Regressive admission control enabled by real-time QoS measurements

We propose a novel regressive principle to Admission Control (AC) assisted by real-time passive QoS monitoring. This measurement-based AC scheme accepts flows by default, but based on the changes in the network QoS, it makes regressive decisions on the possible flow rejection, thus bringing cognition to the network path. The REgressive Admission Control (REAC) system consists of three modules performing the necessary tasks: QoS measurements, traffic identification, and the actual AC decision making and flow control. There are two major advantages with this new scheme; (i) significant optimization of the connection start-up phase, and ii) continuous QoS knowledge of the accepted streams. In fact, the latter combined with the REAC decisions can enable guaranteed QoS without requiring any QoS support from the network. REAC was tested on a video streaming test bed and proved to have a timely and realistic match between the network's QoS and the video quality.