, mobility, network partitions, fading, energy constraints, obstacles, etc. Due to the intrinsic broadcast nature of wireless medium, dissemination algorithms are one of the most appropriate protocols for communicating devices. Additionally, broadcasting is one of the main low level operation as many applications and even other protocols rely on its service.
Therefore, in this thesis, we are proposing two different broadcasting algorithms for efficiently disseminate messages in the network. The first approach relays on an underlying tree topology, while the second is an energy aware algorithm that tries to reduce the energy consumption. In these unpredictable and changeable networks, protocols need to adapt to the current circumstances and this adaptability is usually obtained using thresholds. The values of these thresholds directly influence the behaviour of the algorithm. Finding them is a complex and critical task as protocols are usually very sensitive. We tackle this problem using approximation algorithms or metaheuristics, more specifically, we use a parallel iterative local search and two evolutionary algorithms. In order to ease this process, we have created a modular experimental framework for evaluating, validating and optimising any algorithm in MANETs.
