No abstract

Network dimensioning is becoming important in order to provide a stable Quality of Service (QoS) to the customers. However, it is not an easy task because the Internet traffic changes dynamically and frequently.We thus need a new approach for network dimensioning suitable to the Internet. Probably, it would be quite different from the existing well-established method for the telephone network. Actually, our method proposed in this thesis tries to establish a new paradigm of network dimensioning for the new communication era of the Internet.In this paper, we consider the network dimensioning approach in order to resolve the bottleneck for stable and QoS-rich data communications in the Internet. For that purpose, a possible cause of the bottleneck must be examined not only within the network, but also at the end hosts, which originated in an intrinsic nature of the Internet architecture. In our study, we first propose a method to identify the bottleneck for improving the performance of the end users. Once the network is found to be bottleneck, the network operator should determine how much the network resources is increased to meet the end user's QoS requirements. So, we next focused on the measurement method for utilization of the bottleneck link between end-to-end hosts. To measure the 1 utilization of the bottleneck link accurately, we propose a new measurement method that can eliminate several kinds of measurement errors. We collected the measurement results from the actual operating Internet, and provide an evidence to support an appropriateness of our approach.Based on the measurements, we finally propose a design framework in order to determine the adequate link capacity for offering a desired QoS level to the end users.

The divergence of people's demands and technologies has been fueling the proliferation of vertically integrated "silo type" systems. In order to maintain systematic simplicity, a common layer is necessary, which is used to quickly develop new applications. This paper discusses the requirements of the service delivery platform (SDP) as the common layer, and proposes our SDP named SPACE. The architecture of SPACE is based on the SOA concept, which provides reusable service elements and an open interface for applications. SPACE includes layered service elements and a messaging function to connect the service element and application. Other key technologies are consistency management and resource control. SPACE creates new values for many players, based on information over a wide area and networking capabilities. However, sometimes the requested application processes incompletely because of the instability of the wireless communication path and the conflict of the device control. Consequently, the SPACE resource will be wasted and additional delay will be caused. To overcome this issue, we propose a resource control mechanism which controls the processing order of service elements. Evaluation results show that the mechanism can reduce processing time in the entire platform system.