Predictive traffic estimation for uplink resource allocation in satellite communications

In this paper, we propose a predictive traffic estimation strategy for the resource allocation to uplinks in the satellite communication. Estimating the traffic arriving during the next epoch is an important part of the dynamic bandwidth resource allocation (DBRA). Based on this estimate, the terminals request the bandwidth resources for transmission during the next epoch or next several epochs. This problem becomes critical when traffic from different quality of service (QoS) require different traffic treatments, and the constraints of the delay and jitter become crucial. This necessitates that such traffic, which are sensitive to delay & jitter, not be queued at all, with dedicated resources assigned to them. At the same time, it is also necessary that the available bandwidth be efficiently used, if such traffic is not active and the lower class traffic needs band-width. Many of the available techniques estimate the traffic requirements based on the queue occupancy and the rate requirements. Reserving precious bandwidth for the highest QoS traffic may result in wasting the bandwidth and may also result in swelling queues of lower priority classes. Furthermore, allocating the resources based on the queue size only may not be enough to give the right estimate of the arrival rates. This study explores the applicability of the predictive estimation techniques for the demand estimation in a QoS environment.