Adaptive Reservation: A New Framework for Multimedia Adaptation

Research on multimedia adaptation usually assumes undifferentiated, best-effort service from the network, and relies on signaling mechanisms such as packet loss rates for feedback. These methods result in unfairness to adaptive applications in the presence of aggressive, non-adaptive applications. A network with enhancements for QoS support, and usage and QoS dependent pricing, can use pricing as a natural incentive to drive adaptive behavior by applications. In this work, we present a framework for the adaptation of multimedia application sending rate and/or choice of network services in response to dynamic network pricing and changes in application requirements. The adaptation is based on the (monetary) value of a service as perceived by the user, relative to the price charged by the network. Experimental results show that perceived-value-based adaptation allows bandwidth to be shared fairly among competing users. When network resources are scarce, bandwidth is shown to be distributed among competing applications (and among media streams belonging to a single multimedia system) according to their relative elasticity of demand, indicated by the sensitivity of the perceived value to the bandwidth.