Caracterização de Perda de Pacotes em Redes Wi-Fi com Modelos Ocultos de Markov em Camadas 1 e 2

In Wi-Fi networks, packet losses can occur due to different causes and are most often characterized by the occurrence of bursts. Markov chains are applied to model these losses, such as the traditional Gilbert-Elliot model that was designed to capture the temporal dependency of losses. Loss models are developed to capture temporal dependency characteristics, but usually considering the phenomena observable in the physical layer. However, it is observed that the access control in the link layer also has a great influence on packet losses. In this paper, a new packet loss model for IEEE 802.11b/g/n networks is proposed based on the grouping of SNR characteristics and channel occupation using Hidden Markov Model (HMM). Results show that the use of channel occupation is fundamental for modeling packet losses. When considering this factor, the proposed model presents better adherence to the real network operation situation and can be used to improve computational simulations and also to develop error recovery strategies.