Efficient bandwidth allocation for QoS guarantee in multimedia networks

Abstract

As the demand for new wireless services has been explosively increasing, it is expected to support various multimedia applications with guaranteeing Quality of Service (QoS). Different multimedia applications have different traffic characteristics, and accordingly have different QoS requirements. For example, real-time multimedia applications such as voice-over-Internet protocol (VoIP) are delay sensitive. So, these applications can tolerate some packet losses, but have stringent delay constraints. Accordingly, for these applications, it is important not to suffer from excessive delay and contention with other applications for QoS guarantee. On the other hand, non-real-time multimedia applications, such as e-mail and data file, can tolerate a little delay but required reliable packet transmission without loss. So, for these applications it is important to guarantee reliable packet transmission. Therefore, providing QoS guarantee which is appropriate to traffic characteristics has been considered as a challenging issue for multimedia applications. In order to guarantee QoS requirements of the service traffic, it is crucial to know that how much bandwidth should be allocated for the service traffic. In addition, if we can estimate the pertinent amount of bandwidth and predict the performance, we can utilize the limited network bandwidth more efficiently. Therefore, in this dissertation, we investigate efficient bandwidth allocation methods to satisfy QoS requirements for the three kinds of service traffic, long-range dependence (LRD) traffic, video telephony service traffic, and voice-over-Internet protocol (VoIP) service traffic, which have drawn attention as the essential multimedia applications. For a LRD traffic, we propose a new method to adaptively allocate bandwidth according to the amount of input traffic volume. In the proposed method, we divide the input process into two sub-processes, a long time scale process and a short time scale process....