Satellite networks service an important part in wireless communication systems by supporting large amounts of voice, data, video traffic and etc. Satellite communications have a unique advantage that other communication systems cannot support. Because of high altitude of a satellite, it connects various terrestrial terminals which are deployed in a wide area. This feature also enables a satellite to connect terminals in sparsely deployed areas that are difficult to install other means.
In this thesis, several attempts are made to enhance a performance of satellite communications. The main contributions of the thesis are as follows.
First, we propose a new resource prediction method for the Demand Assigned Multiple Access (DAMA) scheme in satellite networks. An inaccurate prediction of future traffic causes a degradation of QoS and utilization due to the long propagation delay in satellite networks. The Dynamic Leveling Scheme(DLS) use a leveling method to modify its prediction to a discrete one to change the precision of the prediction result. This new scheme has two features: 1) It enhances the probability of successful prediction and 2) it can be applied to any type of existing prediction method.
Second, we devise a distributed resource sharing method to overcome a propagation delay of satellite link. In the Dynamic Bandwidth Sharing (DBS) scheme, terminals check a utilization of a granted resource and share their idle resource to other terminals that need more resources. In other to quickly negotiate between STs(Satellite Terminals), the DBS uses local networks as a control channel for the resource share instead of satellite link. Local networks support direct communication between STs. Because local networks are just responsible for control messages, they do not have to support a high capacity but needs a low transmission delay for fast negotiations.
Finally, we evaluate a performance when both DLS and DBS schemes are adopted in satellite networks. The DLS overcomes the limitation of the static leveling scheme and an inaccuracy of the prediction algorithm. However, a low level increases the block size of DLS and degrades the utilization of satellite link. The DBS enhances a performance of request/grant process of DAMA. However, its performance is influenced by a slot length of local networks and it must use local networks as a control channel. In addition to, terminals need neighbor terminals, which have an idle resource, to use shared resources of them. The DBS shares the idle resource caused by the low level of the DLS and the DLS reduces a probability of prediction failure regardless of local networks.