ProCon - PrognÃstico de congestionamento de trÃfego de redes usando wavelets

AUTOR(ES)
DATA DE PUBLICAÇÃO

2004

RESUMO

LAN and WAN networks have showed a traffic behavior different from the commonly assumed in the traffic stochastic modelling. This new pattern of observed behavior motivates the use of self-similar stochastic processes for traffic modelling purposes. The behavior of the current networks traffic is chaotic, possessing statistical properties which include complex temporal correlations. In this scenario, a problem not yet solved is related to traffic congestion. The current congestion control algorithms implemented in the TCP protocol, for example, are reactive and based only on packets discard. There are preventive schemes of congestion control that treats with the problem at the output queues of the network nodes, using actives queues management mechanisms. However, they do not avoid the use of the discard policies. In the last decades, it has been observed that considerable research effort in network management has been concentrated on performance issues, mainly in preventive congestion control mechanisms. Main reasons for this is the growth of network applications, not only in amount but also in complexity, and to the number of users that use networks, particularly the Internet. This work proposes and implements a mechanism for forecasting possible congestions through the identification of the tendency component in time series that represent the network real traffic, using discrete wavelet transform. The proposed mechanism produced the wanted effect, since can detect possible cases of congestion

ASSUNTO(S)

congestionamento de redes engenharia de trÃfego modelo de trÃfego ciencia da computacao transformada de wavelet

ACESSO AO ARTIGO

Documentos Relacionados