Integration of Streaming and Elastic Traffic in a Single UMTS Cell: Modeling and Performance Analysis

Various papers that study the integration of elastic and streaming traffic have been published recently [1, 2, 3, 4, 5, 6]. In terms of bandwidth sharing policy, the classical approach is to give head-of-line priority to packets of streaming flows in order to offer packet delay and loss guarantees [1, 2, 4]; alternatively, adaptive streaming flows (that are TCP-friendly and mimic elastic flows) are considered in [3, 5, 6]. In terms of modeling approach, while Markovian models have been developed in [7] for the exact analysis of the integrated-services system, they can be numerically cumbersome. Hence, a fluid model is proposed in [2, 3, 4, 5, 6] to provide closed form limit results and approximations. These results can serve as performance bounds, and hence yield useful insight. In this study, we consider an integrated-services system (in the context of a single UMTS cell) that gives priority to streaming flows through bandwidth reservation (as opposed to head-of-line marking), and ensures stability through admission control on both types of flows (similar to [2]). We develop various analysis approaches based on a fluid model and evaluate these techniques against simulation in terms of various performance measures such as expected response time and blocking probabilities.