Técnicas de otimização não-linear aplicada a algoritmos DSM

Abstract

This M.Sc. thesis presents state-of-art algorithms for dynamic spectrum management (DSM). The strategy and description of each DSM algorithm are presented and briefly discussed as well as a comparative analysis between them for better understanding. This analysis was conducted considering the rate gain (kbps) obtained from computer simulations and the results are presented by tables and rate region curves. Additionally, it was conducted in laboratory a set of direct and coupling transfer function measurements. Later on, those measurements were used during DSM algorithms simulations; this dissertation focused in the IWF, ISB and SCALE algorithms. The results obtained from the transfer function measurements showed better performance compared to the results with transfer function 1% worst-case model. The later considers 1% of pairs present the biggest crosstalk level in all transfer function frequencies. In the case where standardized channels were employed, ISB and SCALE obtained similar performances whereas IWF had a performance quite similar to SSM. On the other hand, these three algorithms had similar gain in simulations evolving the measured channels because the low crosstalk level.