Tesis doctoral de Marisa Da Silva Maximiano
The frequency assignment (fap) is a problem that has risen in the telecommunication area in the last decades, due to the enormously growth of the mobile users. That is the reason why it has been so important, especially in the gsm (global system for mobile communications) networks since it represents one of the most important and critical tasks for current (and future) mobile communication operators. the main problem in fap is that in gsm networks a fixed frequency spectrum is available, but it is also very scarce. Therefore frequencies need to be reused throughout the network, and consequently, interferences may occur and some separation constraints may be violated. Therefore, the fap aims to minimize these unwanted occurrences, in order to provide the network users with a good quality of service. Due to the usable range of frequency spectrum being very limited, the fap becomes very important to guaranty the traffic performance on mobile communications. As a result, in the gsm technology fap is under the most relevant and significant problems. in the above context, the need to develop new strategies in order to address this problem becomes extremely important. Generally speaking, the key ambition is to study and develop different approaches based on new and also hybrid metaheuristics. In fact, to deal with the realistic problem we work with a complex mathematical formulation, which takes in consideration the requirements of real-world gsm networks. The main contribution of this work is a study developed to solve fap using distinct evolutionary algorithms (ea). Specifically, it considers the use of population-based metaheuristics and trajectory-based metaheuristics. eas are a class of stochastic search and optimization methods. Their algorithms are based on the principles of natural biological evolution, and have received an increasing interest in the last decades. Eas encompass a range of different methods inspired in the natural evolution, including genetic algorithms, genetic programming, and also other related techniques, e.G., Differential evolution (de), etc. first of all, several already known metaheuristics have been designed and adapted to be analyzed with a single objective formulation of fap. Initially,our goal was the application of several metaheuristics considering only the level of interferences occurring in the network. Following this, and considering that solving real-life engineering problems it is not an easy task because normally they represent a multiobjective optimization scenario, a more complex formulation is being investigated. Indeed, with fap, two different and conflicting objectives that characterize a multiobjective optimization (mo) problem can be considered, in which an assignment of frequencies is sought to a high number of transmitters in a way as efficient as possible. Therefore, we have designed a more complex formulation of fap considering it as a multiobjective optimization problem. This new multiobjective formulation of fap considers as objectives to improvement of the interference costs and the separation costs. although we have started with a single-objective formulation, our main contribution in addressing fap was focused on comparing different multiobjective metaheuristics, specially comparing population-based metaheuristics against the ones based on trajectory. Evaluating the performance of multiobjective algorithms is far from being a trivial task. Therefore, we have incorporated several complementary measurements to assess the quality of the accomplished solutions. First of all, the validation of results uses the common statistical comparisons and after that we used the hypervolume indicator that defines the volume of the objective space dominated by the pareto front. Besides the hypervolume, the coverage relation was also used as additional metrics in order to assess which the best pareto front is. two real-world instances of fap, currently being operated, will be used in the experiments scenario. Comparisons with other well-known algorithms are also performed in order to assess the efficiency of our approaches.
Datos académicos de la tesis doctoral «Aplicando metaheurísticas multiobjectivo al problema de asignación de frecuencias en redes gsm«
- Título de la tesis: Aplicando metaheurísticas multiobjectivo al problema de asignación de frecuencias en redes gsm
- Autor: Marisa Da Silva Maximiano
- Universidad: Extremadura
- Fecha de lectura de la tesis: 26/09/2011
Dirección y tribunal
- Director de la tesis
- Miguel ángel Vega Rodríguez
- Tribunal
- Presidente del tribunal: Juan manuel Sánchez pérez
- Juan Antonio Gómez pulido (vocal)
- victor Manuel Basto fernandes (vocal)
- oscar Gutierrez blanco (vocal)