Predictive resource allocation for real time video traffic in broadband satellite networks

摘要

Satellites are attractive for broadband networks because they provide fairly hgih bit-rate connections over a wide area. However, satellite networks are characterised by having a fixed amount of shared bandwidth between all users, and by relativel ylarge propagation delays. These problems pose a challenge when implementing a high performance dynamic resource allocation scheme to effectively share the bandwidth among a large population of Earth terminals, with heterogeneous mix of traffic and different QOS requirements. This paper proposes the implementation of a predictive resource allocation scheme for the real time video component of the traffic in a broadband satellite network environment. Two predictive implementations are investigated: one based on time series analysis, and the other based on neural networks.