Supporting Quality of Service is an important objective for future mobile systems, and requires resource reservation and admission control to achieve. In this thesis, we introduce a scalable scheme to admission control termed Virtual Bottleneck Cell. Our approach is designed to scale to many users and hand-offs, while simultaneously controlling "hot spots". The key technique is to hierarchically control the virtual system, ensuring QoS objectives are satisfied without requiring accurate predictions of the users' future locations. We develop a simple analytical model to study the system and illustrate several key components of the approach. We formulate the problem of how to group the cells to form the virtual system as an optimization problem and propose a heuristic adaptive clustering algorithm as its solution. Finally, we perform simulations in a two-dimensional network to compare the performance obtained with VBC and adaptive clustering with alternate schemes, including the optimal offline algorithm.